wehub-resource-sync
106 lines
4.9 KiB
Python
106 lines
4.9 KiB
Python
"""End-to-end regression test: ``omnigent claude --resume`` restores history.
|
|
|
|
Reproduces the user-reported bug: running
|
|
|
|
omnigent claude --server <url> --resume <conv_id>
|
|
|
|
attached a Claude Code session with **no history** — the prior conversation
|
|
was silently lost because the runner auto-created a *fresh* Claude terminal
|
|
(no ``--resume``) for the CLI-driven session, racing and beating the CLI's own
|
|
cold-resume launch. The fix gates the runner's auto-create to host-spawned
|
|
(web-UI) sessions only; CLI-driven sessions keep their own cold-resume launch.
|
|
|
|
This is the robust, outcome-based check: it drives the REAL ``omnigent
|
|
claude`` CLI to create a conversation that knows a passphrase, resumes it, and
|
|
then — through the server's message API, the same path the web UI uses —
|
|
verifies Claude answers a follow-up with that passphrase. The passphrase
|
|
coming back proves the resumed session actually had the earlier turn. The
|
|
shared flow lives in :func:`assert_native_cli_resume_restores_history` (see
|
|
``tests/e2e/_native_resume_helpers.py``); the deterministic
|
|
``--resume``-injection mechanism is separately guarded by the runner unit
|
|
tests in ``tests/runner/test_app_sessions_native.py``.
|
|
|
|
Environment requirements (why this is opt-in, not pure-CI)
|
|
----------------------------------------------------------
|
|
* **Opt-in only**: set ``OMNIGENT_E2E_CLAUDE_NATIVE=1`` to run. claude-native
|
|
needs an *interactive* Claude login (OAuth/Enterprise) anchored to the real
|
|
``$HOME`` — it cannot be relocated into CI. The ``claude`` binary IS present
|
|
in CI (claude-sdk installs it), so gating on binary presence alone would let
|
|
this run unauthenticated and hang the TUI. The env-var gate keeps it out of
|
|
CI; a developer with a logged-in Claude opts in explicitly.
|
|
* Run it the same way as the host claude-native test::
|
|
|
|
OMNIGENT_E2E_CLAUDE_NATIVE=1 \
|
|
.venv/bin/python -m pytest tests/e2e/test_claude_native_cli_resume_e2e.py \
|
|
--profile oss \
|
|
--llm-api-key "$(databricks auth token -p oss \
|
|
| python -c 'import sys,json;print(json.load(sys.stdin)["access_token"])')" \
|
|
-v
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import os
|
|
import shutil
|
|
from pathlib import Path
|
|
|
|
import pytest
|
|
|
|
from tests.e2e._native_resume_helpers import assert_native_cli_resume_restores_history
|
|
|
|
# ``resume_test_server`` is provided by tests/e2e/conftest.py (shared with the
|
|
# codex resume test) and resolved by pytest fixture discovery.
|
|
|
|
# Opt-in only — see module docstring. The `claude` binary alone is not a
|
|
# sufficient gate (present-but-unauthenticated in CI hangs the TUI), so
|
|
# require an explicit env var that only a developer with a logged-in Claude
|
|
# sets.
|
|
pytestmark = pytest.mark.skipif(
|
|
os.environ.get("OMNIGENT_E2E_CLAUDE_NATIVE") != "1" or shutil.which("claude") is None,
|
|
reason=(
|
|
"claude-native CLI resume e2e needs an interactive Claude login; set "
|
|
"OMNIGENT_E2E_CLAUDE_NATIVE=1 (and have `claude` installed + logged in) to run"
|
|
),
|
|
)
|
|
|
|
|
|
def test_claude_native_cli_resume_restores_history(
|
|
resume_test_server: str,
|
|
tmp_path: Path,
|
|
request: pytest.FixtureRequest,
|
|
) -> None:
|
|
"""
|
|
Cross-context ``omnigent claude --resume`` restores conversation history.
|
|
|
|
Drives the real ``omnigent claude --server …`` CLI (gateway routing
|
|
via the config-home auth block from the pytest ``--profile``) to teach
|
|
Claude a passphrase, **deletes Claude's local transcript** for that session,
|
|
then resumes — so the resume cannot reuse Claude's own on-disk transcript
|
|
and must go through Omnigent' cold-resume *synthesis* (rebuild the
|
|
transcript from server-side items and hand it to ``claude --resume``). A
|
|
recall message is sent through the server and Claude must reply with the
|
|
passphrase, proving the resumed session loaded the prior turn.
|
|
|
|
This is the exact scenario a user hits resuming a conversation created
|
|
elsewhere / in another cwd / on another machine — the path that was
|
|
silently losing history. It exercises both halves of the fix end-to-end:
|
|
the runner re-creating the terminal on a reused daemon runner (the ensure
|
|
path) and ``_ensure_local_claude_resume_transcript`` synthesizing the
|
|
transcript. (We deliberately do NOT also test the same-machine fast path —
|
|
that only exercises Claude resuming its own file, not Omnigent' code, and
|
|
running a second back-to-back CLI flow thrashes the shared host daemon.)
|
|
|
|
:param resume_test_server: Base URL of the allow-list-free test server.
|
|
:param tmp_path: Per-test temp dir.
|
|
:param request: Pytest request — reads ``--profile`` for the LLM gateway.
|
|
"""
|
|
profile = request.config.getoption("--profile")
|
|
assert profile, "this test requires --profile (e.g. --profile oss) for the LLM gateway"
|
|
assert_native_cli_resume_restores_history(
|
|
harness="claude",
|
|
server=resume_test_server,
|
|
profile=profile,
|
|
tmp_path=tmp_path,
|
|
force_cold_resume=True,
|
|
)
|