Merge 66654bb1f41582cae2c70bfca3c03865759eb17e 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/extractors.py

@@ -679,7 +679,12 @@ from .mediasite import (
     MediasiteNamedCatalogIE,
 )
 from .medici import MediciIE
-from .megaphone import MegaphoneIE
+from .megaphone import (
+    MegaphoneIE,
+    MegaphoneEpisodeIE,
+    MegaphonePlaylistIE,
+    MegaphoneChannelIE,
+)
 from .meipai import MeipaiIE
 from .melonvod import MelonVODIE
 from .meta import METAIE

youtube_dl/extractor/megaphone.py

@@ -4,28 +4,49 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..utils import js_to_json
+from ..utils import (
+    clean_html,
+    dict_get,
+    get_element_by_class,
+    js_to_json,
+    parse_duration,
+    parse_iso8601,
+    str_or_none,
+    try_get,
+)
+from ..compat import (
+    compat_etree_Element,
+    compat_etree_fromstring,
+    compat_str,
+    compat_xpath,
+)
 
 
 class MegaphoneIE(InfoExtractor):
     IE_NAME = 'megaphone.fm'
     IE_DESC = 'megaphone.fm embedded players'
-    _VALID_URL = r'https://player\.megaphone\.fm/(?P<id>[A-Z0-9]+)'
-    _TEST = {
-        'url': 'https://player.megaphone.fm/GLT9749789991?"',
+    _PLAYER_URL_TEMPL = 'https://player.megaphone.fm/%s'
+    _VALID_URL_TEMPL = _PLAYER_URL_TEMPL.replace('.', r'\.')
+    _VALID_URL = _VALID_URL_TEMPL % r'(?P<id>[A-Z0-9]+)'
+    _JSON_URL_TEMPL = _PLAYER_URL_TEMPL % 'playlist/episode/%s'
+    _TESTS = [{
+        'url': 'https://player.megaphone.fm/GLT9749789991',
         'md5': '4816a0de523eb3e972dc0dda2c191f96',
         'info_dict': {
             'id': 'GLT9749789991',
             'ext': 'mp3',
             'title': '#97 What Kind Of Idiot Gets Phished?',
-            'thumbnail': r're:^https://.*\.png.*$',
-            'duration': 1776.26375,
-            'author': 'Reply All',
+            'thumbnail': r're:^https://.*\.png(?:\?.+)?$',
+            'duration': 2013.36,
+            'uploader': 'Reply All',
+            'upload_date': '20170518',
+            'timestamp': 1495101600,
+            'description': 'md5:8fc2ba1da0efb099ef928df127358a90',
         },
-    }
+    }]
 
-    def _real_extract(self, url):
-        video_id = self._match_id(url)
+    def _old_real_extract(self, url, video_id):
+        """version for pages before React-ification"""
         webpage = self._download_webpage(url, video_id)
 
         title = self._og_search_property('audio:title', webpage)
@@ -45,11 +66,223 @@ class MegaphoneIE(InfoExtractor):
             'thumbnail': thumbnail,
             'title': title,
             'author': author,
-            'duration': episode_data['duration'],
+            'duration': episode_data.get('duration'),
             'formats': formats,
         }
 
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        episode_json = self._download_json(self._JSON_URL_TEMPL % video_id, video_id, fatal=False)
+        if episode_json is False:
+            # probably, no pages match the old structure, but try anyway
+            return self._old_real_extract(url, video_id)
+        entries = []
+        for e in try_get(episode_json, lambda x: x['episodes'], list) or []:
+            title = try_get(e, lambda x: x['title'], compat_str)
+            if not title:
+                continue
+            video_url = dict_get(e, ('episodeUrlHRef', 'audioURL'))
+            if not video_url:
+                continue
+            entry = {
+                'id': e.get('UID') or video_id,
+                'title': title,
+                'description': clean_html(e.get('summary')),
+                'alt_title': e.get('subtitle'),
+                'formats': [{'url': video_url}],
+                'thumbnail': e.get('imageUrl'),
+                'duration': parse_duration(e.get('duration')),
+                'timestamp': parse_iso8601(e.get('pubDate')),
+            }
+            uploader = episode_json.get('podcastTitle')
+            if uploader:
+                entry['uploader'] = uploader
+                entry['author'] = uploader
+            entries.append(entry)
+        if entries:
+            if len(entries) == 1:
+                return entries[0]
+            return self.playlist_result(entries, playlist_id=video_id, playlist_title=episode_json.get('podcastTitle'))
+
     @classmethod
     def _extract_urls(cls, webpage):
         return [m[0] for m in re.findall(
             r'<iframe[^>]*?\ssrc=["\'](%s)' % cls._VALID_URL, webpage)]
+
+
+class MegaphoneEpisodeIE(MegaphoneIE):
+    IE_NAME = 'megaphone.fm:episode'
+    IE_DESC = 'megaphone.fm episode'
+    _VALID_URL_TEMPL = r'https://playlist\.megaphone\.fm/?%s'
+    _VALID_URL = _VALID_URL_TEMPL % r'\?e=(?P<id>[A-Z0-9]+)'
+    _JSON_URL_TEMPL = MegaphoneIE._PLAYER_URL_TEMPL % 'playlist/episode/%s'
+    _TESTS = [{
+        'url': 'https://playlist.megaphone.fm/?e=PAN7405681599',
+        'md5': '7fa866c3af93caac7e13a579c183f6ab',
+        'info_dict': {
+            'id': 'PAN7405681599',
+            'ext': 'mp3',
+            'title': 'Nirvana - Nevermind: 30 Years Later with Danny Goldberg',
+            'thumbnail': r're:^https://.*\.jpe?g(?:\?.+)?$',
+            'duration': 3576.94,
+            'uploader': 'Cobras & Fire: Comedy / Rock Talk Show',
+            'upload_date': '20210810',
+            'timestamp': 1628578800,
+            'description': 'md5:8f5623a8b22d3be4420c4570d0e36b69',
+        },
+    }]
+
+
+class MegaphonePlaylistIE(MegaphoneEpisodeIE):
+    IE_NAME = 'megaphone.fm:playlist'
+    IE_DESC = 'megaphone.fm playlist'
+    _VALID_URL = MegaphoneEpisodeIE._VALID_URL_TEMPL % r'\?p=(?P<id>[A-Z0-9]+)'
+    _JSON_URL_TEMPL = MegaphoneIE._PLAYER_URL_TEMPL % 'playlist/%s'
+    _TESTS = [{
+        'url': 'https://playlist.megaphone.fm/?p=DEM6640968282',
+        'info_dict': {
+            'id': 'DEM6640968282',
+            'title': 'Lightbulb Productions',
+        },
+        'playlist_mincount': 6,
+    }, {
+        'url': 'https://playlist.megaphone.fm/?p=DEM6640968282',
+        'md5': '71fbb6616c75aa2cc972e978683dffd4',
+        'info_dict': {
+            'id': 'DEM6640968282',
+            'ext': 'mp3',
+            'title': 'Open Source World',
+            'thumbnail': r're:^https://.*\.jpe?g(?:\?.+)?$',
+            'duration': 754.38,
+            'uploader': 'Lightbulb Productions',
+            'upload_date': '20200602',
+            'timestamp': 1591070400,
+            'description': 'md5:a06a5a078c0d98bb023626615fb1432d',
+        },
+        'params': {
+            'noplaylist': True,
+        },
+    }]
+
+    def _real_extract(self, url):
+        entries = super(MegaphonePlaylistIE, self)._real_extract(url)
+        if entries:
+            noplaylist = self._downloader.params.get('noplaylist')
+            if noplaylist:
+                self.to_screen('Downloading just the first episode because of --no-playlist')
+                return entries['entries'][0]
+        return entries
+
+
+class MegaphoneChannelIE(MegaphoneIE):
+    IE_NAME = 'megaphone.fm:channel'
+    IE_DESC = 'megaphone.fm channel'
+    _VALID_URL = r'https://cms\.megaphone\.fm/channel/(?P<id>[A-Z0-9]+)(?:\?selected=(?P<clip_id>[A-Z0-9]+))?'
+    _TESTS = [{
+        'url': 'https://cms.megaphone.fm/channel/ADL3707263633',
+        'info_dict': {
+            'id': 'ADL3707263633',
+            'title': 'Pax Britannica',
+            'description': 'md5:7b4002330ffe4abcb81d97ab9b56fede',
+        },
+        'playlist_mincount': 98,
+    }, {
+        'url': 'https://cms.megaphone.fm/channel/ADL3707263633?selected=ADL9449136081',
+        'md5': '42901d1112c059a8de374046e0b1ed25',
+        'info_dict': {
+            'id': 'ADL9449136081',
+            'title': '02.23 - Nolumus Leges Angliae Mutari',
+            'description': 'md5:d35989ec81de7199b3020bc919ab7c0d',
+            'ext': 'mp3',
+            'thumbnail': r're:^https://.*\.png(?:\?.+)?$',
+            'duration': 2470.09,
+            'extractor_key': 'Megaphone',
+            'upload_date': '20210711',
+            'timestamp': 1625961600,
+            'uploader': 'Pax Britannica',
+        },
+    }, {
+        'url': 'https://cms.megaphone.fm/channel/ADL3707263633',
+        'md5': '81156c760235d45a9133a9ea9ccbb7d0',
+        'info_dict': {
+            'id': 'ADL9716153485',
+            'title': '02.24 - Give Unto Caesar His Due',
+            'description': 'The First English Civil War begins.',
+            'ext': 'mp3',
+            'duration': 1860,
+        },
+        'params': {
+            'noplaylist': True,
+        },
+    }]
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url).groupdict()
+        video_id = mobj['id']
+        # If clip is selected, dl that instead
+        clip_id = mobj.get('clip_id')
+        if clip_id:
+            return self.url_result(self._PLAYER_URL_TEMPL % clip_id, ie='Megaphone', video_id=clip_id)
+        webpage = self._download_webpage(url, video_id)
+        # Useful information is split between a JS JSON.parse() call and
+        # a <div class="public-ep-list>...</div> HTML element
+        playlist_json = self._search_regex(
+            r"var\s+playlist\s*=\s*JSON\.parse\s*\(\s*('[^<]+?')\s*\)\s*;",
+            webpage, 'playlist JSON')
+        # The data must be parsed twice: once for JSON.parse, once for Python
+        playlist_json = self._parse_json(
+            playlist_json, video_id,
+            transform_source=lambda s: self._parse_json(s, video_id, transform_source=js_to_json))
+        entries = []
+        ep_list = None
+        # Support --no-playlist to get the first item when no explicit selection
+        noplaylist = self._downloader.params.get('noplaylist')
+        for ep_num, episode in enumerate(playlist_json):
+            ep_url = episode.get('mp3')
+            ep_title = episode.get('title')
+            ep_id = episode.get('uid')
+            if not (ep_url and ep_id):
+                continue
+            entry = {
+                'url': ep_url,
+                'id': ep_id,
+            }
+            if ep_num == 0:
+                # As there are items to process, initialise the public-ep-list
+                ep_list = self._search_regex(
+                    r'(?s)<div\s[^>]*?class\s*=\s*("|\')public-ep-list\1[^>]*>.*?(?P<ep_list><div\s[^>]*?id\s*=.+?</div>\s*</div>)\s*</div>',
+                    webpage, 'episode list', default=None, group='ep_list')
+                if ep_list:
+                    # Although the page itself isn't well-formed as XML, the
+                    # <div class="public-ep-list>...</div> HTML element is (apparently)
+                    # So for "easy" extraction, make it into a compat_etree_Element
+                    ep_list = '<?xml version="1.0" encoding="UTF-8"?><eplist>%s</eplist>' % ep_list
+                    ep_list = ep_list.encode('utf-8')
+                    ep_list = try_get(ep_list, compat_etree_fromstring, compat_etree_Element)
+            if ep_list is not None:
+                # Use the subset XPath syntax to extract the additional data (counted from 0)
+                ep_info = ep_list.find(compat_xpath("div[@id='%d']" % ep_num))
+                if ep_info is not None:
+                    if not ep_title:
+                        title = ep_info.find(compat_xpath(".//div[@class='ep-title']"))
+                        if title:
+                            ep_title = str_or_none(title.text)
+                    description = ep_info.find(compat_xpath(".//div[@class='ep-summary']"))
+                    if description is not None:
+                        entry['description'] = str_or_none(description.text)
+                    duration = ep_info.find(compat_xpath(".//div[@class='ep-duration']"))
+                    if duration is not None:
+                        entry['duration'] = parse_duration(duration.text)
+            if not ep_title:
+                continue
+            entry['title'] = ep_title
+            if ep_num == 0 and noplaylist:
+                self.to_screen('Downloading just the first episode because of --no-playlist')
+                return entry
+            entries.append(entry)
+        if entries:
+            title = (self._og_search_property('audio:artist', webpage, default=None)
+                     or get_element_by_class('title', webpage))
+            description = get_element_by_class('summary', webpage)
+            return self.playlist_result(entries, video_id, title, description)

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)
@@ -1204,9 +1208,10 @@ 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(
-                        ('' if x in (None, JS_Undefined) else _js_toString(x))
-                        for x in obj)
+                    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':
                     assertion(not argvals, 'does not take any arguments')
                     obj.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(',')],