guard ctype calls against out-of-range wide characters

[zayeem06]

Out-of-bounds ctype read on wide input characters

The C runtime passes UTF16/UTF32 code points straight into isprint()/isdigit(). On a lexer error displayRecognitionError calls isprint(ex->c), where ex->c is filled from is->_LA() on the char stream, so with a UTF16/UTF32 input stream it can be any code point up to 0x10FFFF. printableUTF16 and toInt32_UTF16 do the same on their UINT16 characters. The C standard only allows an argument representable as unsigned char or EOF, so anything above 255 reads past the ctype classification table.

I gated each call with a <= 255 range check so wide characters skip the classification instead of indexing out of bounds. In-range bytes keep the exact same path, and a wide char in displayRecognitionError now falls through to the existing hex branch. Worth a look at whether keeping the wide char as-is rather than escaping it in printableUTF16 matches what you'd want there.