LT

LT(LLM Translator) is autonomous translate agent that helps with Linux Kernel docs' translation. It supports block-based translation of large documents, translation integrity detection, and targeted retranslation (only retranslating missing segments to reduce costs).

Project Structure

• src/llm_factory.py — Qwen/OpenAI client wrapper and QwenChatModel.
• src/text_chunker.py — Markdown chunker based on structure and token limits.
• src/translator.py — Main translation workflow: chunking, translation, merging, summary comparison, and focused retranslation.
• src/summary_generator.py — Generates summaries and performs source/translation comparison.
• src/markdown_parser.py — Parses YAML front matter and content.
• src/rst_processor.py — reStructuredText-specific processor (block parsing/reconstruction).
• src/universal_translator.py — 通用编排器，按文件格式选择处理器并管理逐块/整篇翻译与重译策略。
• src/config_manager.py — Configuration and API key management.
• tests/ — Various test scripts.
• requirements.txt — Python dependencies.
• main.py — Command-line interface for translation.

Attention

The api key is configured in config.ini.

Run locally

git clone https://github.com/open-atom-hust-club/LT.git
cd LT
pdm install
python main.py --model qwen-plus --provider qwen translate tests/ldm.md -o tests/ldm_translated.md

# 单独运行 RST 测试（不会与 Markdown 的 `pdm run test` 冲突）
# 已在 `pyproject.toml` 中添加了一个独立脚本 `test-rst`：

pdm run test-rst

或显式运行：

pdm run python main.py --model qwen-plus --provider qwen translate tests/w1-generic.rst -o tests/w1-generic_translated.rst

Testing

Run the translation test using PDM:

pdm test         # 运行 Markdown 测试 (ldm.md)
# 或运行 RST 专用测试：
pdm run test-rst  # 运行 RST 测试 (w1-generic.rst)

This command will translate the test Markdown file tests/ldm.md to tests/ldm_translated.md using the qwen-plus model.