ducoterra
4971e2859d
Remove the blanket .agent/ gitignore so the phase roadmap, user stories, reports, and PLAN.md are versioned with the code. Only runtime artifacts (.agent/phase-sessions/, .agent/pipeline.log) remain ignored. Update AGENTS.md git protocol rule to match.
1.8 KiB
1.8 KiB
Phase 10 complete — all gates green, committed as 6ec6181.
Implemented
GET /api/documents/content?source=…&path=…(app/api/docs.py,DocContentinapp/schemas.py) — DB-only lookup, 404{detail:"document not found"},formatfrom lowercased suffix (textfallback)- Extracted the escape-first markdown renderer to
frontend/assets/markdown.js(classic script, relative<script src>onindex.html+document.html); rendering unchanged - New dark-themed viewer page
frontend/document.html+document.js: back link (history-aware), title, meta badges (source/format/mono path/indexed/chunks), md → ≤46rem rendered column, other formats → escaped<pre class="doc-raw">, designed 404 not-found card, landmarks/skip-link/aria-live/focus-to-main - Chat source chips + Sources-table path links → viewer in a new tab (
target="_blank" rel="noopener", percent-encoded URLs); no-CDN + emoji guards extended to the viewer; README UI section
Verification
uv run pytest --cov=app --cov-report=term-missing→ 201 passed, 99% onapp/(>90%)uv run ruff check .→ clean;uv run pyright→ 0 errorsuv run pytest tests/e2e/test_document_viewer.py -v --no-cov→ 5 passed (isolated)- All 9 existing story E2E suites re-run in isolation → all pass (incl. XSS-proofing E2E:
<script>fixture renders escaped, no dialog)
Notable
- E2E source values are
docs(imported-dir basename) with slash-bearing paths — exactly what the query encoding handles; updated the now-stale chip-href assertion in the phase-03 suite (/sources.html→ encoded viewer URL) and kept the cell-leveltitle(phase-02 contract) alongside the link's.
Next pending phase: none — 10_story_document_viewer.md was the last in todo/ (roadmap complete).