Closed dhdaines closed 1 month ago
Note also that you could also just add {r'\w'}
to all_word_characters in the same way as you do for the default pipeline.
In actual fact we should add \w
to them, because otherwise they will remove numbers at the end of search terms, which is almost certainly not what you want for a lot of applications! But... this is not bug-compatible with lunr-languages, so it might just need a documented workaround.
You may not really want to do this, it seems the trimmers in lunr-languages are full of weird junk: https://github.com/MihaiValentin/lunr-languages/issues/66
Hmm. It turns out, actually, that lunr-languages code is generated programmatically as well. So it doesn't make a lot of sense to parse it to create these. I'm closing this PR and will come up with a better way to do this.
See #149 (doesn't fix the whole thing)