[youtube] Relax html5 player regexes (closes #18465, closes #18466)
[youtube-dl] / youtube_dl / extractor / youtube.py
index c582ab2ff7cfc1445b40a7c989de945b39a14c52..954853e009bfc2ec312a3d3ea95b5bec3ecdf112 100644 (file)
@@ -48,6 +48,7 @@ from ..utils import (
     unified_strdate,
     unsmuggle_url,
     uppercase_escape,
+    url_or_none,
     urlencode_postdata,
 )
 
@@ -1104,7 +1105,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
 
     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(?:-new)?|(?:/[a-z]{2}_[A-Z]{2})?/base)?\.(?P<ext>[a-z]+)$',
+            r'.*?-(?P<id>[a-zA-Z0-9_-]+)(?:/watch_as3|/html5player(?:-new)?|(?:/[a-z]{2,3}_[A-Z]{2})?/base)?\.(?P<ext>[a-z]+)$',
             player_url)
         if not id_m:
             raise ExtractorError('Cannot identify player %r' % player_url)
@@ -1386,8 +1387,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             self._downloader.report_warning(err_msg)
             return {}
 
-    def _mark_watched(self, video_id, video_info):
-        playback_url = video_info.get('videostats_playback_base_url', [None])[0]
+    def _mark_watched(self, video_id, video_info, player_response):
+        playback_url = url_or_none(try_get(
+            player_response,
+            lambda x: x['playbackTracking']['videostatsPlaybackUrl']['baseUrl']) or try_get(
+            video_info, lambda x: x['videostats_playback_base_url'][0]))
         if not playback_url:
             return
         parsed_playback_url = compat_urlparse.urlparse(playback_url)
@@ -1830,7 +1834,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                             else:
                                 player_version = self._search_regex(
                                     [r'html5player-([^/]+?)(?:/html5player(?:-new)?)?\.js',
-                                     r'(?:www|player)-([^/]+)(?:/[a-z]{2}_[A-Z]{2})?/base\.js'],
+                                     r'(?:www|player(?:_ias)?)-([^/]+)(?:/[a-z]{2,3}_[A-Z]{2})?/base\.js'],
                                     player_url,
                                     'html5 player', fatal=False)
                                 player_desc = 'html5 player %s' % player_version
@@ -2122,7 +2126,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
 
         self._sort_formats(formats)
 
-        self.mark_watched(video_id, video_info)
+        self.mark_watched(video_id, video_info, player_response)
 
         return {
             'id': video_id,