CEFR Highlight

A VSCode/Cursor extension that highlights English vocabulary by CEFR level (A1–C2) using semantic tokens. Hover over any word to see its level, part of speech, and Chinese translation.

Features

• Semantic highlighting — words are colored by CEFR level in .cefr files
• Phrase detection — multi-word phrases are recognized with higher priority than individual words and given a background highlight
• Hover tooltips — hover over a word or phrase to see its CEFR level, part of speech, and Chinese translation (via Google Translate)
• Configurable — customize colors, minimum level threshold, and phrase background

Color Legend

Level	Default Color	Emoji	Description
A1	#BDBDBD Grey	⚪	Beginner
A2	#43A047 Green	🟢	Elementary
B1	#1E88E5 Blue	🔵	Intermediate
B2	#8E24AA Purple	🟣	Upper-intermediate
C1	#CDDC39 Lime	🟡	Advanced
C2	#FB8C00 Orange	🟠	Proficiency

By default, only words at B2 and above are highlighted.

Configuration

All settings live under cefrHighlight.* in your VSCode/Cursor settings:

Setting	Default	Description
cefrHighlight.minimumLevel	B2	Only highlight words at or above this level
cefrHighlight.colors.A1	#BDBDBD	Foreground color for A1 words
cefrHighlight.colors.A2	#43A047	Foreground color for A2 words
cefrHighlight.colors.B1	#1E88E5	Foreground color for B1 words
cefrHighlight.colors.B2	#8E24AA	Foreground color for B2 words
cefrHighlight.colors.C1	#CDDC39	Foreground color for C1 words
cefrHighlight.colors.C2	#FB8C00	Foreground color for C2 words
cefrHighlight.phraseBackground	rgba(255, 235, 59, 0.15)	Background color for multi-word phrases

Example — show only C1+ words:

{
  "cefrHighlight.minimumLevel": "C1"
}

Vocabulary Data

The CEFR index contains ~72,800 words and ~3,000 phrases, merged from four sources (in priority order):

1. CEFR-J Vocabulary Profile — 7,020 manually curated entries by language researchers
2. EFLLex — 13,843 frequency-graded lemmas from the UCLouvain graded reader corpus
3. Oxford 14K — 10,618 words classified via ML on Oxford 5000 features
4. Maximax67 Words-CEFR-Dataset — 62,238 entries computed from Google N-Gram frequencies, filtered to remove obscure words

Architecture

cefr-highlight/
├── server/          Rust language server (tower-lsp)
│   ├── src/
│   │   ├── main.rs       LSP lifecycle, semantic tokens, hover + translation
│   │   ├── cefr.rs       CEFR index loading, lookup with inflection handling
│   │   ├── tokenizer.rs  Word/phrase tokenization with longest-match priority
│   │   └── translate.rs  Google Translate free API client with in-memory cache
│   └── data/
│       └── cefr_index.json   Vocabulary database (embedded at compile time)
├── client/          TypeScript extension client
│   └── src/
│       └── extension.ts  Language client, color sync, phrase decorations
└── package.json     Extension manifest and configuration schema

• Server (Rust / tower-lsp): loads the CEFR dictionary at startup, tokenizes documents, returns semantic tokens and hover info via LSP, fetches Chinese translations on demand
• Client (TypeScript / vscode-languageclient): spawns the server binary over stdio, syncs color configuration to editor semantic token rules, applies phrase background decorations via a custom cefr/phraseRanges notification

Building

# Build the language server
cd server && cargo build --release

# Build the extension client
cd client && npm install && npm run compile

Installation

After building, symlink the project into your extensions directory:

ln -s /path/to/cefr-highlight ~/.cursor-server/extensions/kamome.cefr-highlight-0.3.0

Then reload the window (Ctrl+Shift+P → Developer: Reload Window).

Usage

1. Create or open a file with the .cefr extension
2. Words at or above the configured minimum CEFR level will be highlighted
3. Hover over any highlighted word to see its level, part of speech, and Chinese translation

Development

Press F5 in VSCode/Cursor to launch the Extension Development Host.

License

MIT