[youtube:tab] Fix extraction with cookies provided (closes #27005)

This commit is contained in:
Sergey M․ 2020-11-17 03:40:02 +07:00
parent 91dcde8a38
commit 6699b6ce41
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D
1 changed files with 13 additions and 6 deletions

View File

@ -2723,7 +2723,7 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor):
'itct': ctp, 'itct': ctp,
} }
def _entries(self, tab): def _entries(self, tab, identity_token):
continuation = None continuation = None
slr_contents = tab['sectionListRenderer']['contents'] slr_contents = tab['sectionListRenderer']['contents']
for slr_content in slr_contents: for slr_content in slr_contents:
@ -2768,16 +2768,20 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor):
if not continuation: if not continuation:
continuation = self._extract_continuation(is_renderer) continuation = self._extract_continuation(is_renderer)
headers = {
'x-youtube-client-name': '1',
'x-youtube-client-version': '2.20201112.04.01',
}
if identity_token:
headers['x-youtube-identity-token'] = identity_token
for page_num in itertools.count(1): for page_num in itertools.count(1):
if not continuation: if not continuation:
break break
browse = self._download_json( browse = self._download_json(
'https://www.youtube.com/browse_ajax', None, 'https://www.youtube.com/browse_ajax', None,
'Downloading page %d' % page_num, 'Downloading page %d' % page_num,
headers={ headers=headers, query=continuation, fatal=False)
'x-youtube-client-name': '1',
'x-youtube-client-version': '2.20201030.01.00',
}, query=continuation, fatal=False)
if not browse: if not browse:
break break
response = try_get(browse, lambda x: x[1]['response'], dict) response = try_get(browse, lambda x: x[1]['response'], dict)
@ -2848,8 +2852,11 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor):
title = channel_title or channel_id title = channel_title or channel_id
if tab_title: if tab_title:
title += ' - %s' % tab_title title += ' - %s' % tab_title
identity_token = self._search_regex(
r'\bID_TOKEN["\']\s*:\s*["\'](.+?)["\']', webpage,
'identity token', default=None)
return self.playlist_result( return self.playlist_result(
self._entries(selected_tab['content']), self._entries(selected_tab['content'], identity_token),
playlist_id=channel_external_id or channel_id, playlist_id=channel_external_id or channel_id,
playlist_title=title) playlist_title=title)