lingua-py

Why we included this project

Knowing what language a piece of text is written in matters before you can do almost anything else with it. Lingua answers that question reliably even for single words and short mixed-language snippets, the case where most detection libraries fall apart. It slots in as a preprocessing step ahead of text classification, spell checking, or routing support emails to the right regional team, and needs no neural network or word dictionaries to do it. The methods are rule-based plus statistical Naive Bayes, and the Python bindings sit on a Rust core, so it stays fast, light, and fully offline once installed. You get a small focused library instead of a large ML framework to learn.