@lexical/code (No Prism)

Drop-in replacement for @lexical/code with Prism.js removed.

What's preserved

• CodeNode, CodeHighlightNode, CodeExtension
• Keyboard handlers (Tab/Shift+Tab, Alt+Arrow line movement, Home/End)
• Gutter line numbers
• DOM import/export (including GitHub code tables)
• Language metadata and maps
• Full API surface — identical exports to upstream @lexical/code

What's removed

• prismjs dependency (0 bytes of Prism in your bundle)
• Syntax highlighting / tokenization (code renders as plain text)

Merge strategy

Only FacadePrism.ts differs from upstream. All other source files (index.ts, CodeNode.ts, CodeHighlightNode.ts, CodeExtension.ts, CodeHighlighterPrism.ts, FlatStructureUtils.ts) are byte-identical to upstream @lexical/code.

When upstream updates, merge conflicts are isolated to FacadePrism.ts only.

Additional fork-only files:

• src/shared/invariant.ts — shim for monorepo-internal shared/invariant
• tsconfig.json — paths aliases for shared/invariant and @lexical/code
• tsup.config.ts — esbuild alias for shared/invariant

Usage

Override @lexical/code in your package manager:

pnpm

{
  "pnpm": {
    "overrides": {
      "@lexical/code": "npm:@anthropic/lexical-code-no-prism@^0.41.0"
    }
  }
}

npm

{
  "overrides": {
    "@lexical/code": "npm:@anthropic/lexical-code-no-prism@^0.41.0"
  }
}

yarn

{
  "resolutions": {
    "@lexical/code": "npm:@anthropic/lexical-code-no-prism@^0.41.0"
  }
}

Custom Tokenizer

You can provide your own tokenizer (e.g., Shiki) via the Tokenizer interface:

import { registerCodeHighlighting, type Tokenizer } from '@lexical/code';

const myTokenizer: Tokenizer = {
  defaultLanguage: 'javascript',
  tokenize(code, language) { /* return tokens */ },
  $tokenize(codeNode, language) { /* return LexicalNode[] */ },
};

registerCodeHighlighting(editor, myTokenizer);

License

MIT (forked from facebook/lexical)