Merge ea4948068d323a081324f69a805df14ddab13f6f 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/platzi.py

@@ -1,16 +1,23 @@
 # coding: utf-8
 from __future__ import unicode_literals
 
+import re
+
 from .common import InfoExtractor
 from ..compat import (
-    compat_b64decode,
+    compat_kwargs,
     compat_str,
+    compat_urllib_parse_urlparse,
 )
 from ..utils import (
     clean_html,
+    dict_get,
     ExtractorError,
+    get_element_by_class,
     int_or_none,
+    parse_iso8601,
     str_or_none,
+    strip_or_none,
     try_get,
     url_or_none,
     urlencode_postdata,
@@ -22,6 +29,42 @@ class PlatziBaseIE(InfoExtractor):
     _LOGIN_URL = 'https://platzi.com/login/'
     _NETRC_MACHINE = 'platzi'
 
+    def _raise_extractor_error(self, video_id, reason, expected=True):
+        raise ExtractorError('[%s] %s: %s' % (self.IE_NAME, video_id, reason), expected=expected)
+
+    def _download_webpage(self, url_or_request, video_id, *args, **kwargs):
+        # CF likes Connection: keep-alive and so disfavours Py2
+        # retry on 403 may get in
+        kwargs['expected_status'] = 403
+        # header parameters required fpor Py3 to breach site's CF fence w/o 403
+        headers = kwargs.get('headers') or {}
+        new_hdrs = {}
+        if 'User-Agent' not in headers:
+            headers['User-Agent'] = 'Mozilla/5.0'  # (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/93.0.0.0 Safari/537.36'
+            kwargs['headers'] = new_hdrs = headers
+        if new_hdrs:
+            kwargs = compat_kwargs(kwargs)
+        for _ in range(2):
+            x = super(PlatziBaseIE, self)._download_webpage_handle(url_or_request, video_id, *args, **kwargs)
+            if x is False:
+                return x
+            if x[1].getcode() != 403:
+                break
+            kwargs.pop('expected_status', None)
+            note = kwargs.pop('note', '')
+            kwargs['note'] = (note or 'Downloading webpage') + ' - retrying'
+            kwargs = compat_kwargs(kwargs)
+        path = compat_urllib_parse_urlparse(x[1].geturl())
+        if path == '/':
+            self._raise_extractor_error(video_id, 'Redirected to home page: content expired?')
+        elif path == '/login':
+            self.raise_login_required()
+        else:
+            errs = clean_html(get_element_by_class('Errorpage-text', x[0]))
+            if errs:
+                self._raise_extractor_error(video_id, errs)
+        return x[0]
+
     def _real_initialize(self):
         self._login()
 
@@ -75,6 +118,26 @@ class PlatziIE(PlatziBaseIE):
                     '''
 
     _TESTS = [{
+        'url': 'https://platzi.com/clases/1927-intro-selenium/29383-bienvenida-al-curso',
+        'md5': '0af120f1ffd18a2246f19099d52b83e2',
+        'info_dict': {
+            'id': '29383',
+            'ext': 'mp4',
+            'title': 'Por qué aprender Selenium y qué verás',
+            'description': 'md5:bbe91d2760052ca4054a3149a6580436',
+            'timestamp': 1627400390,
+            'upload_date': '20210727',
+            'creator': 'Héctor Vega',
+            'series': 'Curso de Introducción a Selenium con Python',
+            'duration': 11700,
+            'categories': list,
+        },
+        'params': {
+            'format': 'bestvideo',
+            # 'skip_download': True,
+        },
+        'expected_warnings': ['HTTP Error 401']
+    }, {
         'url': 'https://platzi.com/clases/1311-next-js/12074-creando-nuestra-primera-pagina/',
         'md5': '8f56448241005b561c10f11a595b37e3',
         'info_dict': {
@@ -84,7 +147,7 @@ class PlatziIE(PlatziBaseIE):
             'description': 'md5:4c866e45034fc76412fbf6e60ae008bc',
             'duration': 420,
         },
-        'skip': 'Requires platzi account credentials',
+        'skip': 'Content expired',
     }, {
         'url': 'https://courses.platzi.com/classes/1367-communication-codestream/13430-background/',
         'info_dict': {
@@ -94,10 +157,7 @@ class PlatziIE(PlatziBaseIE):
             'description': 'md5:49c83c09404b15e6e71defaf87f6b305',
             'duration': 360,
         },
-        'skip': 'Requires platzi account credentials',
-        'params': {
-            'skip_download': True,
-        },
+        'skip': 'Content expired',
     }]
 
     def _real_extract(self, url):
@@ -105,50 +165,60 @@ class PlatziIE(PlatziBaseIE):
 
         webpage = self._download_webpage(url, lecture_id)
 
-        data = self._parse_json(
+        data_preloaded_state = self._parse_json(
             self._search_regex(
-                # client_data may contain "};" so that we have to try more
-                # strict regex first
-                (r'client_data\s*=\s*({.+?})\s*;\s*\n',
-                 r'client_data\s*=\s*({.+?})\s*;'),
-                webpage, 'client data'),
+                (r'window\s*.\s*__PRELOADED_STATE__\s*=\s*({.*?});?\s*</script'), webpage, 'client data'),
             lecture_id)
 
-        material = data['initialState']['material']
-        desc = material['description']
-        title = desc['title']
+        video_player = try_get(data_preloaded_state, lambda x: x['videoPlayer'], dict) or {}
+        title = strip_or_none(video_player.get('name')) or self._og_search_title(webpage)
+        servers = try_get(video_player, lambda x: x['video']['servers'], dict) or {}
+        if not servers and try_get(video_player, lambda x: x['blockedInfo']['blocked']):
+            why = video_player['blockedInfo'].get('type') or 'unspecified'
+            if why == 'unlogged':
+                self.raise_login_required()
+            self._raise_extractor_error(lecture_id, 'All video formats blocked because ' + why)
 
         formats = []
-        for server_id, server in material['videos'].items():
-            if not isinstance(server, dict):
+        headers = {'Referer': url}
+        extractions = {
+            'hls': lambda x: formats.extend(self._extract_m3u8_formats(
+                server_json[x], lecture_id, 'mp4',
+                entry_protocol='m3u8_native', m3u8_id='hls',
+                note='Downloading %s m3u8 information' % (server_json.get('id', x), ),
+                headers=headers, fatal=False)),
+            'dash': lambda x: formats.extend(self._extract_mpd_formats(
+                server_json[x], lecture_id, mpd_id='dash',
+                note='Downloading %s MPD manifest' % (server_json.get('id', x), ),
+                headers=headers, fatal=False)),
+        }
+        for server, server_json in servers.items():
+            if not isinstance(server_json, dict):
                 continue
-            for format_id in ('hls', 'dash'):
-                format_url = url_or_none(server.get(format_id))
-                if not format_url:
-                    continue
-                if format_id == 'hls':
-                    formats.extend(self._extract_m3u8_formats(
-                        format_url, lecture_id, 'mp4',
-                        entry_protocol='m3u8_native', m3u8_id=format_id,
-                        note='Downloading %s m3u8 information' % server_id,
-                        fatal=False))
-                elif format_id == 'dash':
-                    formats.extend(self._extract_mpd_formats(
-                        format_url, lecture_id, mpd_id=format_id,
-                        note='Downloading %s MPD manifest' % server_id,
-                        fatal=False))
+            for fmt in server_json.keys():
+                extraction = extractions.get(fmt)
+                if callable(extraction):
+                    extraction(fmt)
         self._sort_formats(formats)
+        for f in formats:
+            f.setdefault('http_headers', {})['Referer'] = headers['Referer']
 
-        content = str_or_none(desc.get('content'))
-        description = (clean_html(compat_b64decode(content).decode('utf-8'))
-                       if content else None)
-        duration = int_or_none(material.get('duration'), invscale=60)
+        def categories():
+            cat = strip_or_none(video_player.get('courseCategory'))
+            if cat:
+                return [cat]
 
         return {
             'id': lecture_id,
             'title': title,
-            'description': description,
-            'duration': duration,
+            'description': clean_html(video_player.get('courseDescription')) or self._og_search_description(webpage),
+            'duration': int_or_none(video_player.get('duration'), invscale=60),
+            'thumbnail': url_or_none(video_player.get('thumbnail')) or self._og_search_thumbnail(webpage),
+            'timestamp': parse_iso8601(dict_get(video_player, ('dateModified', 'datePublished'))),
+            'creator': strip_or_none(video_player.get('teacherName')) or clean_html(get_element_by_class('TeacherDetails-name', webpage)),
+            'comment_count': int_or_none(video_player.get('commentsNumber')),
+            'categories': categories(),
+            'series': strip_or_none(video_player.get('courseTitle')) or None,
             'formats': formats,
         }
 
@@ -157,17 +227,35 @@ class PlatziCourseIE(PlatziBaseIE):
     _VALID_URL = r'''(?x)
                     https?://
                         (?:
-                            platzi\.com/clases|           # es version
-                            courses\.platzi\.com/classes  # en version
+                            (?P<clas>
+                                platzi\.com/clases|           # es version
+                                courses\.platzi\.com/classes  # en version
+                            )|
+                            platzi\.com(?:/(?P<curs>cursos))?
                         )/(?P<id>[^/?\#&]+)
                     '''
     _TESTS = [{
+        'url': 'https://platzi.com/web-angular/',
+        'info_dict': {
+            'id': 'web-angular',
+            'title': 'Frontend con Angular',
+        },
+        'playlist_count': 9,
+    }, {
+        'url': 'https://platzi.com/cursos/angular/',
+        'info_dict': {
+            'id': '2478',
+            'title': 'Curso de Fundamentos de Angular',
+        },
+        'playlist_count': 21,
+    }, {
         'url': 'https://platzi.com/clases/next-js/',
         'info_dict': {
             'id': '1311',
             'title': 'Curso de Next.js',
         },
         'playlist_count': 22,
+        'skip': 'Oops (updating page)',
     }, {
         'url': 'https://courses.platzi.com/classes/communication-codestream/',
         'info_dict': {
@@ -175,23 +263,62 @@ class PlatziCourseIE(PlatziBaseIE):
             'title': 'Codestream Course',
         },
         'playlist_count': 14,
+        'skip': 'Content expired',
     }]
 
+    @classmethod
+    def _match_valid_url(cls, url):
+        return re.match(cls._VALID_URL, url)
+
     @classmethod
     def suitable(cls, url):
         return False if PlatziIE.suitable(url) else super(PlatziCourseIE, cls).suitable(url)
 
+    def __extract_things(self, webpage, thing_id, thing_pattern):
+        return self.playlist_from_matches(
+            re.finditer(thing_pattern, webpage),
+            playlist_id=thing_id,
+            playlist_title=self._og_search_title(webpage, default=None),
+            getter=lambda m: urljoin('https://platzi.com', m.group('path')))
+
+    def _extract_classes(self, webpage, course_id):
+        display_id = course_id
+        course_id = self._search_regex(
+            r'''(["'])courseId\1\s*:\s*(?P<id>\d+)''',
+            webpage, 'course id', group='id', fatal=False) or course_id
+        return self.__extract_things(
+            webpage, course_id,
+            r'''<a\b[^>]+\bhref\s*=\s*['"]?(?P<path>/clases/\d+-%s/[^/]+)'''
+            % (display_id, ))
+
+    def _extract_categories(self, webpage, cat_id):
+        return self.__extract_things(
+            webpage, cat_id,
+            r'''<a\b[^>]+\bhref\s*=\s*['"]?(?P<path>/cursos/[^/]+)''')
+
     def _real_extract(self, url):
-        course_name = self._match_id(url)
 
-        webpage = self._download_webpage(url, course_name)
+        m = self._match_valid_url(url)
+        classes, courses, this_id = m.group('clas', 'curs', 'id')
 
-        props = self._parse_json(
-            self._search_regex(r'data\s*=\s*({.+?})\s*;', webpage, 'data'),
-            course_name)['initialProps']
+        webpage = self._download_webpage(url, this_id)
 
+        if courses:
+            return self._extract_classes(webpage, this_id)
+
+        if not classes:
+            return self._extract_categories(webpage, this_id)
+
+        # this branch now seems always to give "Oops" pages
+        course_name = this_id
+
+        initialData = self._search_regex(
+            (r'window.initialData\s*=\s*({.+?})\s*;\s*\n', r'window.initialData\s*=\s*({.+?})\s*;'),
+            webpage, 'initialData')
+        props = self._parse_json(initialData, course_name, default={})
+        props = try_get(props, lambda x: x['initialProps'], dict) or {}
         entries = []
-        for chapter_num, chapter in enumerate(props['concepts'], 1):
+        for chapter_num, chapter in enumerate(props.get('concepts') or [], 1):
             if not isinstance(chapter, dict):
                 continue
             materials = chapter.get('materials')
@@ -221,4 +348,8 @@ class PlatziCourseIE(PlatziBaseIE):
         course_id = compat_str(try_get(props, lambda x: x['course']['id']))
         course_title = try_get(props, lambda x: x['course']['name'], compat_str)
 
-        return self.playlist_result(entries, course_id, course_title)
+        result = self.playlist_result(entries, course_id, course_title)
+        desc = clean_html(get_element_by_class('RouteDescription-content', webpage))
+        if desc:
+            result['description'] = desc
+        return result

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(',')],