def _load_doc(self, url, html):
try:
dom = page_parser.parse_unicode(html, url, notify=logging.info)
except:
logging.error("raw doc can not be parsed", url)
return None
return dom
def load_dom(url, encoding = None):
body = load_body(url, encoding)
if body == None:
return None
dom = page_parser.parse_unicode(body, url)
return dom
