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

fix parsing torrents list

Browse source
This commit is contained in:
Romain Bignon 2010-08-13 23:32:27 +02:00
commit a8c3fc166f
1 changed files with 10 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

15
weboob/backends/gazelle/pages/torrents.py
View file

@ -17,7 +17,7 @@
import re import re
from logging import warning from logging import warning, debug
from weboob.tools.misc import html2text from weboob.tools.misc import html2text
from weboob.tools.browser import BasePage from weboob.tools.browser import BasePage
@ -50,6 +50,9 @@ class TorrentsPage(BasePage):
table = table[0] table = table[0]
current_group = None current_group = None
for tr in table.findall('tr'): for tr in table.findall('tr'):
if tr.attrib.get('class', '') == 'colhead':
# ignore
continue
if tr.attrib.get('class', '') == 'group': if tr.attrib.get('class', '') == 'group':
tds = tr.findall('td') tds = tr.findall('td')
current_group = u'' current_group = u''
@ -72,7 +75,9 @@ class TorrentsPage(BasePage):
i = 0 i = 0
elif len(tds) in (8,9): elif len(tds) in (8,9):
# An alone torrent # An alone torrent
i = len(tds) - 7 i = len(tds) - 1
while i >= 0 and tds[i].find('a') is None:
i -= 1
else: else:
# Useless title # Useless title
continue continue
@ -87,8 +92,8 @@ class TorrentsPage(BasePage):
continue continue
id = id.group(1) id = id.group(1)
size = self.unit(*tds[i+3].text.split()) size = self.unit(*tds[i+3].text.split())
seeders = int(tds[i+5].text) seeders = int(tds[-2].text)
leechers = int(tds[i+6].text) leechers = int(tds[-1].text)
torrent = Torrent(id, torrent = Torrent(id,
title, title,
@ -98,7 +103,7 @@ class TorrentsPage(BasePage):
leechers=leechers) leechers=leechers)
yield torrent yield torrent
else: else:
print tr.attrib debug('unknown attrib: %s' % tr.attrib)
def get_torrent(self, id): def get_torrent(self, id):
table = self.document.getroot().cssselect('div.thin') table = self.document.getroot().cssselect('div.thin')