[youtube] Fix extraction (closes #20758, closes #20759, closes #20761, closes #20762...

[youtube-dl] / youtube_dl / extractor / youtube.py

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 0bc6cc7a73fc436d090cfded42617c6466679f9f..1bc2c27adacb40b2cd5e8ff2fb2e56b92eece38b 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -351,7 +351,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                             (?:www\.)?hooktube\.com/|
                             (?:www\.)?yourepeat\.com/|
                             tube\.majestyc\.net/|
-                            (?:www\.)?invidio\.us/|
+                            (?:(?:www|dev)\.)?invidio\.us/|
+                            (?:www\.)?invidiou\.sh/|
                             (?:www\.)?invidious\.snopyta\.org/|
                             (?:www\.)?invidious\.kabi\.tk/|
                             (?:www\.)?vid\.wxzm\.sx/|
@@ -483,7 +484,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         # RTMP (unnamed)
         '_rtmp': {'protocol': 'rtmp'},
     }
-    _SUBTITLE_FORMATS = ('ttml', 'vtt')
+    _SUBTITLE_FORMATS = ('srv1', 'srv2', 'srv3', 'ttml', 'vtt')
 
     _GEO_BYPASS = False
 
@@ -1651,7 +1652,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                         view_count = extract_view_count(get_video_info)
                     if not video_info:
                         video_info = get_video_info
-                    if 'token' in get_video_info:
+                    get_token = get_video_info.get('token') or get_video_info.get('account_playback_token')
+                    if get_token:
                         # Different get_video_info requests may report different results, e.g.
                         # some may report video unavailability, but some may serve it without
                         # any complaint (see https://github.com/ytdl-org/youtube-dl/issues/7362,
@@ -1661,7 +1663,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                         # due to YouTube measures against IP ranges of hosting providers.
                         # Working around by preferring the first succeeded video_info containing
                         # the token if no such video_info yet was found.
-                        if 'token' not in video_info:
+                        token = video_info.get('token') or video_info.get('account_playback_token')
+                        if not token:
                             video_info = get_video_info
                         break
 
@@ -1670,7 +1673,15 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                 r'(?s)<h1[^>]+id="unavailable-message"[^>]*>(.+?)</h1>',
                 video_webpage, 'unavailable message', default=None)
 
-        if 'token' not in video_info:
+        if not video_info:
+            unavailable_message = extract_unavailable_message()
+            if not unavailable_message:
+                unavailable_message = 'Unable to extract video data'
+            raise ExtractorError(
+                'YouTube said: %s' % unavailable_message, expected=True, video_id=video_id)
+
+        token = video_info.get('token') or video_info.get('account_playback_token')
+        if not token:
             if 'reason' in video_info:
                 if 'The uploader has not made this video available in your country.' in video_info['reason']:
                     regions_allowed = self._html_search_meta(