iftypeset/STATUS.md

iftypeset status (pubstyle)

Updated: 2026-01-03
Project root: /root/ai-workspace/iftypeset/

What exists (working today)

• Spec + schema: spec/schema/rule.schema.json, spec/manifest.yaml
• Profiles: spec/profiles/*.yaml (web_pdf, print_pdf, dense_tech, memo, slide_deck)
• Post-render QA gates: spec/quality_gates.yaml
• Rule registry (seeded): spec/rules/**.ndjson
• Indexes (derived): spec/indexes/*.json (rebuildable)
• CLI: validate-spec, report, lint, render-html, render-pdf, qa, emit-css
• Ephemeral extraction helpers: tools/ (Chicago OCR is grep-only, temp files deleted)
• Forgejo integration note: forgejo/README.md
• Fixtures + tests: fixtures/ and tests/
• CI script: scripts/ci.sh (validate-spec, report, unit tests)

Rule corpus snapshot

From out/coverage-report.json:

• Total rules: 307
• By category: citations 61, numbers 62, punctuation 55, layout 46, headings 32, tables 23, typography 15, links 5, accessibility 4, code 4
• By enforcement: manual 186, typeset 62, lint 46, postrender 13
• By severity: must 28, should 263, warn 16

Current rule batches

• spec/rules/accessibility/v1_accessibility_001.ndjson (4)
• spec/rules/citations/v1_citations_001.ndjson (16)
• spec/rules/citations/v1_citations_002.ndjson (45)
• spec/rules/code/v1_code_001.ndjson (4)
• spec/rules/headings/v1_headings_001.ndjson (12)
• spec/rules/headings/v1_headings_002.ndjson (20)
• spec/rules/layout/v1_layout_001.ndjson (12)
• spec/rules/layout/v1_layout_002.ndjson (30)
• spec/rules/layout/v1_layout_003.ndjson (4)
• spec/rules/links/v1_links_001.ndjson (5)
• spec/rules/numbers/v1_numbers_001.ndjson (12)
• spec/rules/numbers/v1_numbers_002.ndjson (50)
• spec/rules/punctuation/v1_punctuation_001.ndjson (15)
• spec/rules/punctuation/v1_punctuation_002.ndjson (40)
• spec/rules/tables/v1_tables_001.ndjson (8)
• spec/rules/tables/v1_tables_002.ndjson (15)
• spec/rules/typography/v1_typography_001.ndjson (8)
• spec/rules/typography/v1_typography_002.ndjson (7)

How to validate and inspect

• Validate spec + rebuild indexes:
  • PYTHONPATH=src python3 -m iftypeset.cli validate-spec --spec spec --build-indexes
• Lint:
  • PYTHONPATH=src python3 -m iftypeset.cli lint --input fixtures/sample.md --out out --profile web_pdf
• Render HTML + CSS:
  • PYTHONPATH=src python3 -m iftypeset.cli render-html --input fixtures/sample.md --out out --profile web_pdf
• Render PDF (if renderer installed):
  • PYTHONPATH=src python3 -m iftypeset.cli render-pdf --input fixtures/sample.md --out out --profile web_pdf
• Run QA gates (HTML fallback if no PDF):
  • PYTHONPATH=src python3 -m iftypeset.cli qa --out out --profile web_pdf
• Coverage report:
  • PYTHONPATH=src python3 -m iftypeset.cli report --spec spec --out out --build-indexes
• Emit CSS for a profile:
  • PYTHONPATH=src python3 -m iftypeset.cli emit-css --spec spec --profile web_pdf --out out-css
• Run unit tests:
  • python3 -m unittest discover -s tests -p 'test_*.py'

Key constraints (don’t drift)

• No bulk OCR/transcription of books into repo. Rules must be paraphrased and pointer-backed.
• source_refs must be pointers, not quotes; include (scan pN) only as a single page hint.
• Chicago extraction may use OCR ephemerally only to locate pointers; do not persist OCR output.

Next work (highest leverage)

• Add new batches for: figures, frontmatter, backmatter, abbreviations, i18n, and expand accessibility.
• Grow post-render QA rule coverage (widows/orphans, heading keeps, overflow) beyond the current seed set.
• Add a real PDF-layout analyzer when a stable renderer is selected (widows/orphans, overflow).