Compare
Sorry, something went wrong.
No results found
v3.2.1 — Provenance polish + full Wumenguan documentation
Provenance panel improvements:
- Clickable witness URLs (open in browser, truncated with full tooltip)
- Full production method text (removed 400-char truncation)
- Documents section: discovers + displays .md files from provenance/
and docs/curation/exemplars/ as collapsible expanders
- Minimal markdown renderer (headings, **bold**, `code`, bullet lists,
numbered lists, | tables |, --- separators) — no NuGet dependencies
- Tables horizontally scrollable with auto-sizing columns
- Panel takes full column height when study panel is hidden
- CBETA fallback enhanced with title/author/license from TEI header
- "Source:" label now before the translation source combobox
OpenZenTexts repo (Fabulu/OpenZenTexts c13034d):
Published full Wumenguan provenance chain (+5,916 lines):
- 60KB diplomatic witness draft (leaf-by-leaf with evidence)
- Scan-derived case ordering proof
- 6 rebuild chunks (editorial assembly per case range)
- 5 witness READMEs (NDL 1632, Waseda 1752, Wumen Huikai, Korea 1882, Wikisource)
- Captured Wikisource wikitext as comparison witness
- Wumen Huikai corroboration mappings
- Transcription system docs + leaf scorecard
- PROCESS_LOG_TEMPLATE.md for future transcriptions
Tests: 882 pass / 2 skip / 0 fail