kanji-lint

Detects kanji outside the 常用漢字 (jōyō kanji) list in plain text and Markdown manuscripts, reporting each hit with code-point-accurate line/column position.

Install / build

npm install
npm run build

CLI usage

kanji-lint manuscript.md
kanji-lint "docs/**/*.md"
kanji-lint manuscript.md --allow-words allow-words.txt
kanji-lint manuscript.md --allow-kanji allow-kanji.txt
kanji-lint manuscript.md --format json

Allow-list files: one entry per line; blank lines and lines starting with # are skipped. --allow-kanji entries must each be exactly one character.

Exit codes: 0 no issues, 1 issues found, 2 execution error (missing file, empty glob match, malformed allow-list file).

Library usage

import { check } from "kanji-lint";

const result = check("齟齬がないか確認してください。", { format: "text" });

Data provenance

src/data/jouyou-kanji.ts contains the 2010-revision 常用漢字表 (2,136 characters). The list was cross-verified against two independently maintained sources (joyo-kanji and joyo-kanji-counts on npm) — identical character set, identical order, no duplicates — and is guarded by integrity tests in test/data.test.ts.

v0.1 scope

See the project spec for full scope. Notably out of scope for v0.1: 表外音訓 checks, ひらく/閉じる judgments, context-based replacement suggestions, proper-noun/jinmeiyō-kanji classification, variation selectors, Unicode normalization diffs, auto-fix.