Skip to content

[chore](i18n) Remove Japanese docs and the EN->JA translation pipeline - #4075

Merged
morningman merged 1 commit into
apache:masterfrom
morningman:remove-ja
Aug 19, 2026
Merged

[chore](i18n) Remove Japanese docs and the EN->JA translation pipeline#4075
morningman merged 1 commit into
apache:masterfrom
morningman:remove-ja

Conversation

@morningman

Copy link
Copy Markdown
Contributor

Versions

  • dev
  • 4.x
  • 3.x
  • 2.1 or older (not covered by version/language sync gate)

(Applies to all versions: the Japanese tree carried current / 4.x / 3.x / 2.1 copies.)

Languages

  • Chinese
  • English

Japanese only — Chinese and English content are untouched.

What this does

Removes the Japanese locale from the website entirely, together with the pipeline that produced it.

Deleted

  • ja-source/ — 5,703 files, the Japanese doc source tree
  • ja-build/ — the prebuilt Japanese site copied into build/ja at deploy time
  • scripts/i18n/ — all 9 scripts. Every one was added by the original feat(website): i18n ja (#3468) commit, and nothing outside the translation workflow referenced them.
  • .github/workflows/manual-i18n-translate-workflow.yaml — the EN→JA translation pipeline

Deploy workflows

cron-deploy-website.yml and manual-deploy-website.yml no longer guard on ja-build or copy it into build/ja. Their build step already passed only --locale en --locale zh-CN, so nothing else about the build changed.

Site config

  • docusaurus.config.js: locales['en', 'zh-CN']; dropped the ja localeConfig, the /ja/search sitemap exclusion, 'ja' from the search languages, and the ja/docs / ja/course route bases
  • config/custom-docusaurus-plugin.js, scripts/update_search_hash.js, and the start:ja / write-translations:ja package scripts
  • src/theme/TOC/index.tsx: dropped the Japanese UI-string branch
  • scripts/patch-search-tokenize.js: removed the two Japanese-only node_modules patches (kana tokenizer preference, zh+ja mixed-mode tokenizer). The four locale-aware search patches stay — zh-CN still needs them.

Docs governance

The ja locale, the ja_source content root, and the i18n-sync-locale-candidate rule are removed from .docs-governance/rules.yml, manifest.js, lint-links.js, lint-seo.js, and lint-i18n-sync.js, with matching test updates. i18n-sync-policy.md, translation-glossary.yml, the exceptions.yml example, and the PR template's Japanese checkbox are updated to match.

website-quality-governance/broken-links-baseline.md: the ja-source/ rows are dropped and its 502 internal errors subtracted from the internal tables (exact per-root counts from the same scan). The external info count has no per-root breakdown and still reflects the original scan; the file now says so.

Notes for reviewers

Two deliberate judgment calls:

  • The locale dropdown's "Coming Soon" badge is kept. ja was its only member, so COMING_SOON_LOCALES is now an empty set; the markup and styles are untouched so a future locale can use it.
  • src/utils/tokenize.ts is left alone. Its ["ja", "jp", "th"] list is a vendored copy of the upstream search-plugin util (added well before the Japanese work), and jp / th are equally unused there. Removing just ja would be arbitrary.

Docs Checklist

  • Checked by AI
  • Test Cases Built
  • Updated required version and language counterparts, or explained why not
  • If only one language changed, confirmed whether source/translation counterparts need sync

yarn docs-governance:test passes 37/37. No site build was run locally.

🤖 Generated with Claude Code

Drops the Japanese locale from the website entirely.

Removed:
- `ja-source/` (5703 files), the Japanese doc source tree
- `ja-build/`, the prebuilt Japanese site copied into `build/ja` at deploy
- `scripts/i18n/`, all 9 scripts, each added by the original
  `feat(website): i18n ja (apache#3468)` commit and referenced only by the
  translation workflow
- `.github/workflows/manual-i18n-translate-workflow.yaml`, the EN->JA
  translation pipeline

Deploy workflows (`cron-deploy-website.yml`, `manual-deploy-website.yml`)
no longer guard on `ja-build` or copy it into `build/ja`. Their build step
already passed only `--locale en --locale zh-CN`.

Site config: `locales` is now `['en', 'zh-CN']`; the `ja` localeConfig,
`/ja/search` sitemap exclusion, `ja` search language, and `ja/docs`
route bases are gone, along with the ja branches in
`custom-docusaurus-plugin.js`, `update_search_hash.js`, the TOC UI strings,
and the `start:ja` / `write-translations:ja` package scripts.
`patch-search-tokenize.js` loses its two Japanese-only node_modules patches;
the four locale-aware search patches stay, since zh-CN still needs them.

Docs governance: ja locale, `ja_source` content root, and the
`i18n-sync-locale-candidate` rule are removed from the rules, manifest, and
linters, with matching test updates. `yarn docs-governance:test` passes 37/37.

The locale dropdown's "Coming Soon" badge mechanism is kept; its locale set
is now empty since `ja` was its only member.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
@morningman
morningman merged commit 6e27522 into apache:master Aug 19, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant