def handle_starttag(self, tag, attrs) :
if tag == 'a' :
for attr in attrs :
if attr[0] == 'href' :
joined = urlparse.urljoin(self.baseurl, attr[1])
#print joined
if not joined.startswith('http') :
break
if joined.endswith('.torrent') :
# ok, this should be torrent
self.href = 1
self.cur_link = (joined, None)
else :
try :
link = urlopen(joined,referer=self.baseurl)
except Exception:
break
if is_torrent(link):
self.href = 1
m = None
if link.info().has_key('Content-Disposition') :
m = content_dispo_re.search(link.info()['Content-Disposition'])
if m :
self.cur_link = (joined, m.group('filename'))
else :
self.cur_link = (joined, None)
break
def fetch(self):
self.prefetch()
media_list = self.filter.process(self.tracker)
if self.policy(policy.VERIFY_LINK):
for media in media_list:
if not (is_movie(media.link) or is_torrent(media.link)) :
self.log.warn('link failed: link is not movie or torrent: %s\n' % str(media.link))
return media_list