Qwen Code Translator

A documentation translation tool specifically designed for the github project. Automatically sync documentation from GitHub repositories, translate with Qwen AI, and build a multilingual Nextra documentation site.

Features

• 🌍 Multi-language Support: Translate documentation to Chinese (zh), German (de), French (fr), Russian (ru), Japanese (ja), and Spanish (es)
• 🤖 Qwen AI Translation: Powered by Qwen API for high-quality technical document translation
• 📚 Nextra Integration: Automatically generates a modern documentation site using Nextra
• 🔄 Git Synchronization: Automatically syncs with source repositories to keep translations up-to-date
• ⚡ CLI Interface: Easy-to-use command-line interface for all operations
• 📝 Smart Translation: Preserves code blocks, links, and technical terms while translating content
• 🚀 Parallel Processing: Translates multiple languages concurrently for faster processing

Installation

npm install -g @qwen-code/translator

Quick Start

1. Initialize a new translation project:

   qwen-translator init

2. Configure environment variables:

   cp .env.example .env
   # Edit .env file and add your Qwen API key

3. Sync source repository documents:

   qwen-translator sync

4. Translate documents:

   qwen-translator markdown

5. Start the documentation site:

   npm install
   npm run dev

Commands

init

Initialize a new translation project with interactive configuration. Sets up project structure, copies Nextra template, and creates configuration files.

sync [options]

Sync source repository documents and automatically translate changed files into the target languages.

• -f, --force: Force sync all documents (ignores previous sync records)
• -d, --detect-only: Only detect and list changed files; skip translation and write nothing (no content/ updates, no last-sync.json/changelog changes). Useful for previewing what a sync would translate.
• -s, --source-only: Detect changes and write the source-language docs (content/<sourceLanguage> and .source-docs), but skip translation. Does not advance last-sync.json or write the changelog, so a later sync (with a key) still re-detects and translates these files. Useful for refreshing the source docs now and translating later.

No API key for detection / source-only. The translator is lazy-loaded, so --detect-only, --source-only, and a normal sync that finds zero changes do not require OPENAI_API_KEY. A normal sync that has changes validates the key before writing any files, so a missing key fails fast without leaving partially-written output. Note that detection still clones/updates the source repo under .temp-source-repo.

markdown [options]

Translate the Markdown documents under content/<sourceLanguage>/ into the target languages. Translates the whole directory by default, or a single file with -f.

• -l, --language <lang>: Specify target language (zh, de, fr, ru, ja, pt-BR, es)
• -f, --file <file>: Specify single file to translate

meta [options]

Translate Nextra navigation files (_meta.ts) into the target languages.

• -l, --language <lang>: Specify target language
• -f, --file <file>: Specify a single _meta.ts file to translate

config

View and manage project configuration interactively.

status

Show current project status, configuration, and environment setup.

Configuration

The tool creates a translation.config.json file during initialization:

{
  "name": "project-name",
  "sourceRepo": "https://github.com/QwenLM/qwen-code.git",
  "docsPath": "docs",
  "sourceLanguage": "en",
  "targetLanguages": ["zh", "de", "fr", "ru"],
  "outputDir": "content"
}

Environment Variables

Create a .env file with the following variables:

# Required: OpenAI-compatible API key
OPENAI_API_KEY=your_api_key_here

# Optional: API configuration (defaults shown)
OPENAI_BASE_URL=https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1
QWEN_MODEL=deepseek-v4-flash
QWEN_MAX_TOKENS=32768

Alternative API Endpoints

You can also use other compatible endpoints:

# DashScope (Aliyun)
OPENAI_BASE_URL=https://dashscope.aliyuncs.com/compatible-mode/v1/
QWEN_MODEL=qwen-turbo

Requirements

• Node.js >= 18.0.0
• npm >= 8.0.0
• OpenAI-compatible API key

Project Structure

After initialization, your project will have:

├── content/
│   ├── en/           # Source language documents (synced from repo)
│   ├── zh/           # Chinese translations
│   ├── de/           # German translations
│   ├── fr/           # French translations
│   └── ru/           # Russian translations
├── app/              # Next.js app directory (from template)
│   ├── [lang]/       # Dynamic language routing
│   └── layout.tsx    # App layout
├── .source-docs/     # Raw synced documentation
├── .temp-source-repo/# Temporary git clone (auto-managed)
├── translation.config.json
├── translation-changelog.json  # Translation history
├── last-sync.json    # Sync tracking
├── .env.example
├── next.config.mjs
└── package.json

How It Works

1. Sync: The tool clones/updates the source repository and detects changed files
2. Parse: Markdown documents are parsed while preserving structure and metadata
3. Translate: Content is segmented and translated using Qwen AI, with smart handling of code blocks and technical terms
4. Generate: Translated content is organized into a Nextra-based multilingual documentation site
5. Serve: The Next.js app serves the documentation with language switching

Development

# Install dependencies
npm install

# Build the CLI tool
npm run build

# Run in development mode (watch mode)
npm run dev

# Run tests
npm test

License

MIT

Contributing

This tool is specifically designed for the Qwen Code project. For feature requests or bug reports, please open an issue in the project repository.