Import or query hunspell?

Question

Hi,
We have a hunspell-based spell-checking dictionary. Other than
trying to programatically generate a humongous word-list to cover every possible combination of root+affix (+affix) that it allows, is there any way of getting Paratext to use it?

anon542642

Ancienne publication - affichée dans sa langue d'origine

Answer 1

I don’t remember how I did it… I could try to look up notes if needed… but I expanded a hunspell dictionary into a PT wordlist using some pre-existing program for creating all possible combinations. The resulting file was large, but works perfectly fine. Is there a reason you want to use the dictionary directly instead of going that route?

Ancienne publication - affichée dans sa langue d'origine

Comments

mnjames
February 14

I don’t remember how I did it… I could try to look up notes if needed… but I
expanded a hunspell dictionary into a PT wordlist using some pre-existing
program for creating all possible combinations. The resulting file was large,
but works perfectly fine.

Practically speaking, the resultant list will be getting very large - every
verb potentially turns into multiple participles, and every participle can act
as a nominal and each then end up in sg/plural versions in nominative,
oblique(accusative), locative, dative, ablative, genitive (aggreeing with
gender and number of the posessed item) cases and that doesn’t add up to the
right number so I’ve forgotten some.

Is there a reason you want to use the dictionary
directly instead of going that route?

Ideally, yes:
The dictionary is still being worked on, and sometimes errors are
spotted, or decisions about spelling get revisited, especially for
loan words where there are several options about which noun class
they might go in: that’s a language-community-preference thing
that might develop over time.

Ideally I’d hope there could be a way for paratext’s current version
of manually checked words (e.g. for names of one-off towns, ancient kings,
etc) to sit beside the hunspell dict.

I’ve never looked at the ParaText code, but I’d assume there would be benefits
in using the Enchant library (https://abiword.github.io/enchant/),
as that’s backend-agnostic.

anon542642

Ancienne publication - affichée dans sa langue d'origine

---

There is the laboursome possibility to do spellchecking in Indesign, (which supports hunspell). It means exporting via PubAssist into Indesign, and then do a round-trip back to sfm files.

Ancienne publication - affichée dans sa langue d'origine