package for Belarusian orthography conversion. The engine behind the taraskevizatar web app

Content

The package converts the academic orthography to the classical one. The main function is tarask, there are also alphabetic and experimental phonetic pipelines, splitIntoChunks for worker-ready chunks, and the TaraskConfig class with options for the alphabet (cyrillic, latin, arabic), i by j replacing, variations, wrappers and the g letter.

It ships in 5 forms: an library with types, a CDN dle with a global variable that works synchronously without initialization, a tarask CLI with prebuilt binaries that streams multi-gigabyte inputs in parallel, library, and JSON dictionaries in the releases.

 install taraskevizer
tarask --latin ’планета’
tarask < ./cyr-text.txt > ./lat-text.txt

History

After a while, someone asked me in issues on to move logic of conversion to an -package. And I moved this with json-generation to another repository. For building I chose , it’s simple builder, that allows use plugins. It was simpler to create a plugin for json-generation with than with .

Then, I set up CI/CD with . The workflow is building project and publish it as -package, after successful publication it triggers a workflow in the app repository, that makes build and publish to .

Later I replaced dling with transpiling and removed from the deps. In the autumn of 2026 I moved the conversion core to compiled to , so the browser conversion became 3-4 times faster, and added the prebuilt tarask binaries to every release.