Use CFStringGetCharacters instead of calling CFStringGetBytes twice

[AreaZR]

This can get the character information much faster.

[al45tair]

This can get the character information much faster.

Do you have measurements to prove that? The issue here is that there's a good chance the CFStrings we're using here are ASCII, in which case doing CFStringGetCharacters() is going to convert to UTF-16 and then String() is going to convert that back to UTF-8, plus the buffer you've allocated in the meantime is twice as large as needed.

[AreaZR]

This can get the character information much faster.

Do you have measurements to prove that? The issue here is that there's a good chance the CFStrings we're using here are ASCII, in which case doing CFStringGetCharacters() is going to convert to UTF-16 and then String() is going to convert that back to UTF-8, plus the buffer you've allocated in the meantime is twice as large as needed.

if let ptr = CFStringGetCStringPtr is the fast path. That does UTF-8. Also UTF-16 is how they are natively stored.

[al45tair]

if let ptr = CFStringGetCStringPtr is the fast path. That does UTF-8. Also UTF-16 is how they are natively stored.

Well, CFStringGetCStringPtr() is the fast path, since CFString stores ASCII bytes in many cases. I like the CFStringGetCharactersPtr() call, because in the case where the CFString is UTF-16, that might indeed be faster, but using CFStringGetCharacters() rather than CFStringGetBytes() is unlikely to be any faster and might actually be slower. Why? Because if CFStringGetCharactersPtr() fails, it means the string storage isn't contiguous UTF-16. In the case, there's some kind of conversion going to happen, and if you use CFStringGetCharacters() then your buffer is twice the size and you've got to traverse it at least twice (once to fill it, maybe converting to UTF-16 in the process, and once in String to convert it to UTF-8).

In addition, you said

This can get the character information much faster.

If that's true, you should have numbers to back it up.