Improve encoding detection.

This commit is contained in:
Kovid Goyal 2008-02-25 17:28:13 +00:00
parent f94102bc87
commit dc11f5051f

View File

@ -344,7 +344,10 @@ def entity_to_unicode(match, exceptions=[], encoding='cp1252'):
num = int(ent[1:]) num = int(ent[1:])
if encoding is None or num > 255: if encoding is None or num > 255:
return unichr(num) return unichr(num)
return chr(num).decode(encoding) try:
return chr(num).decode(encoding)
except UnicodeDecodeError:
return unichr(num)
try: try:
return unichr(name2codepoint[ent]) return unichr(name2codepoint[ent])
except KeyError: except KeyError: