import re
from sgmllib import SGMLParser, SGMLParseError
from BeautifulSoup import BeautifulSoup, CData
+
from .bkmk_ph_util import HTMLParser
+from compat import string_type
DEFAULT_CHARSET = "cp1251" # Stupid default for Russian Cyrillic
def parse_html(html_text, charset=None, log=None):
+ if not html_text:
+ return None
root = _parse_html(html_text, charset)
if root is None:
return None
else:
parts = []
for part in title:
- if not isinstance(part, basestring):
+ if not isinstance(part, string_type):
part = part.decode()
parts.append(part.strip())
title = ''.join(parts)
if meta_content:
meta_charset = _charset = meta_content.lower()
- if title and (_charset or meta_charset):
- title = title.encode(_charset or meta_charset)
+ #if title and (_charset or meta_charset):
+ # title = title.encode(_charset or meta_charset)
meta = head.find(_find_refresh, recursive=False)
if meta: