[youtube] improve title and description extraction(closes #21934)

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

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index aa316ba88abf9eb91c093e1d7f85c25c13b39542..9a182fcf67138bf347c677fe69ca73bcc4fc8290 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -1820,16 +1820,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         video_details = try_get(
             player_response, lambda x: x['videoDetails'], dict) or {}
 
-        # title
-        if 'title' in video_info:
-            video_title = video_info['title'][0]
-        elif 'title' in player_response:
-            video_title = video_details['title']
-        else:
+        video_title = video_info.get('title', [None])[0] or video_details.get('title')
+        if not video_title:
             self._downloader.report_warning('Unable to extract video title')
             video_title = '_'
 
-        # description
         description_original = video_description = get_element_by_id("eow-description", video_webpage)
         if video_description:
 
@@ -1854,11 +1849,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             ''', replace_url, video_description)
             video_description = clean_html(video_description)
         else:
-            fd_mobj = re.search(r'<meta name="description" content="([^"]+)"', video_webpage)
-            if fd_mobj:
-                video_description = unescapeHTML(fd_mobj.group(1))
-            else:
-                video_description = ''
+            video_description = self._html_search_meta('description', video_webpage) or video_details.get('shortDescription')
 
         if not smuggled_data.get('force_singlefeed', False):
             if not self._downloader.params.get('noplaylist'):
@@ -2455,6 +2446,8 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
         'info_dict': {
             'title': '29C3: Not my department',
             'id': 'PLwP_SiAcdui0KVebT0mU9Apz359a4ubsC',
+            'uploader': 'Christiaan008',
+            'uploader_id': 'ChRiStIaAn008',
         },
         'playlist_count': 95,
     }, {
@@ -2463,6 +2456,8 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
         'info_dict': {
             'title': '[OLD]Team Fortress 2 (Class-based LP)',
             'id': 'PLBB231211A4F62143',
+            'uploader': 'Wickydoo',
+            'uploader_id': 'Wickydoo',
         },
         'playlist_mincount': 26,
     }, {
@@ -2471,6 +2466,8 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
         'info_dict': {
             'title': 'Uploads from Cauchemar',
             'id': 'UUBABnxM4Ar9ten8Mdjj1j0Q',
+            'uploader': 'Cauchemar',
+            'uploader_id': 'Cauchemar89',
         },
         'playlist_mincount': 799,
     }, {
@@ -2488,13 +2485,17 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
         'info_dict': {
             'title': 'JODA15',
             'id': 'PL6IaIsEjSbf96XFRuNccS_RuEXwNdsoEu',
+            'uploader': 'milan',
+            'uploader_id': 'UCEI1-PVPcYXjB73Hfelbmaw',
         }
     }, {
         'url': 'http://www.youtube.com/embed/_xDOZElKyNU?list=PLsyOSbh5bs16vubvKePAQ1x3PhKavfBIl',
         'playlist_mincount': 485,
         'info_dict': {
-            'title': '2017 華語最新單曲 (2/24更新)',
+            'title': '2018 Chinese New Singles (11/6 updated)',
             'id': 'PLsyOSbh5bs16vubvKePAQ1x3PhKavfBIl',
+            'uploader': 'LBK',
+            'uploader_id': 'sdragonfang',
         }
     }, {
         'note': 'Embedded SWF player',
@@ -2503,13 +2504,16 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
         'info_dict': {
             'title': 'JODA7',
             'id': 'YN5VISEtHet5D4NEvfTd0zcgFk84NqFZ',
-        }
+        },
+        'skip': 'This playlist does not exist',
     }, {
         'note': 'Buggy playlist: the webpage has a "Load more" button but it doesn\'t have more videos',
         'url': 'https://www.youtube.com/playlist?list=UUXw-G3eDE9trcvY2sBMM_aA',
         'info_dict': {
             'title': 'Uploads from Interstellar Movie',
             'id': 'UUXw-G3eDE9trcvY2sBMM_aA',
+            'uploader': 'Interstellar Movie',
+            'uploader_id': 'InterstellarMovie1',
         },
         'playlist_mincount': 21,
     }, {
@@ -2534,6 +2538,7 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
         'params': {
             'skip_download': True,
         },
+        'skip': 'This video is not available.',
         'add_ie': [YoutubeIE.ie_key()],
     }, {
         'url': 'https://youtu.be/yeWKywCrFtk?list=PL2qgrgXsNUG5ig9cat4ohreBjYLAPC0J5',
@@ -2545,7 +2550,6 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
             'uploader_id': 'backuspagemuseum',
             'uploader_url': r're:https?://(?:www\.)?youtube\.com/user/backuspagemuseum',
             'upload_date': '20161008',
-            'license': 'Standard YouTube License',
             'description': 'md5:800c0c78d5eb128500bffd4f0b4f2e8a',
             'categories': ['Nonprofits & Activism'],
             'tags': list,
@@ -2732,6 +2736,8 @@ class YoutubeChannelIE(YoutubePlaylistBaseInfoExtractor):
         'info_dict': {
             'id': 'UUKfVa3S1e4PHvxWcwyMMg8w',
             'title': 'Uploads from lex will',
+            'uploader': 'lex will',
+            'uploader_id': 'UCKfVa3S1e4PHvxWcwyMMg8w',
         }
     }, {
         'note': 'Age restricted channel',
@@ -2741,6 +2747,8 @@ class YoutubeChannelIE(YoutubePlaylistBaseInfoExtractor):
         'info_dict': {
             'id': 'UUs0ifCMCm1icqRbqhUINa0w',
             'title': 'Uploads from Deus Ex',
+            'uploader': 'Deus Ex',
+            'uploader_id': 'DeusExOfficial',
         },
     }, {
         'url': 'https://invidio.us/channel/UC23qupoDRn9YOAVzeoxjOQA',
@@ -2825,6 +2833,8 @@ class YoutubeUserIE(YoutubeChannelIE):
         'info_dict': {
             'id': 'UUfX55Sx5hEFjoC3cNs6mCUQ',
             'title': 'Uploads from The Linux Foundation',
+            'uploader': 'The Linux Foundation',
+            'uploader_id': 'TheLinuxFoundation',
         }
     }, {
         # Only available via https://www.youtube.com/c/12minuteathlete/videos
@@ -2834,6 +2844,8 @@ class YoutubeUserIE(YoutubeChannelIE):
         'info_dict': {
             'id': 'UUVjM-zV6_opMDx7WYxnjZiQ',
             'title': 'Uploads from 12 Minute Athlete',
+            'uploader': '12 Minute Athlete',
+            'uploader_id': 'the12minuteathlete',
         }
     }, {
         'url': 'ytuser:phihag',
@@ -2927,7 +2939,7 @@ class YoutubePlaylistsIE(YoutubePlaylistsBaseInfoExtractor):
         'playlist_mincount': 4,
         'info_dict': {
             'id': 'ThirstForScience',
-            'title': 'Thirst for Science',
+            'title': 'ThirstForScience',
         },
     }, {
         # with "Load more" button
@@ -2944,6 +2956,7 @@ class YoutubePlaylistsIE(YoutubePlaylistsBaseInfoExtractor):
             'id': 'UCiU1dHvZObB2iP6xkJ__Icw',
             'title': 'Chem Player',
         },
+        'skip': 'Blocked',
     }]