Implement Unicode Character Classes in Brackets

[raskad]

This PR implements Unicode Character Classes in Brackets.

[ridiculousfish]

This change looks good, thank you! However rather than tracking the unicode properties as an extra field(s) in BracketContents, we should instead replace the character class with its associated ranges. That goes for both CharacterClass and UnicodePropertyEscape. That will both speed up and simplify matching.

Probably the place to do it is immediately, during parse(), so that we don't have to carry around the character classes in the IR.

You can make that change as part of this, or we can defer it and do it later, and you can land what you have.

[raskad]

This change looks good, thank you! However rather than tracking the unicode properties as an extra field(s) in BracketContents, we should instead replace the character class with its associated ranges. That goes for both CharacterClass and UnicodePropertyEscape. That will both speed up and simplify matching.

Probably the place to do it is immediately, during parse(), so that we don't have to carry around the character classes in the IR.

Very nice idea. That would probably also enable some range optimizations in some cases. My only concern would be that that would copy the static range tables and in some cases potentially increase the size of the compiled regex a lot.

I will test that out on a different branch :)