• apps/translate — CLI tool for translation, written in TypeScript
• ml/text/ — Machine learning models for text processing, written in Python
  • ./code_ner — Named Entity Recognition model for detecting inline code in MUI docs (not used at the moment)

JavaScript:

• yarn install
• cp .env.example .env and fill in the values

Python (not used at the moment):

• Install Anaconda, see https://www.anaconda.com/products/distribution
• Install anaconda-project: conda install anaconda-project
• Download dependencies: cd ml/text && anaconda-project prepare
• Enter the correct environment: conda activate envs/default

You will need a patched version of the MUI repo. Clone it from https://github.com/oversett/mui-repo.

To translate the entire repo (NB: just the docs/data/material folder is about 300K chars, so free DeepL API quota is not enough to translate the whole repo):

yarn translate mui repo ../mui-repo

To translate only a subset of the repo:

yarn translate mui repo ../mui-repo base/components/tabs   # can list multiple paths

To preview the translation (at http://localhost:3000):

cd ../mui-repo && yarn && yarn docs:dev

To deploy the translation, you also need to clone https://github.com/oversett/mui-translated. Here is an example for the ru translation:

cd ..
rm -rf mui-translated/ru
cp -R mui-repo mui-translated/ru  # Assuming mui-repo is already translated locally
rm -rf mui-translated/ru/.git

cd mui-translated
git add -A && git commit -m "Update ru translation"
git push

# Netlify should deploy automatically but the build fails for some reason.
# You need 'netlify-cli' to do a manual deploy.
cd ru
yarn
netlify build
netlify deploy --prod