[youtube] improve Youtube Music Auto-generated description parsing(closes #20742)

youtube_dl/extractor/youtube.py

@@ -1088,7 +1088,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             },
         },
         {
-            # artist and track fields should return non-null, per issue #20599
+            # Youtube Music Auto-generated description
             'url': 'https://music.youtube.com/watch?v=MgNrAu2pzNs',
             'info_dict': {
                 'id': 'MgNrAu2pzNs',
@@ -1109,11 +1109,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             },
         },
         {
+            # Youtube Music Auto-generated description
             # Retrieve 'artist' field from 'Artist:' in video description
             # when it is present on youtube music video
-            # Some videos have release_date and no release_year -
-            # (release_year should be extracted from release_date)
-            # https://github.com/ytdl-org/youtube-dl/pull/20742#issuecomment-485740932
             'url': 'https://www.youtube.com/watch?v=k0jLE7tTwjY',
             'info_dict': {
                 'id': 'k0jLE7tTwjY',
@@ -1134,6 +1132,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             },
         },
         {
+            # Youtube Music Auto-generated description
             # handle multiple artists on youtube music video
             'url': 'https://www.youtube.com/watch?v=74qn0eJSjpA',
             'info_dict': {
@@ -1155,6 +1154,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             },
         },
         {
+            # Youtube Music Auto-generated description
             # handle youtube music video with release_year and no release_date
             'url': 'https://www.youtube.com/watch?v=-hcAI0g-f5M',
             'info_dict': {
@@ -2161,36 +2161,25 @@ def extract_meta(field):
 
         track = extract_meta('Song')
         artist = extract_meta('Artist')
-        album = None
-        release_date = None
-        release_year = None
 
-        description_info = video_description.split('\n\n')
+        # Youtube Music Auto-generated description
-        # If the description of the video has the youtube music auto-generated format, extract additional info
+        album = release_date = release_year = None
-        if len(description_info) >= 5 and description_info[-1] == 'Auto-generated by YouTube.':
+        if video_description:
-            track_artist = description_info[1].split(' · ')
+            mobj = re.search(r'(?s)Provided to YouTube by [^\n]+\n+(?P<track>[^·]+)·(?P<artist>[^\n]+)\n+(?P<album>[^\n]+)(?:.+?℗\s*(?P<release_year>\d{4})(?!\d))?(?:.+?Released on\s*:\s*(?P<release_date>\d{4}-\d{2}-\d{2}))?(.+?\nArtist\s*:\s*(?P<clean_artist>[^\n]+))?', video_description)
-            if len(track_artist) >= 2:
+            if mobj:
-                if track is None:
+                if not track:
-                    track = track_artist[0]
+                    track = mobj.group('track').strip()
-                if artist is None:
+                if not artist:
-                    artist = re.search(r'Artist: ([^\n]+)', description_info[-2])
+                    artist = mobj.group('clean_artist') or ', '.join(a.strip() for a in mobj.group('artist').split('·'))
-                    if artist:
+                album = mobj.group('album'.strip())
-                        artist = artist.group(1)
+                release_year = mobj.group('release_year')
-                    if artist is None:
+                release_date = mobj.group('release_date')
-                        artist = track_artist[1]
+                if release_date:
-                        # handle multiple artists
+                    release_date = release_date.replace('-', '')
-                        if len(track_artist) > 2:
+                    if not release_year:
-                            for i in range(2, len(track_artist)):
+                        release_year = int(release_date[:4])
-                                artist += ', %s' % track_artist[i]
+                if release_year:
-            release_year = re.search(r'℗ ([0-9]+)', video_description)
+                    release_year = int(release_year)
-            if release_year:
-                release_year = int_or_none(release_year.group(1))
-            album = description_info[2]
-            if description_info[4].startswith('Released on: '):
-                release_date = description_info[4].split(': ')[1].replace('-', '')
-                # extract release_year from release_date if necessary
-                if release_year is None:
-                    release_year = int_or_none(release_date[0:4])
 
         m_episode = re.search(
             r'<div[^>]+id="watch7-headline"[^>]*>\s*<span[^>]*>.*?>(?P<series>[^<]+)</a></b>\s*S(?P<season>\d+)\s*•\s*E(?P<episode>\d+)</span>',