nojhan/weboob-devel
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

correctly find title

Browse source
This commit is contained in:
Romain Bignon 2011-08-29 20:18:22 +02:00
commit 98cf41b454
1 changed files with 4 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

8
weboob/backends/phpbb/pages/forum.py
View file

@ -95,7 +95,7 @@ class TopicPage(PhpBBPage):
self.tot_pages = int(strongs[1].text.strip()) self.tot_pages = int(strongs[1].text.strip())
try: try:
url = self.parser.select(self.document.getroot(), 'h2 a', 1).attrib['href'] url = self.document.xpath('//h2/a')[-1].attrib['href']
except BrokenPageError: except BrokenPageError:
url = self.url url = self.url
v = urlsplit(url) v = urlsplit(url)
@ -159,14 +159,14 @@ class TopicPage(PhpBBPage):
id = div.attrib['id'][1:] id = div.attrib['id'][1:]
post = Post(self.forum_id, self.topic_id, id) post = Post(self.forum_id, self.topic_id, id)
title_tags = body.cssselect('h3 a') title_tags = body.xpath('//h3/a')
if len(title_tags) == 0: if len(title_tags) == 0:
title_tags = self.document.getroot().cssselect('h2 a') title_tags = self.document.xpath('//h2/a')
if len(title_tags) == 0: if len(title_tags) == 0:
title = u'' title = u''
self.logger.warning('Unable to parse title') self.logger.warning('Unable to parse title')
else: else:
title = title_tags[0].text.strip() title = title_tags[-1].text.strip()
post.title = self.forum_title + title post.title = self.forum_title + title
for a in profile.cssselect('dt a'): for a in profile.cssselect('dt a'):