Allow urlparse and urlsplit of bytes URLs

[koogoro]

Right now, urlparse(url) and urlsplit(url) doesn't work if url is a bytes. This is because the only overload that can match this case is the one where url: bytes | bytearray, but even this one doesn't match since the default value for scheme is "", which isn't bytes | bytearray | None. Adding Literal[""] to the types for scheme won't cause the overloads to overlap, since they're differentiated by the type of url, and will allow urlparse applied to a single bytes argument to work again.

[hauntsaninja]

Thanks, this matches the runtime better!

[AlexWaygood]

Thanks! Looks like we could actually combine the first two overloads of urlparse and urlsplit in that case — want to do that?

[koogoro]

Is there an easy way to do that? It seems hard to me since they still have different return types, but if there's a good way to combine them then I'll do it.

[AlexWaygood]

Is there an easy way to do that? It seems hard to me since they still have different return types, but if there's a good way to combine them then I'll do it.

Good point — we can't combine the overloads for urlparse, but we can combine three overloads urlsplit, where the return types are the same for overloads two and three

[AlexWaygood]

Sorry, I meant to combine the second and third overloads, not the first two overloads! Apologies for the garbled review

[AlexWaygood]

Thanks!

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉