I've built Locawise, an open-source Python tool to automate the often tedious process of application localization. It uses AI (OpenAI or VertexAI models) to translate new or changed strings in your language files.
It consists of two main parts:
locawise: A Python CLI tool you run locally. It detects changes in your source language files (JSON, .properties), gets translations, and updates target files. It's context-aware – you can define project context, terminology (glossary), and tone via a YAML config (i18n.yaml) to improve translation quality. locawise-action: A GitHub Action that automates this process. It can run on pushes to your main branch and create a PR with the new translations. The goal is to offer a free, developer-controlled alternative to paid localization platforms. It's designed to be efficient (async, only translates changes) and cost-effective (you control LLM usage).
GitHub (CLI tool): https://github.com/aemresafak/locawise GitHub (Action): https://github.com/aemresafak/locawise-action Quick Tutorial: https://www.youtube.com/watch?v=b_Dz68115lg
I'd love to get your feedback and answer any questions.
aemresafak•8mo ago
Thanks for checking out Locawise. I started building this because I was looking for a more streamlined and cost-effective way to handle localization for my own projects. I wanted something that could leverage the power of modern LLMs for good quality translations but also give me control over context and specific terminology without being locked into an expensive subscription.
The core idea was to make localization an almost invisible part of the development workflow, especially with the GitHub Action – push your code in the source language, and let the translations follow automatically.
This is still evolving, and I'm keen to build it out based on what developers actually need. Looking forward to your thoughts and any suggestions you might have!