def _extract_signature_function(self, video_id, player_url, example_sig):
id_m = re.match(
- r'.*-(?P<id>[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player)?\.(?P<ext>[a-z]+)$',
+ r'.*?-(?P<id>[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player)?\.(?P<ext>[a-z]+)$',
player_url)
if not id_m:
raise ExtractorError('Cannot identify player %r' % player_url)
((?:PL|LL|EC|UU|FL|RD)[0-9A-Za-z-_]{10,})
)"""
_TEMPLATE_URL = 'https://www.youtube.com/playlist?list=%s'
- _MORE_PAGES_INDICATOR = r'data-link-type="next"'
_VIDEO_RE = r'href="\s*/watch\?v=(?P<id>[0-9A-Za-z_-]{11})&[^"]*?index=(?P<index>\d+)'
IE_NAME = 'youtube:playlist'
_TESTS = [{
'info_dict': {
'title': 'JODA7',
}
+ }, {
+ 'note': 'Buggy playlist: the webpage has a "Load more" button but it doesn\'t have more videos',
+ 'url': 'https://www.youtube.com/playlist?list=UUXw-G3eDE9trcvY2sBMM_aA',
+ 'info_dict': {
+ 'title': 'Uploads from Interstellar Movie',
+ },
+ 'playlist_mincout': 21,
}]
def _real_initialize(self):
'Downloading page #%s' % page_num,
transform_source=uppercase_escape)
content_html = more['content_html']
+ if not content_html.strip():
+ # Some webpages show a "Load more" button but they don't
+ # have more videos
+ break
more_widget_html = more['load_more_widget_html']
playlist_title = self._html_search_regex(
class YoutubeChannelIE(InfoExtractor):
IE_DESC = 'YouTube.com channels'
_VALID_URL = r'https?://(?:youtu\.be|(?:\w+\.)?youtube(?:-nocookie)?\.com)/channel/(?P<id>[0-9A-Za-z_-]+)'
- _MORE_PAGES_INDICATOR = 'yt-uix-load-more'
- _MORE_PAGES_URL = 'https://www.youtube.com/c4_browse_ajax?action_load_more_videos=1&flow=list&paging=%s&view=0&sort=da&channel_id=%s'
IE_NAME = 'youtube:channel'
_TESTS = [{
'note': 'paginated channel',
return self.playlist_result(entries, channel_id)
def _entries():
+ more_widget_html = content_html = channel_page
for pagenum in itertools.count(1):
- url = self._MORE_PAGES_URL % (pagenum, channel_id)
- page = self._download_json(
- url, channel_id, note='Downloading page #%s' % pagenum,
- transform_source=uppercase_escape)
- ids_in_page = self.extract_videos_from_page(page['content_html'])
+ ids_in_page = self.extract_videos_from_page(content_html)
for video_id in ids_in_page:
yield self.url_result(
video_id, 'Youtube', video_id=video_id)
- if self._MORE_PAGES_INDICATOR not in page['load_more_widget_html']:
+ mobj = re.search(
+ r'data-uix-load-more-href="/?(?P<more>[^"]+)"',
+ more_widget_html)
+ if not mobj:
break
+ more = self._download_json(
+ 'https://youtube.com/%s' % mobj.group('more'), channel_id,
+ 'Downloading page #%s' % (pagenum + 1),
+ transform_source=uppercase_escape)
+ content_html = more['content_html']
+ more_widget_html = more['load_more_widget_html']
+
return self.playlist_result(_entries(), channel_id)
feed_entries = []
paging = 0
for i in itertools.count(1):
- info = self._download_json(self._FEED_TEMPLATE % paging,
- '%s feed' % self._FEED_NAME,
- 'Downloading page %s' % i)
+ info = self._download_json(
+ self._FEED_TEMPLATE % paging,
+ '%s feed' % self._FEED_NAME,
+ 'Downloading page %s' % i,
+ transform_source=uppercase_escape)
feed_html = info.get('feed_html') or info.get('content_html')
load_more_widget_html = info.get('load_more_widget_html') or feed_html
m_ids = re.finditer(r'"/watch\?v=(.*?)["&]', feed_html)