[youtube:channel] Fix the extraction of autogenerated channels

The ajax pages are empty, now it looks directly in the channel's /videos page
This commit is contained in:
Jaime Marquínez Ferrándiz 2013-11-15 11:51:45 +01:00
parent feee2ecfa9
commit b9643eed7c
1 changed files with 24 additions and 13 deletions

View File

@ -1594,20 +1594,31 @@ class YoutubeChannelIE(InfoExtractor):
# Download channel page
channel_id = mobj.group(1)
video_ids = []
url = 'https://www.youtube.com/channel/%s/videos' % channel_id
channel_page = self._download_webpage(url, channel_id)
if re.search(r'channel-header-autogenerated-label', channel_page) is not None:
autogenerated = True
else:
autogenerated = False
# Download all channel pages using the json-based channel_ajax query
for pagenum in itertools.count(1):
url = self._MORE_PAGES_URL % (pagenum, channel_id)
page = self._download_webpage(url, channel_id,
u'Downloading page #%s' % pagenum)
page = json.loads(page)
ids_in_page = self.extract_videos_from_page(page['content_html'])
video_ids.extend(ids_in_page)
if self._MORE_PAGES_INDICATOR not in page['load_more_widget_html']:
break
if autogenerated:
# The videos are contained in a single page
# the ajax pages can't be used, they are empty
video_ids = self.extract_videos_from_page(channel_page)
else:
# Download all channel pages using the json-based channel_ajax query
for pagenum in itertools.count(1):
url = self._MORE_PAGES_URL % (pagenum, channel_id)
page = self._download_webpage(url, channel_id,
u'Downloading page #%s' % pagenum)
page = json.loads(page)
ids_in_page = self.extract_videos_from_page(page['content_html'])
video_ids.extend(ids_in_page)
if self._MORE_PAGES_INDICATOR not in page['load_more_widget_html']:
break
self._downloader.to_screen(u'[youtube] Channel %s: Found %i videos' % (channel_id, len(video_ids)))