Souta 3072681c18 fix(docs-ci): slugify keeps Unicode letters so JA anchors are checkable ... The slugify in check-docs-links.mjs stripped all non-ASCII via `[^a-z0-9\s\-]`, so every Japanese / CJK / accented heading collapsed to the same empty string `""`. Both the heading-set entry AND the parsed anchor would slugify to `""` → silent match → any bad JA anchor slipped through the gate. Concrete demonstration before the fix: '概要' -> '' '詳細' -> '' '使い方' -> '' '何ができるか' -> '' After the fix (Unicode-aware via `\p{L}\p{N}` + NFC): '概要' -> '概要' '詳細' -> '詳細' '使い方' -> '使い方' '何ができるか' -> '何ができるか' Verified by injecting a deliberately-bad JA anchor (`[[ja/migration/from-obsidian-sync#存在しない見出し]]`) and confirming the checker now reports (slug "存在しない見出し" not in docs/ja/migration/from-obsidian-sync.md) instead of silently passing. ASCII-only headings produce identical slugs as before (`Backup & restore` → `backup-restore`), so no regression on the EN tree. This matters now that the JA tree (PRs #313 + #314) is shipping in 1.0.46 — the gate would have given false confidence on the new content. Carryover from PR #309 review.		2026-05-10 23:25:11 +09:00
..
check-docs-links.mjs	fix(docs-ci): slugify keeps Unicode letters so JA anchors are checkable	2026-05-10 23:25:11 +09:00