Support for hunspell dictionaries

[Marko-98]

Hi everyone! 👋

I understand that Harper will be English-only at least for a longer period of time now. But I was wondering if you could add support for hunspell dictionaries.

This way, users of other languages could just import hunspell dictionary file for their language and still have a basic spell check until Harper fully implements support their language.

Thanks!

[hippietrail]

I'm pretty sure we'd add support for a second language if we had a dedicated "champion" for the language willing to do the work and testing needed. They would need a decent grasp of the grammar and/or linguistics of their language and preferably but not necessarily Rust coding experience.

Our dictionary format actually started out as the Hunspell dictionary format. It's tied to the annotation.json file in Harper's case.

But note that all (perhaps only almost all?) of the linters are English-specific. A quick hack do implement what you ask would be to simply turn off all the linters besides the spellcheck one, when an outside dictionary is used.

Note also that the English dictionary is compiled into the binary. So we'd need to choose between also building a separate binary with another language dictionary compiled in, or adding support to load other language dictionaries in dynamically. I experimented with this a few months ago for Spanish and realized that the LSP backend was loading the dictionary basically every keypress. @elijah-potter mentioned in another thread on the topic that it wouldn't be too hard to implement properly, and he knows the codebase a lot better than I do of course (-:

[Marko-98]

I'd be satisfied even with just basic spell check because current Firefox spell check implementation sucks and, unfortunately, there's no change in sight. 🫤