fsecada01/Pygentic-AI
0
0
Fork 0
mirror of https://github.com/fsecada01/Pygentic-AI.git synced 2026-05-12 12:15:00 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
93 Commits 4 Branches 0 Tags
dev_deploy
Commit Graph

4 Commits

Author SHA1 Message Date
Francis Secada
dd2b2e8a33 feat(pdf): improve download filename with company names and date 
- Generate descriptive filenames: swot-{company}-vs-{competitor}-{date}.pdf
- Sanitize company names for filesystem safety
- Include date in YYYY-MM-DD format
- Handle multiple comparison entities (e.g., "plus2" for 2 additional)

Examples:
- Single entity: "swot-Apple-2026-02-04.pdf"
- With comparison: "swot-Apple-vs-Microsoft-2026-02-04.pdf"
- Multiple comparisons: "swot-Apple-vs-Microsoft-plus2-2026-02-04.pdf"

Fixes: Downloaded PDF had generic filename with .txt suffix

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2026-02-04 17:04:39 -05:00
Francis Secada
1eb6c33fa3 fix(tests): fix empty HTML responses and PDF assertion thresholds 
- Fix empty HTML response tests by calling endpoints directly with mocked requests
  - TestClient cookies don't work with SessionMiddleware's encrypted sessions
  - Changed to call get_status() and get_result() directly with MagicMock requests
  - Updated assertions to use response.body.decode() instead of response.text

- Fix PDF assertion thresholds
  - Lowered size threshold from 10KB to 2KB (realistic for minimal SWOT PDFs)
  - Fixed buffer position checks (use seek(0,2) for reliable end position)
  - Removed unreliable text search in compressed PDFs

Regression tests for:
- SessionMiddleware session handling in integration tests
- PDF size validation for minimal reports

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2026-02-04 16:45:56 -05:00
Francis Secada
6745b3c870 test: add comprehensive regression tests for PDF download error handling 
Tests mock the exact production bug scenario:
- Session exists but result is None → 404 with proper Response

Regression Tests:
- test_download_pdf_without_session_returns_404
  → No session ID error path
- test_download_pdf_without_result_returns_404
  → Session but no result (EXACT production bug)
- test_download_pdf_error_paths_use_response_not_streaming
  → Verifies Response used, not StreamingResponse

Additional Tests:
- test_download_pdf_returns_pdf_file
  → Successful PDF generation and download
- test_download_pdf_uses_cache
  → PDF caching behavior validation
- test_download_pdf_filename_format
  → Filename format verification

Mocking Strategy:
- Use unittest.mock.MagicMock for request.session
- Mock session.get() to return test session_id
- Call download_pdf() handler directly with mocked request
- Avoids TestClient session middleware complications

Coverage Improvements:
- pdf_service.py: 31% → 99%
- pdf_cache.py: 37% → 76%
- router.py: 35% → 53%
- Overall: 49% → 63%

All tests pass and prevent regression of AttributeError bug.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2026-02-04 15:59:36 -05:00
Francis Secada
1be156ed7c feat: add pytest infrastructure and regression tests 
Implements comprehensive testing framework to prevent regressions like:
- PDF style name conflicts (BodyText)
- StreamingResponse type errors
- Template not found after refactors

Test Infrastructure:
- pytest.ini: Configuration with coverage, markers, asyncio support
- conftest.py: Shared fixtures (test_client, test_db, sample_data)
- GitHub Actions CI: Automated testing on push/PR
- Directory structure: tests/{unit,integration,fixtures}

Integration Tests (test_analyze_flow.py):
- Regression: analyze endpoint returns empty response (not status.html)
- Status polling with OOB swaps
- Session creation and management
- First poll returns container + items
- Subsequent polls return only new items
- Result endpoint with/without data

Integration Tests (test_pdf_export.py):
- Regression: PDF generation returns BytesIO (not int)
- Regression: No ReportLab style name conflicts
- PDF download endpoint with streaming response
- PDF caching behavior
- Valid PDF format verification
- Filename format validation

Unit Tests (test_pdf_service.py):
- Content hash generation and consistency
- PDF generator initialization
- Custom style creation without conflicts
- SwotAnalysis model validation

CI/CD:
- GitHub Actions workflow for automated testing
- Python 3.13 support
- Coverage reporting with codecov integration

Test Markers:
- @pytest.mark.unit: Fast, isolated tests
- @pytest.mark.integration: Multi-component tests
- @pytest.mark.pdf: PDF-related tests
- @pytest.mark.api: API endpoint tests

Fixtures:
- test_client: FastAPI TestClient
- test_db_session: SQLite in-memory database
- sample_swot_analysis: Mock SWOT data
- clear_caches: Auto-cleanup between tests

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2026-02-04 15:18:07 -05:00