Merge 1231ce7d561f7a269ccc7d4b404b47184b63f826 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
-        test_id = m.group('id')
+        assert m, '{0!r} should follow URL format'.format(url)
+        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/go.py

@@ -20,10 +20,12 @@ class GoIE(AdobePassIE):
         'abc': {
             'brand': '001',
             'requestor_id': 'ABC',
+            'go': True,
         },
         'freeform': {
             'brand': '002',
             'requestor_id': 'ABCFamily',
+            'go': True,
         },
         'watchdisneychannel': {
             'brand': '004',
@@ -40,6 +42,7 @@ class GoIE(AdobePassIE):
         'disneynow': {
             'brand': '011',
             'resource_id': 'Disney',
+            'go': True,
         },
         'fxnow.fxnetworks': {
             'brand': '025',
@@ -48,15 +51,19 @@ class GoIE(AdobePassIE):
     }
     _VALID_URL = r'''(?x)
                     https?://
-                        (?:
-                            (?:(?P<sub_domain>%s)\.)?go|
-                            (?P<sub_domain_2>abc|freeform|disneynow|fxnow\.fxnetworks)
+                        (?P<sub_domain>
+                            (?:(?:%(sites)s)\.)?go|fxnow\.fxnetworks|
+                            (?:www\.)?(?:%(sites)s)
                         )\.com/
                         (?:
                             (?:[^/]+/)*(?P<id>[Vv][Dd][Kk][Aa]\w+)|
                             (?:[^/]+/)*(?P<display_id>[^/?\#]+)
                         )
-                    ''' % '|'.join(list(_SITE_INFO.keys()))
+                    ''' % {
+        'sites':
+            # keep _SITE_INFO in scope
+            (lambda s: '|'.join(re.escape(k) for k in s.keys() if s[k].get('go')))(_SITE_INFO),
+    }
     _TESTS = [{
         'url': 'http://abc.go.com/shows/designated-survivor/video/most-recent/VDKA3807643',
         'info_dict': {
@@ -70,14 +77,9 @@ class GoIE(AdobePassIE):
             'skip_download': True,
         },
         'skip': 'This content is no longer available.',
-    }, {
-        'url': 'http://watchdisneyxd.go.com/doraemon',
-        'info_dict': {
-            'title': 'Doraemon',
-            'id': 'SH55574025',
     },
-        'playlist_mincount': 51,
-    }, {
+        # Disney XD app withdrawn 2018: 'http://watchdisneyxd.go.com/doraemon',
+        {
         'url': 'http://freeform.go.com/shows/shadowhunters/episodes/season-2/1-this-guilty-blood',
         'info_dict': {
             'id': 'VDKA3609139',
@@ -85,23 +87,37 @@ class GoIE(AdobePassIE):
             'title': 'This Guilty Blood',
             'description': 'md5:f18e79ad1c613798d95fdabfe96cd292',
             'age_limit': 14,
+            'duration': 2544,
+            'episode': 'Episode 1',
+            'thumbnail': r're:https?://.*?\.jpg',
+            'season_number': 2,
+            'episode_number': 1,
+            'season': 'Season 2',
+            'series': 'Shadowhunters',
         },
         'params': {
-            'geo_bypass_ip_block': '3.244.239.0/24',
+            'geo_bypass_ip_block': '6.48.97.0/24',
             # m3u8 download
             'skip_download': True,
         },
     }, {
-        'url': 'https://abc.com/shows/the-rookie/episode-guide/season-02/03-the-bet',
+        'url': 'https://abc.com/shows/the-rookie/episode-guide/season-04/13-fight-or-flight',
         'info_dict': {
-            'id': 'VDKA13435179',
+            'id': 'VDKA26139487',
             'ext': 'mp4',
-            'title': 'The Bet',
-            'description': 'md5:c66de8ba2e92c6c5c113c3ade84ab404',
+            'title': 'Fight or Flight',
+            'episode': 'Episode 13',
+            'thumbnail': r're:https?://.*?\.jpg',
+            'series': 'The Rookie',
+            'description': 'md5:8caf91de9806a1c5f79823059fe80267',
+            'duration': 2575,
+            'episode_number': 13,
+            'season_number': 4,
+            'season': 'Season 4',
             'age_limit': 14,
         },
         'params': {
-            'geo_bypass_ip_block': '3.244.239.0/24',
+            'geo_bypass_ip_block': '6.48.97.0/24',
             # m3u8 download
             'skip_download': True,
         },
@@ -112,9 +128,13 @@ class GoIE(AdobePassIE):
             'ext': 'mp4',
             'title': 'First Look: Better Things - Season 2',
             'description': 'md5:fa73584a95761c605d9d54904e35b407',
+            'duration': 161,
+            'series': 'Better Things',
+            'age_limit': 14,
+            'thumbnail': r're:https?://.*?\.jpg',
         },
         'params': {
-            'geo_bypass_ip_block': '3.244.239.0/24',
+            'geo_bypass_ip_block': '6.48.97.0/24',
             # m3u8 download
             'skip_download': True,
         },
@@ -125,11 +145,20 @@ class GoIE(AdobePassIE):
             'ext': 'mp4',
             'title': 'Pilot',
             'description': 'md5:74306df917cfc199d76d061d66bebdb4',
+            'season': 'Season 1',
+            'episode': 'Episode 1',
+            'series': 'Modern Family',
+            'season_number': 1,
+            'duration': 1301,
+            'episode_number': 1,
+            'age_limit': 0,
+            'thumbnail': r're:https?://.*?\.jpg',
         },
         'params': {
             # m3u8 download
             'skip_download': True,
         },
+        'skip': 'The content you’re looking for is not available',
     }, {
         'url': 'http://abc.go.com/shows/the-catch/episode-guide/season-01/10-the-wedding',
         'only_matching': True,
@@ -147,28 +176,66 @@ class GoIE(AdobePassIE):
     }, {
         'url': 'https://disneynow.com/shows/minnies-bow-toons/video/happy-campers/vdka4872013',
         'only_matching': True,
+    }, {
+        'url': 'https://www.freeform.com/shows/cruel-summer/episode-guide/season-01/01-happy-birthday-jeanette-turner',
+        'only_matching': True,
+    }, {
+        # will expire and need to be replaced by a newer episode
+        'url': 'https://disneynow.com/shows/ducktales-disney-channel/season-03/episode-22-the-last-adventure/vdka22464554',
+        'info_dict': {
+            'id': 'VDKA22464554',
+            'ext': 'mp4',
+            'title': 'The Last Adventure!',
+            'description': 'The Duck Family uncovers earth-shattering secrets in a final standoff with FOWL, with the future of adventure hanging in the balance.',
+            'season_number': 3,
+            'episode': 'Episode 22',
+            'season': 'Season 3',
+            'episode_number': 22,
+            'age_limit': 7,
+            'series': 'DuckTales',
+            'thumbnail': r're:https?://.*?\.jpg',
+        },
+        'params': {
+            # AES-128: ffmpeg
+            'skip_download': True,
+        },
+    }, {
+        'url': 'https://disneynow.com/shows/ducktales-disney-channel',
+        'info_dict': {
+            'title': 'DuckTales TV Show',
+            'id': 'SH553923438',
+        },
+        'playlist_mincount': 30,
+        'expected_warnings': ['Ignoring subtitle tracks', ],
     }]
 
     def _extract_videos(self, brand, video_id='-1', show_id='-1'):
         display_id = video_id if video_id != '-1' else show_id
-        return self._download_json(
-            'http://api.contents.watchabc.go.com/vp2/ws/contents/3000/videos/%s/001/-1/%s/-1/%s/-1/-1.json' % (brand, show_id, video_id),
-            display_id)['video']
+        data = self._download_json(
+            'http://api.contents.watchabc.go.com/vp2/ws/contents/4000/videos/%s/001/-1/%s/-1/%s/-1/-1.json' % (brand, show_id, video_id),
+            display_id, fatal=False)
+        return try_get(data, lambda x: x['video'], list)
+
+    def _raise_extractor_error(self, video_id, msg, expected=True):
+        raise ExtractorError('%s said: %s: %s' % (self.IE_NAME, video_id, msg), expected=expected)
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
-        sub_domain = mobj.group('sub_domain') or mobj.group('sub_domain_2')
+        sub_domain = re.sub(r'^(?:www\.)?(.+?)(?:\.go)?$', r'\1', mobj.group('sub_domain') or '')
         video_id, display_id = mobj.group('id', 'display_id')
+        display_id = display_id or video_id
         site_info = self._SITE_INFO.get(sub_domain, {})
         brand = site_info.get('brand')
         if not video_id or not site_info:
-            webpage = self._download_webpage(url, display_id or video_id)
+            webpage = self._download_webpage(url, display_id, expected_status=[404])
+            if re.search(r'(?:>|=\s*")Page not found\s', webpage):
+                self._raise_extractor_error(display_id, 'Page not found')
             data = self._parse_json(
                 self._search_regex(
                     r'["\']__abc_com__["\']\s*\]\s*=\s*({.+?})\s*;', webpage,
                     'data', default='{}'),
-                display_id or video_id, fatal=False)
-            # https://abc.com/shows/modern-family/episode-guide/season-01/101-pilot
+                display_id, fatal=False)
+            # https://abc.com/shows/ExtractorErrmodern-family/episode-guide/season-01/101-pilot
             layout = try_get(data, lambda x: x['page']['content']['video']['layout'], dict)
             video_id = None
             if layout:
@@ -181,34 +248,46 @@ class GoIE(AdobePassIE):
                     (
                         # There may be inner quotes, e.g. data-video-id="'VDKA3609139'"
                         # from http://freeform.go.com/shows/shadowhunters/episodes/season-2/1-this-guilty-blood
-                        r'data-video-id=["\']*(VDKA\w+)',
+                        r'data-video-id\s*=\s*["\']{,2}(VDKA\w+)',
                         # page.analytics.videoIdCode
                         r'\bvideoIdCode["\']\s*:\s*["\']((?:vdka|VDKA)\w+)',
+                        # This obsolete pattern, if matched, overrides detection of DisneyNow show pages
                         # https://abc.com/shows/the-rookie/episode-guide/season-02/03-the-bet
-                        r'\b(?:video)?id["\']\s*:\s*["\'](VDKA\w+)'
+                        # r'\b(?:video)?id["\']\s*:\s*["\'](VDKA\w+)'
                     ), webpage, 'video id', default=video_id)
             if not site_info:
                 brand = self._search_regex(
                     (r'data-brand=\s*["\']\s*(\d+)',
-                     r'data-page-brand=\s*["\']\s*(\d+)'), webpage, 'brand',
-                    default='004')
+                     r'data-page-brand=\s*["\']\s*(\d+)',
+                     r'"brand"\s*:\s*"(\d+)"'), webpage, 'brand',
+                    # default to a brand that has an active domain
+                    default='011')
                 site_info = next(
-                    si for _, si in self._SITE_INFO.items()
+                    si for si in self._SITE_INFO.values()
                     if si.get('brand') == brand)
             if not video_id:
                 # show extraction works for Disney, DisneyJunior and DisneyXD
                 # ABC and Freeform has different layout
                 show_id = self._search_regex(r'data-show-id=["\']*(SH\d+)', webpage, 'show id')
                 videos = self._extract_videos(brand, show_id=show_id)
-                show_title = self._search_regex(r'data-show-title="([^"]+)"', webpage, 'show title', fatal=False)
+                show_title = (
+                    self._search_regex(r'data-show-title="([^"]+)"', webpage, 'show title', default=None)
+                    or self._og_search_title(webpage, fatal=False))
+                if show_title:
+                    show_title = re.sub(r'^(?:Watch\s+)?(.+?)(?:\s*\|\s*Disney.*)?$', r'\1', show_title)
                 entries = []
                 for video in videos:
                     entries.append(self.url_result(
                         video['url'], 'Go', video.get('id'), video.get('title')))
                 entries.reverse()
                 return self.playlist_result(entries, show_id, show_title)
-        video_data = self._extract_videos(brand, video_id)[0]
-        video_id = video_data['id']
+        video_data = self._extract_videos(brand, video_id)
+        if not video_data:
+            self._raise_extractor_error(video_id, 'No video found')
+        video_id = try_get(video_data, lambda x: x[0]['id'], compat_str)
+        if not video_id:
+            self._raise_extractor_error(video_id, 'No video data', expected=False)
+        video_data = video_data[0]
         title = video_data['title']
 
         formats = []
@@ -249,7 +328,7 @@ class GoIE(AdobePassIE):
                             self.raise_geo_restricted(
                                 error['message'], countries=['US'])
                     error_message = ', '.join([error['message'] for error in errors])
-                    raise ExtractorError('%s said: %s' % (self.IE_NAME, error_message), expected=True)
+                    self._raise_extractor_error(video_id, error_message)
                 asset_url += '?' + entitlement['uplynkData']['sessionKey']
                 formats.extend(self._extract_m3u8_formats(
                     asset_url, video_id, 'mp4', m3u8_id=format_id or 'hls', fatal=False))
@@ -262,7 +341,8 @@ class GoIE(AdobePassIE):
                 if re.search(r'(?:/mp4/source/|_source\.mp4)', asset_url):
                     f.update({
                         'format_id': ('%s-' % format_id if format_id else '') + 'SOURCE',
-                        'preference': 1,
+                        'quality': 1,
+                        # 'preference': 1,
                     })
                 else:
                     mobj = re.search(r'/(\d+)x(\d+)/', asset_url)

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'/(?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$',
-        r'\b(?P<id>vfl[a-zA-Z0-9_-]+)\b.*?\.js$',
+        r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})/(?:tv-)?player',
+        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$',
+        r'\b(?P<id>vfl[a-zA-Z0-9_-]{6,})\b.*?\.js$',
     )
     _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(cls, player_url):
-        for player_re in cls._PLAYER_INFO_RE:
-            id_m = re.search(player_re, player_url)
-            if id_m:
-                break
-        else:
-            raise ExtractorError('Cannot identify player %r' % player_url)
-        return id_m.group('id')
+    def _extract_player_info(self, player_url):
+        try:
+            return self._search_regex(
+                self._PLAYER_INFO_RE, player_url, 'player info', group='id')
+        except ExtractorError as e:
+            raise ExtractorError(
+                'Cannot identify player %r' % (player_url,), cause=e)
 
     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)
-        if var_ay:
-            func_code = (func_code[0], ';\n'.join((var_ay, func_code[1])))
+        func_code = self._extract_sig_fn(jsi, func_name)
 
         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,
-            ('captions', 'playerCaptionsTracklistRenderer', T(dict)))
+            (player_response, api_player_response),
+            (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(',')],