I was messing around with trying to getting nltk to work, but quickly ran into the issue that importing nltk fails right off the bat because some of the corpus modules (like panlex_lite) import sqlite3, but pyodide specifically excludes sqlite3 in remove_modules.txt. Even if you just want to use a sentence lexer, you’re out of luck because the lexers end up depending on some of those corpora.

I played around a bit with including a WASMized sqlite3 (e.g.:
https://github.com/mandel59/sqlite-wasm or https://github.com/kripken/sql.js/) but didn’t get too far.

Wondering if it’s worth the effort to get it working?