rtf reader: unichr() causes ValueError () about pyth HOT 3 CLOSED

This is a legitimate bug but I'm not sure what the correct fix is. It's possible to construct a surrogate pair to represent the character, with something like:

struct.pack('<L', 0x10000).decode('utf-32')

Which will "work", but perhaps cause other bugs down the line with e.g. slicing. So maybe we shouldn't do it.

Other alternatives include replacing it with '?', or just raising a different exception type. I haven't decided.

from pyth.

I like the solution: replacing with ? + log message

from pyth.

I've tried the struct trick above. It means that plugins should never trust len() of unicode strings, or slice them. But that's probably true anyway.

from pyth.