[Pitch] Regex-powered string processing algorithms

[benlings]

@Michael_Ilseman

contains and starts(with:) have existing analogues and it makes sense to extend them for regex. The ??? entry is the ~= overload as the analogue for == (or elementsEqual )

I think it would be unfortunate if the only way to spell this operation would be with the ~= operator. The (only?) other override of ~= that I could find in the stdlib is for Range, where it is effectively an alias for .contains(...). Would it make sense to add a Regex analogue of elementsEqual?

extension BidirectionalCollection where SubSequence == Substring {
  public func elementsEqual(_ other: some RegexComponent) -> Bool 
}

I'll admit that the naming is much less clear than for contains(_:) and starts(with:), but it is at least consistent. ~= could then be an alias for this.

@Michael_Ilseman

I think whole-match is the most useful and intuitive semantics here (and consistent with e.g. switching over an array of characters). A match-anywhere is surprising as it's easy for a regex to accidentally have some match somewhere inside an unknown or untested input.

Yours and @Karl 's arguments have convinced me that this is the right choice .

@Michael_Ilseman

Match-anywhere should probably be opt-in and I am interested in making opting-in more ergonomic.

@Michael_Ilseman

I think you'd need to work around the lack of variadic generics by generating a ton of overloads like we're doing . We definitely want a better way, and that's future work at this point.

It would definitely be nice to make abstracting over composing regexes easier. With the current proposals, it's possible to compose regexes (which is already a huge improvement over what's available elsewhere). As I discovered above, putting this composition into a function is much harder! Your idea for regex literal interpolation looks like it might help with this?