task-p4.5.1

docs/implementation/task-p.4.5.1-visual-qa-checklist.md

@@ -0,0 +1,37 @@
+# Task P.4.5.1 Visual QA Checklist
+
+## Automated checks
+
+- PDF generation succeeds for every fixture
+- baseline manifest exists for every fixture
+- page count matches stored baseline
+- rendered section order matches stored baseline
+- signature block remains on the last page
+- `items_table` row shape remains seven columns
+- runtime produces no error-severity issues
+- 500-item performance fixture generates without crash
+
+## Manual checks
+
+- company block alignment
+- customer block alignment
+- quotation code and quotation date alignment
+- product table header repetition on multi-page item fixtures
+- table borders remain continuous across page breaks
+- long-description rows do not clip text
+- Thai and English mixed text remain readable
+- signature block spacing remains intact
+- topic spacing and numbering remain consistent
+- footer copy remains aligned
+
+## Review scope
+
+- `ALLA`
+- `ONVALLA`
+- empty state
+- single item
+- standard 4-item case
+- 20-item pagination case
+- 100-item heavy pagination case
+- long-description wrapping case
+- multi-topic continuation case

docs/implementation/task-p.4.5.1-visual-regression-report.md

@@ -0,0 +1,65 @@
+# Task P.4.5.1 Visual Regression Report
+
+## Scope
+
+This task establishes reusable visual regression baselines for Product Item quotation templates without changing quotation runtime architecture.
+
+Implemented outputs:
+
+- baseline fixture generation under [fixtures/pdf-visual/README.md](/C:/Users/mtpphtaps/Documents/gitea/alla-allaos-fullstack/fixtures/pdf-visual/README.md)
+- automated visual-regression audit script at [scripts/audit-pdf-visual-regression.ts](/C:/Users/mtpphtaps/Documents/gitea/alla-allaos-fullstack/scripts/audit-pdf-visual-regression.ts)
+- manual QA checklist at [docs/implementation/task-p.4.5.1-visual-qa-checklist.md](/C:/Users/mtpphtaps/Documents/gitea/alla-allaos-fullstack/docs/implementation/task-p.4.5.1-visual-qa-checklist.md)
+
+## Generated fixture set
+
+Baselines are organized by reusable scenario:
+
+- `quotation-empty`
+- `quotation-single-item`
+- `quotation-standard`
+- `quotation-20-items`
+- `quotation-100-items`
+- `quotation-long-description`
+- `quotation-multi-topic`
+
+Each scenario stores `ALLA` and `ONVALLA` golden PDFs plus manifest metadata.
+
+## Automated validation coverage
+
+The visual regression script validates:
+
+- PDF generation success
+- page count stability
+- section order stability
+- Product Item page presence through runtime structure
+- `items_table` seven-column contract
+- signature block remaining on the last rendered page
+- runtime issue severity
+- 500-item generation performance smoke check
+
+Current run report is written to:
+
+- `artifacts/pdf-visual/visual-regression-report.json`
+- `artifacts/pdf-visual/visual-regression-report.md`
+
+## Screenshot strategy
+
+This environment does not currently provide a PDF-to-image renderer such as `pdftoppm`, Ghostscript, or ImageMagick, and `sharp` cannot rasterize the generated PDFs here.
+
+Because of that:
+
+- golden PDF baselines are generated now
+- current-run PDFs are generated now
+- manifest-based regression comparison is automated now
+- pixel screenshot comparison is documented but not automated in this environment
+
+If a renderer is added later, the current fixture layout can be extended without changing the dataset contract.
+
+## Acceptance mapping
+
+- visual fixtures created = PASS
+- golden PDF baselines created = PASS
+- future regression comparison path created = PASS
+- automated layout validation added where practical = PASS
+- screenshot automation documented with environment limitation = PASS
+- manual visual sign-off still required for true pixel review = PENDING