On Wed, Jul 03, 2024 at 11:35:10AM -0700, Darrick J. Wong wrote: > I'm not sure how exactly to write a classifier here -- the 'invisible' > and 'zero width' ones are obvious, but the 'joiner' code points don't > seem to have any obvious trend to them. > > For now I think I'll take the "conservative" approach and only flag > things that sound like they're supposed to be general metacharacters, > and leave out the modifier codepoints that are ok if they're surrounded > by certain codepoints. But this is a rather manual process. Oh, right. There is no clear identification and you are just doing a manual search based on viѕual output. Yes, there's unfortunately no really good way to automate that.
