learningcircuit--local-deep-research/.github/workflows/pr-triage.yml

name: PR triage labels

# Auto-applies author-class labels at PR open and toggles lifecycle labels
# on review/synchronize events. Labels are defined in .github/labels.yml
# and synced by .github/workflows/labels-sync.yml.
#
# Also applies path-derived topic labels (documentation, tests, docker,
# github-actions, ...) that feed release-notes categorization. These were
# historically applied by the AI code reviewer, which became label-only
# opt-in (#4955), so triage now owns the mechanical subset. Topic labels
# are applied on opened and ready_for_review only, and are additive-only:
# a label a maintainer removed is not re-added on every push, and
# semantic labels (bugfix, feature, security, performance...) stay
# human/AI-applied.
#
# Uses pull_request (not pull_request_target) — fork PRs get a read-only
# token, so label calls return 403. The script catches 403 and continues
# so fork PRs don't show a failing check; maintainers can apply labels
# manually for fork PRs that need them. We accept the no-op behavior on
# forks vs the security cost of pull_request_target running with secrets
# on fork code.
#
# CODEOWNERS list below is mirrored in .github/CODEOWNERS global owners
# (line 6) — keep both in sync when the maintainer roster changes.

on:
  pull_request:
    types: [opened, synchronize, ready_for_review]
  pull_request_review:
    types: [submitted, dismissed]

# No concurrency group — intentionally omitted.
# Previous attempt (#3554, reverted #3599) used cancel-in-progress which
# killed in-progress PR runs before they produced useful results. Race
# between synchronize and pull_request_review can produce transient
# label flapping in rare cases; maintainers can correct manually.

permissions: {}  # Minimal top-level for OSSF Scorecard Token-Permissions

jobs:
  triage:
    runs-on: ubuntu-latest
    timeout-minutes: 5
    permissions:
      issues: write         # PR labels live on the underlying issue resource
      pull-requests: read   # listFiles for path-derived topic labels
    steps:
      - name: Harden the runner (Audit all outbound calls)
        uses: step-security/harden-runner@9af89fc71515a100421586dfdb3dc9c984fbf411 # v2.19.4
        with:
          egress-policy: audit

      - name: Apply triage labels
        uses: actions/github-script@3a2844b7e9c422d3c10d287c895573f7108da1b3 # v9.0.0
        with:
          script: |
            // Keep CODEOWNERS list in sync with .github/CODEOWNERS line 6.
            const CODEOWNERS = ['LearningCircuit', 'hashedviking', 'djpetti'];
            // Known AI-bot accounts that don't carry the [bot] suffix.
            const KNOWN_BOTS = ['moltenbot000', 'mseep-ai', 'Nexus-Digital-Automations'];

            const pr = context.payload.pull_request;
            const issueNumber = pr.number;
            const author = pr.user.login;
            const association = pr.author_association;
            const isBot = author.endsWith('[bot]') || KNOWN_BOTS.includes(author);
            const isInternal = ['OWNER', 'MEMBER', 'COLLABORATOR'].includes(association);

            // 403 on label calls is the expected outcome for fork PRs —
            // pull_request gives forks a read-only token by design. We
            // log and continue so the workflow run stays green and a
            // maintainer can apply labels manually instead of seeing
            // a red check on every fork contribution.
            const isReadOnlyTokenError = (err) => {
              if (err.status !== 403) return false;
              console.log(`Label call returned 403 (read-only token, likely a fork PR). Skipping.`);
              return true;
            };

            const addLabels = async (labels) => {
              if (!labels.length) return;
              try {
                await github.rest.issues.addLabels({
                  owner: context.repo.owner,
                  repo: context.repo.repo,
                  issue_number: issueNumber,
                  labels,
                });
              } catch (err) {
                if (!isReadOnlyTokenError(err)) throw err;
              }
            };

            const removeLabel = async (name) => {
              try {
                await github.rest.issues.removeLabel({
                  owner: context.repo.owner,
                  repo: context.repo.repo,
                  issue_number: issueNumber,
                  name,
                });
              } catch (err) {
                if (err.status === 404) return;  // label wasn't applied
                if (!isReadOnlyTokenError(err)) throw err;
              }
            };

            // Path-derived topic labels for release-notes categorization.
            // Mechanical mappings only — a rule belongs here iff the label
            // follows from *which files* changed, never from what the
            // change means. First-match-per-rule over all changed files.
            const TOPIC_RULES = [
              [/^\.github\/workflows\//, 'github-actions'],
              [/^(\.github\/|\.pre-commit-hooks\/|\.pre-commit-config\.yaml$|scripts\/ci\/)/, 'ci-cd'],
              [/^(Dockerfile|docker-compose|cookiecutter-docker\/|unraid-templates\/)/, 'docker'],
              [/^(docs\/|[^/]+\.md$)/, 'documentation'],
              [/^tests\//, 'tests'],
              [/^src\/local_deep_research\/database\//, 'database'],
              [/^src\/local_deep_research\/(defaults|settings)\//, 'configuration'],
              [/^(pyproject\.toml|pdm\.lock|package\.json|package-lock\.json)$/, 'dependencies'],
              [/^src\/local_deep_research\/(advanced_search_system|web_search_engines)\//, 'research'],
              [/^src\/.*\.py$/, 'python'],
              [/^src\/.*\.(js|mjs)$/, 'javascript'],
              [/^src\/.*\.css$/, 'css'],
            ];

            const topicLabels = async () => {
              const files = await github.paginate(github.rest.pulls.listFiles, {
                owner: context.repo.owner,
                repo: context.repo.repo,
                pull_number: issueNumber,
                per_page: 100,
              });
              const labels = new Set();
              for (const f of files) {
                for (const [pattern, label] of TOPIC_RULES) {
                  if (pattern.test(f.filename)) labels.add(label);
                }
              }
              return [...labels];
            };

            const action = context.payload.action;
            const eventName = context.eventName;

            if (eventName === 'pull_request' && action === 'opened') {
              const labels = [];
              if (isBot) labels.push('bot');
              if (!isInternal && !isBot) labels.push('external-contributor');
              if (association === 'FIRST_TIME_CONTRIBUTOR') labels.push('first-time-contributor');
              if (!isInternal) labels.push('needs-codeowner-review');
              labels.push(...await topicLabels());
              await addLabels(labels);
              return;
            }

            if (eventName === 'pull_request' && (action === 'synchronize' || action === 'ready_for_review')) {
              const current = pr.labels.map((l) => l.name);
              if (current.includes('awaiting-author')) {
                await removeLabel('awaiting-author');
                await addLabels(['awaiting-codeowner']);
              }
              // Draft → ready is a deliberate "the diff is final" moment, so
              // recompute topic labels once there. Deliberately NOT done on
              // synchronize: that would re-add labels a maintainer removed,
              // on every push.
              if (action === 'ready_for_review') {
                await addLabels(await topicLabels());
              }
              return;
            }

            if (eventName === 'pull_request_review' && action === 'submitted') {
              const review = context.payload.review;
              const reviewer = review.user.login;
              // Strict CODEOWNERS-only check. The hardcoded list above must
              // mirror .github/CODEOWNERS line 6. We deliberately do NOT
              // accept any OWNER/MEMBER/COLLABORATOR association as a
              // codeowner: branch protection here may later adopt
              // require_code_owner_reviews=true, at which point clearing
              // needs-codeowner-review on a non-codeowner MEMBER review
              // would be a security-relevant mislabel.
              if (!CODEOWNERS.includes(reviewer)) return;

              if (review.state === 'approved') {
                await removeLabel('needs-codeowner-review');
                await removeLabel('awaiting-codeowner');
                // Also clear awaiting-author: a codeowner can resolve
                // their own changes_requested review by approving without
                // an intervening synchronize (e.g., the author convinced
                // them via comments).
                await removeLabel('awaiting-author');
              } else if (review.state === 'changes_requested') {
                await removeLabel('needs-codeowner-review');
                await removeLabel('awaiting-codeowner');
                await addLabels(['awaiting-author']);
              }
              // 'commented' → no-op.
              return;
            }

            if (eventName === 'pull_request_review' && action === 'dismissed') {
              // GitHub sets review.state to "dismissed" on this event — the
              // original state is not preserved (github/docs#20216). Use the
              // awaiting-author label as the discriminator: it's only set by
              // a codeowner's changes_requested review, so its presence is
              // proof the dismissal is the one we care about. Dismissals of
              // approval/comment reviews are naturally no-ops because
              // awaiting-author won't be present.
              const review = context.payload.review;
              if (!CODEOWNERS.includes(review.user.login)) return;

              const current = pr.labels.map((l) => l.name);
              if (current.includes('awaiting-author')) {
                await removeLabel('awaiting-author');
                await addLabels(['needs-codeowner-review']);
              }
              return;
            }