Merge c555e5e478b3d9bc5861bcbd4a9cfe5d61958758 into 2b4fbfce25902d557b86b003cf48f738129efce4

thx seproDev, bashonly: yt-dlp/yt-dlp#12748

test/test_youtube_signature.py

@@ -84,6 +84,21 @@ _SIG_TESTS = [
         '2aq0aqSyOoJXtK73m-uME_jv7-pT15gOFC02RFkGMqWpzEICs69VdbwQ0LDp1v7j8xx92efCJlYFYb1sUkkBSPOlPmXgIARw8JQ0qOAOAA',
         '0QJ8wRAIgXmPlOPSBkkUs1bYFYlJCfe29xxAj7v1pDL0QwbdV96sCIEzpWqMGkFR20CFOg51Tp-7vj_EMu-m37KtXJ2OySqa0q',
     ),
+    (
+        'https://www.youtube.com/s/player/643afba4/tv-player-ias.vflset/tv-player-ias.js',
+        '2aq0aqSyOoJXtK73m-uME_jv7-pT15gOFC02RFkGMqWpzEICs69VdbwQ0LDp1v7j8xx92efCJlYFYb1sUkkBSPOlPmXgIARw8JQ0qOAOAA',
+        'AAOAOq0QJ8wRAIgXmPlOPSBkkUs1bYFYlJCfe29xx8j7vgpDL0QwbdV06sCIEzpWqMGkFR20CFOS21Tp-7vj_EMu-m37KtXJoOy1',
+    ),
+    (
+        'https://www.youtube.com/s/player/363db69b/player_ias.vflset/en_US/base.js',
+        '2aq0aqSyOoJXtK73m-uME_jv7-pT15gOFC02RFkGMqWpzEICs69VdbwQ0LDp1v7j8xx92efCJlYFYb1sUkkBSPOlPmXgIARw8JQ0qOAOAA',
+        '0aqSyOoJXtK73m-uME_jv7-pT15gOFC02RFkGMqWpz2ICs6EVdbwQ0LDp1v7j8xx92efCJlYFYb1sUkkBSPOlPmXgIARw8JQ0qOAOAA',
+    ),
+    (
+        'https://www.youtube.com/s/player/4fcd6e4a/player_ias.vflset/en_US/base.js',
+        '2aq0aqSyOoJXtK73m-uME_jv7-pT15gOFC02RFkGMqWpzEICs69VdbwQ0LDp1v7j8xx92efCJlYFYb1sUkkBSPOlPmXgIARw8JQ0qOAOAA',
+        'wAOAOq0QJ8ARAIgXmPlOPSBkkUs1bYFYlJCfe29xx8q7v1pDL0QwbdV96sCIEzpWqMGkFR20CFOg51Tp-7vj_EMu-m37KtXJoOySqa0',
+    ),
 ]
 
 _NSIG_TESTS = [
@@ -153,7 +168,7 @@ _NSIG_TESTS = [
     ),
     (
         'https://www.youtube.com/s/player/c57c113c/player_ias.vflset/en_US/base.js',
-        '-Txvy6bT5R6LqgnQNx', 'dcklJCnRUHbgSg',
+        'M92UUMHa8PdvPd3wyM', '3hPqLJsiNZx7yA',
     ),
     (
         'https://www.youtube.com/s/player/5a3b6271/player_ias.vflset/en_US/base.js',
@@ -173,7 +188,7 @@ _NSIG_TESTS = [
     ),
     (
         'https://www.youtube.com/s/player/cfa9e7cb/player_ias.vflset/en_US/base.js',
-        'qO0NiMtYQ7TeJnfFG2', 'k9cuJDHNS5O7kQ',
+        'aCi3iElgd2kq0bxVbQ', 'QX1y8jGb2IbZ0w',
     ),
     (
         'https://www.youtube.com/s/player/8c7583ff/player_ias.vflset/en_US/base.js',
@@ -231,10 +246,6 @@ _NSIG_TESTS = [
         'https://www.youtube.com/s/player/f6e09c70/player_ias_tce.vflset/en_US/base.js',
         'W9HJZKktxuYoDTqW', 'jHbbkcaxm54',
     ),
-    (
-        'https://www.youtube.com/s/player/643afba4/player_ias.vflset/en_US/base.js',
-        'W9HJZKktxuYoDTqW', 'larxUlagTRAcSw',
-    ),
     (
         'https://www.youtube.com/s/player/e7567ecf/player_ias_tce.vflset/en_US/base.js',
         'Sy4aDGc0VpYRR9ew_', '5UPOT1VhoZxNLQ',
@@ -259,6 +270,22 @@ _NSIG_TESTS = [
         'https://www.youtube.com/s/player/643afba4/tv-player-ias.vflset/tv-player-ias.js',
         'ir9-V6cdbCiyKxhr', '2PL7ZDYAALMfmA',
     ),
+    (
+        'https://www.youtube.com/s/player/643afba4/player_ias.vflset/en_US/base.js',
+        'W9HJZKktxuYoDTqW', 'larxUlagTRAcSw',
+    ),
+    (
+        'https://www.youtube.com/s/player/363db69b/player_ias.vflset/en_US/base.js',
+        'eWYu5d5YeY_4LyEDc', 'XJQqf-N7Xra3gg',
+    ),
+    (
+        'https://www.youtube.com/s/player/4fcd6e4a/player_ias.vflset/en_US/base.js',
+        'o_L251jm8yhZkWtBW', 'lXoxI3XvToqn6A',
+    ),
+    (
+        'https://www.youtube.com/s/player/4fcd6e4a/tv-player-ias.vflset/tv-player-ias.js',
+        'o_L251jm8yhZkWtBW', 'lXoxI3XvToqn6A',
+    ),
 ]
 
 
@@ -271,6 +298,8 @@ class TestPlayerInfo(unittest.TestCase):
             ('https://www.youtube.com/s/player/64dddad9/player-plasma-ias-phone-en_US.vflset/base.js', '64dddad9'),
             ('https://www.youtube.com/s/player/64dddad9/player-plasma-ias-phone-de_DE.vflset/base.js', '64dddad9'),
             ('https://www.youtube.com/s/player/64dddad9/player-plasma-ias-tablet-en_US.vflset/base.js', '64dddad9'),
+            ('https://www.youtube.com/s/player/e7567ecf/player_ias_tce.vflset/en_US/base.js', 'e7567ecf'),
+            ('https://www.youtube.com/s/player/643afba4/tv-player-ias.vflset/tv-player-ias.js', '643afba4'),
             # obsolete
             ('https://www.youtube.com/yts/jsbin/player_ias-vfle4-e03/en_US/base.js', 'vfle4-e03'),
             ('https://www.youtube.com/yts/jsbin/player_ias-vfl49f_g4/en_US/base.js', 'vfl49f_g4'),
@@ -280,8 +309,9 @@ class TestPlayerInfo(unittest.TestCase):
             ('https://s.ytimg.com/yts/jsbin/html5player-en_US-vflXGBaUN.js', 'vflXGBaUN'),
             ('https://s.ytimg.com/yts/jsbin/html5player-en_US-vflKjOTVq/html5player.js', 'vflKjOTVq'),
         )
+        ie = YoutubeIE(FakeYDL({'cachedir': False}))
         for player_url, expected_player_id in PLAYER_URLS:
-            player_id = YoutubeIE._extract_player_info(player_url)
+            player_id = ie._extract_player_info(player_url)
             self.assertEqual(player_id, expected_player_id)
 
 
@@ -301,8 +331,8 @@ class TestSignature(unittest.TestCase):
 def t_factory(name, sig_func, url_pattern):
     def make_tfunc(url, sig_input, expected_sig):
         m = url_pattern.match(url)
-        assert m, '%r should follow URL format' % url
+        assert m, '{0!r} should follow URL format'.format(url)
-        test_id = m.group('id')
+        test_id = re.sub(r'[/.-]', '_', m.group('id') or m.group('compat_id'))
 
         def test_func(self):
             basename = 'player-{0}-{1}.js'.format(name, test_id)
@@ -335,12 +365,16 @@ def n_sig(jscode, sig_input):
 
 
 make_sig_test = t_factory(
-    'signature', signature, re.compile(r'.*(?:-|/player/)(?P<id>[a-zA-Z0-9_-]+)(?:/.+\.js|(?:/watch_as3|/html5player)?\.[a-z]+)$'))
+    'signature', signature,
+    re.compile(r'''(?x)
+        .+/(?P<h5>html5)?player(?(h5)(?:-en_US)?-|/)(?P<id>[a-zA-Z0-9/._-]+)
+        (?(h5)/(?:watch_as3|html5player))?\.js$
+    '''))
 for test_spec in _SIG_TESTS:
     make_sig_test(*test_spec)
 
 make_nsig_test = t_factory(
-    'nsig', n_sig, re.compile(r'.+/player/(?P<id>[a-zA-Z0-9_-]+)/.+.js$'))
+    'nsig', n_sig, re.compile(r'.+/player/(?P<id>[a-zA-Z0-9_/.-]+)\.js$'))
 for test_spec in _NSIG_TESTS:
     make_nsig_test(*test_spec)
 

youtube_dl/extractor/common.py

@@ -1415,9 +1415,21 @@ class InfoExtractor(object):
                     continue
                 info[count_key] = interaction_count
 
+        def extract_author(e):
+            author = e.get('author')
+            if not author:
+                return
+            info.update({
+                # author can be an instance of 'Organization' or 'Person' types.
+                # both types can have 'name' property(inherited from 'Thing' type). [1]
+                # however some websites are using 'Text' type instead.
+                # 1. https://schema.org/VideoObject
+                'uploader': author.get('name') if isinstance(author, dict) else author if isinstance(author, compat_str) else None,
+            })
+
         def extract_video_object(e):
             assert e['@type'] == 'VideoObject'
-            author = e.get('author')
+            extract_author(e)
             info.update({
                 'url': url_or_none(e.get('contentUrl')),
                 'title': unescapeHTML(e.get('name')),
@@ -1425,11 +1437,6 @@ class InfoExtractor(object):
                 'thumbnail': url_or_none(e.get('thumbnailUrl') or e.get('thumbnailURL')),
                 'duration': parse_duration(e.get('duration')),
                 'timestamp': unified_timestamp(e.get('uploadDate')),
-                # author can be an instance of 'Organization' or 'Person' types.
-                # both types can have 'name' property(inherited from 'Thing' type). [1]
-                # however some websites are using 'Text' type instead.
-                # 1. https://schema.org/VideoObject
-                'uploader': author.get('name') if isinstance(author, dict) else author if isinstance(author, compat_str) else None,
                 'filesize': float_or_none(e.get('contentSize')),
                 'tbr': int_or_none(e.get('bitrate')),
                 'width': int_or_none(e.get('width')),
@@ -1474,6 +1481,8 @@ class InfoExtractor(object):
                         'title': unescapeHTML(e.get('headline')),
                         'description': unescapeHTML(e.get('articleBody')),
                     })
+                elif item_type == 'SocialMediaPosting':
+                    extract_author(e)
                 elif item_type == 'VideoObject':
                     extract_video_object(e)
                     if expected_type is None:

youtube_dl/extractor/facebook.py

@@ -23,6 +23,7 @@ from ..utils import (
     int_or_none,
     js_to_json,
     limit_length,
+    merge_dicts,
     parse_count,
     qualities,
     sanitized_Request,
@@ -82,8 +83,9 @@ class FacebookIE(InfoExtractor):
         'info_dict': {
             'id': '274175099429670',
             'ext': 'mp4',
-            'title': 're:^Asif Nawab Butt posted a video',
+            'title': 'Asif',
             'uploader': 'Asif Nawab Butt',
+            'uploader_id': '100005115464446',
             'upload_date': '20140506',
             'timestamp': 1399398998,
             'thumbnail': r're:^https?://.*',
@@ -137,18 +139,20 @@ class FacebookIE(InfoExtractor):
             'upload_date': '20160223',
             'uploader': 'Barack Obama',
         },
+        'skip': 'Gif on giphy.com gone',
     }, {
         # have 1080P, but only up to 720p in swf params
         # data.video.story.attachments[].media
         'url': 'https://www.facebook.com/cnn/videos/10155529876156509/',
-        'md5': '9571fae53d4165bbbadb17a94651dcdc',
         'info_dict': {
             'id': '10155529876156509',
             'ext': 'mp4',
-            'title': 'She survived the holocaust — and years later, she’s getting her citizenship s...',
+            'title': 'Holocaust survivor becomes US citizen',
+            'description': 'She survived the holocaust — and years later, she’s getting her citizenship so she can vote for Hillary Clinton http://cnn.it/2eERh5f',
             'timestamp': 1477818095,
             'upload_date': '20161030',
             'uploader': 'CNN',
+            'uploader_id': '5550296508',
             'thumbnail': r're:^https?://.*',
             'view_count': int,
         },
@@ -159,14 +163,17 @@ class FacebookIE(InfoExtractor):
         'info_dict': {
             'id': '1417995061575415',
             'ext': 'mp4',
-            'title': 'md5:1db063d6a8c13faa8da727817339c857',
+            'title': 'Ukrainian Scientists Worldwide | Довгоочікуване відео',
+            'description': 'Довгоочікуване відео',
             'timestamp': 1486648217,
             'upload_date': '20170209',
             'uploader': 'Yaroslav Korpan',
+            'uploader_id': '100000948048708',
         },
         'params': {
             'skip_download': True,
         },
+        'skip': 'This video is only available for registered users',
     }, {
         'url': 'https://www.facebook.com/LaGuiaDelVaron/posts/1072691702860471',
         'info_dict': {
@@ -181,16 +188,21 @@ class FacebookIE(InfoExtractor):
         'params': {
             'skip_download': True,
         },
+        # TODO: parse this webpage
+        'skip': 'This video is only available for registered users (and cannot parse webpage well)',
     }, {
         # data.node.comet_sections.content.story.attachments[].style_type_renderer.attachment.media
         'url': 'https://www.facebook.com/groups/1024490957622648/permalink/1396382447100162/',
         'info_dict': {
-            'id': '1396382447100162',
+            'id': '202882990186699',
             'ext': 'mp4',
-            'title': 'md5:19a428bbde91364e3de815383b54a235',
+
+            'title': 'birb (O v O\") | Hello? Yes your uber ride is here',
+            'description': 'md5:f122681cf504d04be12bb2bdc66c81fd',
             'timestamp': 1486035494,
             'upload_date': '20170202',
             'uploader': 'Elisabeth Ahtn',
+            'uploader_id': '100013949973717',
         },
         'params': {
             'skip_download': True,
@@ -231,8 +243,12 @@ class FacebookIE(InfoExtractor):
         'info_dict': {
             'id': '359649331226507',
             'ext': 'mp4',
-            'title': '#ESLOne VoD - Birmingham Finals Day#1 Fnatic vs. @Evil Geniuses',
+            'title': 'Fnatic vs. EG - Group A - Opening Match - ESL One Birmingham Day 1',
+            'description': '#ESLOne VoD - Birmingham Finals Day#1 Fnatic vs. @Evil Geniuses',
+            'timestamp': 1527084179,
+            'upload_date': '20180523',
             'uploader': 'ESL One Dota 2',
+            'uploader_id': '234218833769558',
         },
         'params': {
             'skip_download': True,
@@ -255,6 +271,7 @@ class FacebookIE(InfoExtractor):
             'id': '10157667649866271',
         },
         'playlist_count': 3,
+        'skip': 'Requires logging in',
     }, {
         # data.nodes[].comet_sections.content.story.attachments[].style_type_renderer.attachment.media
         'url': 'https://m.facebook.com/Alliance.Police.Department/posts/4048563708499330',
@@ -279,6 +296,7 @@ class FacebookIE(InfoExtractor):
             'upload_date': '20161122',
             'timestamp': 1479793574,
         },
+        'skip': 'No video',
     }, {
         # data.video.creation_story.attachments[].media
         'url': 'https://www.facebook.com/watch/live/?v=1823658634322275',
@@ -290,6 +308,20 @@ class FacebookIE(InfoExtractor):
         },
         'playlist_count': 1,
         'skip': 'Requires logging in',
+    }, {
+        # data.node.comet_sections.content.story.attachments[].styles.attachment.media
+        # gone : 'https://www.facebook.com/groups/352666925484718/permalink/1112505706167499/',
+        'url': 'https://www.facebook.com/groups/471738070722101/permalink/471799617382613/',
+        'info_dict': {
+            'id': '101173872166725',
+            'ext': 'mp4',
+            'title': 'md5:0ac174e60ed9d9a5eec588343f6a889c',
+            'description': 'md5:80efdff1371fa5d92758df2fb1b97bfd',
+            'upload_date': '20210517',
+            'timestamp': 1621247381,
+            'uploader': 'md5:34165f897c692f92b13c2cc1eddd8245',
+            'uploader_id': '100957655521680',
+        },
     }]
     _SUPPORTED_PAGLETS_REGEX = r'(?:pagelet_group_mall|permalink_video_pagelet|hyperfeed_story_id_[0-9a-f]+)'
     _api_config = {
@@ -378,8 +410,80 @@ class FacebookIE(InfoExtractor):
         self._login()
 
     def _extract_from_url(self, url, video_id):
-        webpage = self._download_webpage(
+        webpage, urlh = self._download_webpage_handle(
             url.replace('://m.facebook.com/', '://www.facebook.com/'), video_id)
+        login_or_id_check = re.search(r'(?P<login>\blogin\b)|/permalink/|[/=?]' + video_id, urlh.geturl())
+        if login_or_id_check and login_or_id_check.group('login'):
+            self.raise_login_required()
+        elif not login_or_id_check:
+            raise ExtractorError(
+                'Facebook redirected to a generic page (%s): the video may be unavailable or you may need to log in.' % urlh.geturl(),
+                expected=True)
+
+        def extract_metadata(webpage):
+            video_title = self._html_search_regex(
+                r'<h2\s+[^>]*class="uiHeaderTitle"[^>]*>([^<]*)</h2>', webpage,
+                'title', default=None)
+            if not video_title:
+                video_title = self._html_search_regex(
+                    r'(?s)<span class="fbPhotosPhotoCaption".*?id="fbPhotoPageCaption"><span class="hasCaption">(.*?)</span>',
+                    webpage, 'alternative title', default=None)
+
+            # _html_search_meta() finds '' ahead of later non-empty results
+            def html_search_meta_non_empty(patterns, html, name, **kwargs):
+                for p in patterns:
+                    result = self._html_search_meta(p, html, name, **kwargs)
+                    if result:
+                        return result
+
+            if not video_title:
+                video_title = html_search_meta_non_empty(
+                    ['og:title', 'twitter:title', 'description'],
+                    webpage, 'title', default=None)
+            if video_title:
+                video_title = limit_length(video_title, 80)
+            else:
+                video_title = 'Facebook video #%s' % video_id
+            description = html_search_meta_non_empty(
+                ['description', 'og:description', 'twitter:description'],
+                webpage, 'description', default=None)
+            uploader = clean_html(get_element_by_id(
+                'fbPhotoPageAuthorName', webpage)) or self._search_regex(
+                r'ownerName\s*:\s*"([^"]+)"', webpage, 'uploader',
+                default=None) or self._og_search_title(webpage, default=None)
+            timestamp = int_or_none(self._search_regex(
+                r'<abbr[^>]+data-utime=["\'](\d+)', webpage,
+                'timestamp', default=None))
+            thumbnail = html_search_meta_non_empty(
+                ['og:image', 'twitter:image'], webpage, 'thumbnail', default=None)
+            # some webpages contain unretrievable thumbnail urls
+            # like https://lookaside.fbsbx.com/lookaside/crawler/media/?media_id=10155168902769113&get_thumbnail=1
+            # in https://www.facebook.com/yaroslav.korpan/videos/1417995061575415/
+            if thumbnail and not re.search(r'\.(?:jpg|png)', thumbnail):
+                thumbnail = None
+            view_count = parse_count(self._search_regex(
+                [r'\bviewCount\s*:\s*["\']([\d,.]+)', r'"video_view_count"\s*:\s*([\d,.]+)'],
+                webpage, 'view count', default=None))
+            info_dict = {
+                'title': video_title,
+                'description': description,
+                'uploader': uploader,
+                'timestamp': timestamp,
+                'thumbnail': thumbnail,
+                'view_count': view_count,
+            }
+            info_json_ld = self._search_json_ld(webpage, video_id, default={})
+            if not info_json_ld:
+                # expecting title, description, uploader, comment_count and like_count
+                info_json_ld = self._search_json_ld(
+                    webpage, video_id, expected_type='SocialMediaPosting', default={})
+                # if we already have a non-generic title, honor that
+                if info_dict['title'] != 'Facebook video #%s' % video_id:
+                    info_json_ld['title'] = info_dict['title']
+            if info_json_ld.get('title'):
+                info_json_ld['title'] = limit_length(
+                    re.sub(r'\s*\|\s*Facebook$', '', info_json_ld['title']), 80)
+            return merge_dicts(info_json_ld, info_dict)
 
         video_data = None
 
@@ -477,7 +581,7 @@ class FacebookIE(InfoExtractor):
                     entries.append(info)
 
                 def parse_attachment(attachment, key='media'):
-                    media = attachment.get(key) or {}
+                    media = try_get(attachment, lambda x: x[key], dict) or {}
                     if media.get('__typename') == 'Video':
                         return parse_graphql_video(media)
 
@@ -492,7 +596,10 @@ class FacebookIE(InfoExtractor):
                         lambda x: x['attachments']
                     ], list) or []
                     for attachment in attachments:
-                        attachment = try_get(attachment, lambda x: x['style_type_renderer']['attachment'], dict)
+                        attachment = try_get(attachment, [
+                            lambda x: x['style_type_renderer']['attachment'],
+                            lambda x: x['styles']['attachment']
+                        ], dict) or {}
                         ns = try_get(attachment, lambda x: x['all_subattachments']['nodes'], list) or []
                         for n in ns:
                             parse_attachment(n)
@@ -513,8 +620,16 @@ class FacebookIE(InfoExtractor):
                     if not entries:
                         parse_graphql_video(video)
 
+                if len(entries) > 1:
                     return self.playlist_result(entries, video_id)
 
+                video_info = entries[0]
+                webpage_info = extract_metadata(webpage)
+                # honor precise duration in video info
+                if video_info.get('duration'):
+                    webpage_info['duration'] = video_info['duration']
+                return merge_dicts(webpage_info, video_info)
+
         if not video_data:
             m_msg = re.search(r'class="[^"]*uiInterstitialContent[^"]*"><div>(.*?)</div>', webpage)
             if m_msg is not None:
@@ -633,43 +748,12 @@ class FacebookIE(InfoExtractor):
 
         process_formats(formats)
 
-        video_title = self._html_search_regex(
-            r'<h2\s+[^>]*class="uiHeaderTitle"[^>]*>([^<]*)</h2>', webpage,
-            'title', default=None)
-        if not video_title:
-            video_title = self._html_search_regex(
-                r'(?s)<span class="fbPhotosPhotoCaption".*?id="fbPhotoPageCaption"><span class="hasCaption">(.*?)</span>',
-                webpage, 'alternative title', default=None)
-        if not video_title:
-            video_title = self._html_search_meta(
-                'description', webpage, 'title', default=None)
-        if video_title:
-            video_title = limit_length(video_title, 80)
-        else:
-            video_title = 'Facebook video #%s' % video_id
-        uploader = clean_html(get_element_by_id(
-            'fbPhotoPageAuthorName', webpage)) or self._search_regex(
-            r'ownerName\s*:\s*"([^"]+)"', webpage, 'uploader',
-            default=None) or self._og_search_title(webpage, fatal=False)
-        timestamp = int_or_none(self._search_regex(
-            r'<abbr[^>]+data-utime=["\'](\d+)', webpage,
-            'timestamp', default=None))
-        thumbnail = self._html_search_meta(['og:image', 'twitter:image'], webpage)
-
-        view_count = parse_count(self._search_regex(
-            r'\bviewCount\s*:\s*["\']([\d,.]+)', webpage, 'view count',
-            default=None))
-
         info_dict = {
             'id': video_id,
-            'title': video_title,
             'formats': formats,
-            'uploader': uploader,
-            'timestamp': timestamp,
-            'thumbnail': thumbnail,
-            'view_count': view_count,
             'subtitles': subtitles,
         }
+        info_dict.update(extract_metadata(webpage))
 
         return info_dict
 

youtube_dl/extractor/youtube.py

@@ -692,9 +692,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         'invidious': '|'.join(_INVIDIOUS_SITES),
     }
     _PLAYER_INFO_RE = (
-        r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})//(?:tv-)?player',
+        r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})/(?:tv-)?player',
-        r'/(?P<id>[a-zA-Z0-9_-]{8,})/player(?:_ias\.vflset(?:/[a-zA-Z]{2,3}_[a-zA-Z]{2,3})?|-plasma-ias-(?:phone|tablet)-[a-z]{2}_[A-Z]{2}\.vflset)/base\.js$',
+        r'/(?P<id>[a-zA-Z0-9_-]{8,})/player(?:_ias(?:_tce)?\.vflset(?:/[a-zA-Z]{2,3}_[a-zA-Z]{2,3})?|-plasma-ias-(?:phone|tablet)-[a-z]{2}_[A-Z]{2}\.vflset)/base\.js$',
-        r'\b(?P<id>vfl[a-zA-Z0-9_-]+)\b.*?\.js$',
+        r'\b(?P<id>vfl[a-zA-Z0-9_-]{6,})\b.*?\.js$',
     )
     _SUBTITLE_FORMATS = ('json3', 'srv1', 'srv2', 'srv3', 'ttml', 'vtt')
 
@@ -1626,15 +1626,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         """ Return a string representation of a signature """
         return '.'.join(compat_str(len(part)) for part in example_sig.split('.'))
 
-    @classmethod
+    def _extract_player_info(self, player_url):
-    def _extract_player_info(cls, player_url):
+        try:
-        for player_re in cls._PLAYER_INFO_RE:
+            return self._search_regex(
-            id_m = re.search(player_re, player_url)
+                self._PLAYER_INFO_RE, player_url, 'player info', group='id')
-            if id_m:
+        except ExtractorError as e:
-                break
+            raise ExtractorError(
-        else:
+                'Cannot identify player %r' % (player_url,), cause=e)
-            raise ExtractorError('Cannot identify player %r' % player_url)
-        return id_m.group('id')
 
     def _load_player(self, video_id, player_url, fatal=True, player_id=None):
         if not player_id:
@@ -1711,6 +1709,23 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                 '    return %s\n') % (signature_id_tuple, expr_code)
         self.to_screen('Extracted signature function:\n' + code)
 
+    def _extract_sig_fn(self, jsi, funcname):
+        var_ay = self._search_regex(
+            r'''(?x)
+                (?:\*/|\{|\n|^)\s*(?:'[^']+'\s*;\s*)
+                    (var\s*[\w$]+\s*=\s*(?:
+                        ('|")(?:\\\2|(?!\2).)+\2\s*\.\s*split\(\s*('|")\W+\3\s*\)|
+                        \[\s*(?:('|")(?:\\\4|(?!\4).)*\4\s*(?:(?=\])|,\s*))+\]
+                    ))(?=\s*[,;])
+            ''', jsi.code, 'useful values', default='')
+
+        sig_fn = jsi.extract_function_code(funcname)
+
+        if var_ay:
+            sig_fn = (sig_fn[0], ';\n'.join((var_ay, sig_fn[1])))
+
+        return sig_fn
+
     def _parse_sig_js(self, jscode):
         # Examples where `sig` is funcname:
         # sig=function(a){a=a.split(""); ... ;return a.join("")};
@@ -1736,8 +1751,12 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             jscode, 'Initial JS player signature function name', group='sig')
 
         jsi = JSInterpreter(jscode)
-        initial_function = jsi.extract_function(funcname)
+
-        return lambda s: initial_function([s])
+        initial_function = self._extract_sig_fn(jsi, funcname)
+
+        func = jsi.extract_function_from_code(*initial_function)
+
+        return lambda s: func([s])
 
     def _cached(self, func, *cache_id):
         def inner(*args, **kwargs):
@@ -1856,15 +1875,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
 
     def _extract_n_function_code_jsi(self, video_id, jsi, player_id=None):
 
-        var_ay = self._search_regex(
-            r'(?:[;\s]|^)\s*(var\s*[\w$]+\s*=\s*"(?:\\"|[^"])+"\s*\.\s*split\("\W+"\))(?=\s*[,;])',
-            jsi.code, 'useful values', default='')
-
         func_name = self._extract_n_function_name(jsi.code)
 
-        func_code = jsi.extract_function_code(func_name)
+        func_code = self._extract_sig_fn(jsi, func_name)
-        if var_ay:
-            func_code = (func_code[0], ';\n'.join((var_ay, func_code[1])))
 
         if player_id:
             self.cache.store('youtube-nsig', player_id, func_code)
@@ -2136,7 +2149,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                     video_details = merge_dicts(*traverse_obj(
                         (player_response, api_player_response),
                         (Ellipsis, 'videoDetails', T(dict))))
-                    player_response.update(api_player_response or {})
+                    player_response.update(filter_dict(
+                        api_player_response or {}, cndn=lambda k, _: k != 'captions'))
                     player_response['videoDetails'] = video_details
 
         def is_agegated(playability):
@@ -2566,8 +2580,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         }
 
         pctr = traverse_obj(
-            player_response,
+            (player_response, api_player_response),
-            ('captions', 'playerCaptionsTracklistRenderer', T(dict)))
+            (Ellipsis, 'captions', 'playerCaptionsTracklistRenderer', T(dict)))
         if pctr:
             def process_language(container, base_url, lang_code, query):
                 lang_subs = []
@@ -2584,20 +2598,21 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             def process_subtitles():
                 subtitles = {}
                 for caption_track in traverse_obj(pctr, (
-                        'captionTracks', lambda _, v: v.get('baseUrl'))):
+                        Ellipsis, 'captionTracks', lambda _, v: (
+                            v.get('baseUrl') and v.get('languageCode')))):
                     base_url = self._yt_urljoin(caption_track['baseUrl'])
                     if not base_url:
                         continue
+                    lang_code = caption_track['languageCode']
                     if caption_track.get('kind') != 'asr':
-                        lang_code = caption_track.get('languageCode')
-                        if not lang_code:
-                            continue
                         process_language(
                             subtitles, base_url, lang_code, {})
                         continue
                     automatic_captions = {}
+                    process_language(
+                        automatic_captions, base_url, lang_code, {})
                     for translation_language in traverse_obj(pctr, (
-                            'translationLanguages', lambda _, v: v.get('languageCode'))):
+                            Ellipsis, 'translationLanguages', lambda _, v: v.get('languageCode'))):
                         translation_language_code = translation_language['languageCode']
                         process_language(
                             automatic_captions, base_url, translation_language_code,

youtube_dl/jsinterp.py

@@ -678,7 +678,7 @@ class JSInterpreter(object):
             return len(obj)
         try:
             return obj[int(idx)] if isinstance(obj, list) else obj[compat_str(idx)]
-        except (TypeError, KeyError, IndexError) as e:
+        except (TypeError, KeyError, IndexError, ValueError) as e:
             # allow_undefined is None gives correct behaviour
             if allow_undefined or (
                     allow_undefined is None and not isinstance(e, TypeError)):
@@ -1038,6 +1038,10 @@ class JSInterpreter(object):
                     left_val = self._index(left_val, idx)
             if isinstance(idx, float):
                 idx = int(idx)
+            if isinstance(left_val, list) and len(left_val) <= int_or_none(idx, default=-1):
+                # JS Array is a sparsely assignable list
+                # TODO: handle extreme sparsity without memory bloat, eg using auxiliary dict
+                left_val.extend((idx - len(left_val) + 1) * [JS_Undefined])
             left_val[idx] = self._operator(
                 m.group('op'), self._index(left_val, idx) if m.group('op') else None,
                 m.group('expr'), expr, local_vars, allow_recursion)
@@ -1204,7 +1208,8 @@ class JSInterpreter(object):
                 elif member == 'join':
                     assertion(isinstance(obj, list), 'must be applied on a list')
                     assertion(len(argvals) <= 1, 'takes at most one argument')
-                    return (',' if len(argvals) == 0 else argvals[0]).join(
+                    return (',' if len(argvals) == 0 or argvals[0] in (None, JS_Undefined)
+                            else argvals[0]).join(
                                 ('' if x in (None, JS_Undefined) else _js_toString(x))
                                 for x in obj)
                 elif member == 'reverse':
@@ -1364,19 +1369,21 @@ class JSInterpreter(object):
         code, _ = self._separate_at_paren(func_m.group('code'))  # refine the match
         return self.build_arglist(func_m.group('args')), code
 
-    def extract_function(self, funcname):
+    def extract_function(self, funcname, *global_stack):
         return function_with_repr(
-            self.extract_function_from_code(*self.extract_function_code(funcname)),
+            self.extract_function_from_code(*itertools.chain(
+                self.extract_function_code(funcname), global_stack)),
             'F<%s>' % (funcname,))
 
     def extract_function_from_code(self, argnames, code, *global_stack):
         local_vars = {}
 
+        start = None
         while True:
-            mobj = re.search(r'function\((?P<args>[^)]*)\)\s*{', code)
+            mobj = re.search(r'function\((?P<args>[^)]*)\)\s*{', code[start:])
             if mobj is None:
                 break
-            start, body_start = mobj.span()
+            start, body_start = ((start or 0) + x for x in mobj.span())
             body, remaining = self._separate_at_paren(code[body_start - 1:])
             name = self._named_object(local_vars, self.extract_function_from_code(
                 [x.strip() for x in mobj.group('args').split(',')],