qingjian-team/qingjian

Qingjian is a pinyin input method written in Rust that displays translations of candidate words in a language you are learning alongside the normal input interface.

The tool solves the problem of integrating language learning into daily typing by showing translations in English, Japanese, or Spanish next to candidate characters as you input. It works by running pinyin-to-character conversion, word lookup, and a local sentence model entirely on your device. The translation appears as auxiliary information that does not interfere with normal text input. The input method supports whole-sentence input, abbreviated pinyin, spelling correction, shuangpin, and wubi input modes, with the local model adjusting sentence candidates after pauses.

Qingjian suits anyone who types Chinese regularly and wants to passively learn vocabulary in another language without disrupting their workflow. The tool is designed for macOS and Windows, with a Linux version available through Fcitx5 that requires manual service startup. All processing happens locally—pinyin conversion, word lookup, the sentence model, and input statistics remain on your device and are never uploaded. The tool requires no account and allows you to disable or clear input logs independently of statistics tracking.

The project shows consistent development activity with regular updates and maintenance across its codebase. The team actively responds to issues and incorporates user feedback into releases. Documentation is comprehensive and maintained in Chinese, covering installation, input features, learning functionality, and keyboard shortcuts. The development approach emphasizes privacy by design, keeping all language models and user data local rather than relying on cloud services.