import codecs
import os
import re
-from htmlentitydefs import name2codepoint
+try:
+ from html.entities import name2codepoint
+except ImportError:
+ from htmlentitydefs import name2codepoint
from compat import unicode, unichr
def parse_filename(filename, charset=None, log=None):
- fp = open(filename, 'r')
+ fp = open(filename, 'rt')
try:
parser = parse_html(fp.read(), charset=charset, log=log)
finally: