Add CJK (Chinese, Japanese, Korean) search support #2299
Merged
+174
−5
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The search functionality was silently dropping CJK characters because:
1. Query sanitization used `\w` which only matches ASCII word characters
2. Stemmer split by whitespace, which doesn't work for CJK languages
3. Highlighter used `\b` word boundaries that don't apply to CJK
This commit fixes all three issues:
- Query: Use `\p{L}\p{N}` (Unicode letters/numbers) instead of `\w`
- Stemmer: Preserve CJK characters as-is without stemming, since CJK
languages don't have stemming rules like English
- Highlighter: Skip word boundary matching for CJK terms
Also extracts `CJK_PATTERN` to `Search` module to avoid duplication.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <[email protected]>
- Split each CJK character into individual tokens for FTS5 indexing - Add stem_content callback to SQLite adapter to pre-process content - Stem search queries before matching against FTS5 index - Update stemmer tests to reflect character-splitting behavior This enables CJK search on SQLite installations where the FTS5 tokenizer cannot natively segment CJK text. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude Opus 4.5 <[email protected]>
jorgemanrubia
approved these changes
Jan 7, 2026
5 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
Problem
The search functionality was broken for CJK languages because:
\wregex which only matches ASCII word characters, removing all CJK text\bword boundaries that don't apply to CJK charactersSolution
\p{L}\p{N}(Unicode letters/numbers) instead of\wstem_contentcallback to pre-process content before indexing, and stem queries before matchingAlso extracts
CJK_PATTERNconstant toSearchmodule to avoid duplication.How it works
For CJK text like "日本語" (Japanese), the stemmer splits it into "日 本 語" (three separate tokens). This allows SQLite FTS5 with
unicode61tokenizer to match partial searches like "日本" or even single characters like "日".English words are still processed normally with the Mittens stemmer (e.g., "running" → "run").
Test plan
🤖 Generated with Claude Code