[youtube] Fix videos with age gate and encrypted signatures
authorCédric Luthi <cedric.luthi@gmail.com>
Mon, 29 Dec 2014 21:58:14 +0000 (22:58 +0100)
committerCédric Luthi <cedric.luthi@gmail.com>
Mon, 29 Dec 2014 21:58:14 +0000 (22:58 +0100)
The `sts` value is available on the embed webpage, get it from there.

Fixes #4108.

youtube_dl/extractor/youtube.py

index 550e18733e8070650e6cfe3a415cc3abe0112e9d..175e43272e8875e9efc6f91ae33f88d9f118e0b3 100644 (file)
@@ -766,11 +766,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
             age_gate = True
             # We simulate the access to the video from www.youtube.com/v/{video_id}
             # this can be viewed without login into Youtube
+            url = proto + '://www.youtube.com/embed/%s' % video_id
+            embed_webpage = self._download_webpage(url, video_id, 'Downloading embed webpage')
             data = compat_urllib_parse.urlencode({
                 'video_id': video_id,
                 'eurl': 'https://youtube.googleapis.com/v/' + video_id,
                 'sts': self._search_regex(
-                    r'"sts"\s*:\s*(\d+)', video_webpage, 'sts', default=''),
+                    r'"sts"\s*:\s*(\d+)', embed_webpage, 'sts', default=''),
             })
             video_info_url = proto + '://www.youtube.com/get_video_info?' + data
             video_info_webpage = self._download_webpage(
@@ -968,11 +970,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
                 elif 's' in url_data:
                     encrypted_sig = url_data['s'][0]
 
-                    if not age_gate:
-                        jsplayer_url_json = self._search_regex(
-                            r'"assets":.+?"js":\s*("[^"]+")',
-                            video_webpage, 'JS player URL')
-                        player_url = json.loads(jsplayer_url_json)
+                    jsplayer_url_json = self._search_regex(
+                        r'"assets":.+?"js":\s*("[^"]+")',
+                        embed_webpage if age_gate else video_webpage, 'JS player URL')
+                    player_url = json.loads(jsplayer_url_json)
                     if player_url is None:
                         player_url_json = self._search_regex(
                             r'ytplayer\.config.*?"url"\s*:\s*("[^"]+")',