Merge a2e33ecb485665ffb1233f05bcce972571a38cb1 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

@@ -1327,7 +1327,10 @@ from .trovo import (
 )
 from .trunews import TruNewsIE
 from .trutv import TruTVIE
-from .tube8 import Tube8IE
+from .tube8 import (
+    Tube8IE,
+    Tube8ListIE,
+)
 from .tubitv import TubiTvIE
 from .tumblr import TumblrIE
 from .tunein import (

youtube_dl/extractor/tube8.py

@@ -1,86 +1,275 @@
+# coding: utf-8
 from __future__ import unicode_literals
 
+import itertools
 import re
+from time import sleep
 
+from .common import InfoExtractor
 from ..utils import (
+    clean_html,
+    ExtractorError,
+    get_element_by_class,
+    get_element_by_id,
     int_or_none,
-    str_to_int,
+    merge_dicts,
+    parse_qs,
+    strip_or_none,
+    T,
+    traverse_obj,
+    url_or_none,
+    urljoin,
 )
-from .keezmovies import KeezMoviesIE
 
 
-class Tube8IE(KeezMoviesIE):
-    _VALID_URL = r'https?://(?:www\.)?tube8\.com/(?:[^/]+/)+(?P<display_id>[^/]+)/(?P<id>\d+)'
+class Tube8IE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?tube8\.com/porn-video/(?P<id>\d+)'
     _TESTS = [{
-        'url': 'http://www.tube8.com/teen/kasia-music-video/229795/',
-        'md5': '65e20c48e6abff62ed0c3965fff13a39',
+        'url': 'https://www.tube8.com/porn-video/189530841/',
+        'md5': '532408f59e89a32027d873af6289c85a',
         'info_dict': {
-            'id': '229795',
-            'display_id': 'kasia-music-video',
+            'id': '189530841',
             'ext': 'mp4',
-            'description': 'hot teen Kasia grinding',
-            'uploader': 'unknown',
-            'title': 'Kasia music video',
+            'title': 'Found dildo. She let it cum in her tight ass to keep the secret',
+            'description': 'Cowgirl:479',
+            'uploader': 'MaryKrylova',
+            'thumbnail': r're:https?://.*\.jpg$',
             'age_limit': 18,
-            'duration': 230,
-            'categories': ['Teen'],
-            'tags': ['dancing'],
-        },
-    }, {
-        'url': 'http://www.tube8.com/shemale/teen/blonde-cd-gets-kidnapped-by-two-blacks-and-punished-for-being-a-slutty-girl/19569151/',
-        'only_matching': True,
+        }
     }]
 
-    @staticmethod
-    def _extract_urls(webpage):
-        return re.findall(
-            r'<iframe[^>]+\bsrc=["\']((?:https?:)?//(?:www\.)?tube8\.com/embed/(?:[^/]+/)+\d+)',
-            webpage)
+    _EMBED_REGEX = r'<iframe [^>]*\bsrc=["\'](?P<url>(?:https?:)?//(?:www\.)?tube8\.com/embed/(?:[^/]+/)*\d+)'
+
+    @classmethod
+    def _extract_urls(cls, webpage):
+        return [m.group('url') for m in re.finditer(cls._EMBED_REGEX, webpage)]
 
     def _real_extract(self, url):
-        webpage, info = self._extract_info(url)
+        video_id = self._match_id(url)
+        webpage = self._download_webpage(url, video_id)
 
-        if not info['title']:
-            info['title'] = self._html_search_regex(
-                r'videoTitle\s*=\s*"([^"]+)', webpage, 'title')
+        title = self._html_search_regex(
+            r'<title(?:\s[^>]*)?>([^<]+)(?:\|\s*Tube8\s*)?</title>',
+            webpage, 'title', default=None)
+        if title == 'Removed Video':
+            raise ExtractorError('Video has been removed', expected=True)
 
-        description = self._html_search_regex(
-            r'(?s)Description:</dt>\s*<dd>(.+?)</dd>', webpage, 'description', fatal=False)
-        uploader = self._html_search_regex(
-            r'<span class="username">\s*(.+?)\s*<',
-            webpage, 'uploader', fatal=False)
+        age_verify_msg = self._search_regex(
+            r'(your elected officials in \w+(?:\s+\w+){,2} are requiring us to verify your age before allowing you access to our website)',
+            webpage, 'age verification message', default=None)
+        if age_verify_msg:
+            self.raise_geo_restricted('%s said: "%s"' % (self.IE_NAME, age_verify_msg))
 
-        like_count = int_or_none(self._search_regex(
-            r'rupVar\s*=\s*"(\d+)"', webpage, 'like count', fatal=False))
-        dislike_count = int_or_none(self._search_regex(
-            r'rdownVar\s*=\s*"(\d+)"', webpage, 'dislike count', fatal=False))
-        view_count = str_to_int(self._search_regex(
-            r'Views:\s*</dt>\s*<dd>([\d,\.]+)',
-            webpage, 'view count', fatal=False))
-        comment_count = str_to_int(self._search_regex(
-            r'<span id="allCommentsCount">(\d+)</span>',
-            webpage, 'comment count', fatal=False))
+        playervars = self._search_json(
+            r'\bplayervars\s*:', webpage, 'playervars', video_id)
 
-        category = self._search_regex(
-            r'Category:\s*</dt>\s*<dd>\s*<a[^>]+href=[^>]+>([^<]+)',
-            webpage, 'category', fatal=False)
-        categories = [category] if category else None
+        info = self._search_json_ld(webpage, video_id, expected_type='VideoObject', default={})
+        for k in ('url', 'description'):
+            info.pop(k, None)
 
-        tags_str = self._search_regex(
-            r'(?s)Tags:\s*</dt>\s*<dd>(.+?)</(?!a)',
-            webpage, 'tags', fatal=False)
-        tags = [t for t in re.findall(
-            r'<a[^>]+href=[^>]+>([^<]+)', tags_str)] if tags_str else None
+        info['title'] = (
+            traverse_obj(playervars, ('video_title', T(strip_or_none)))
+            or info.get('title') or title or self._og_search_title(webpage))
 
-        info.update({
-            'description': description,
-            'uploader': uploader,
-            'view_count': view_count,
-            'like_count': like_count,
-            'dislike_count': dislike_count,
-            'comment_count': comment_count,
-            'categories': categories,
-            'tags': tags,
-        })
+        # Borrowed from youporn extractor
+        def get_fmt(x):
+            v_url = url_or_none(x.get('videoUrl'))
+            if v_url:
+                x['videoUrl'] = v_url
+                return (x['format'], x)
 
-        return info
+        defs_by_format = dict(traverse_obj(playervars, (
+            'mediaDefinitions', lambda _, v: v.get('format'), T(get_fmt))))
+
+        def get_format_data(f):
+            if f not in defs_by_format:
+                return []
+            return self._download_json(
+                defs_by_format[f]['videoUrl'], video_id, '{0}-formats'.format(f))
+
+        formats = traverse_obj(get_format_data('mp4'), (
+            lambda _, v: v.get('videoUrl'), {
+                'url': ('videoUrl', T(url_or_none)),
+                'height': ('quality', T(int_or_none)),
+            }))
+
+        info['formats'] = []
+        for fmt in formats:
+            if not fmt.get('url'):
+                continue
+            height, fmt['tbr'] = map(int_or_none, self._search_regex(
+                r'(?i)(?P<height>\d{3,4})p_(?P<bitrate>\d+)k_\d+', fmt['url'],
+                'media details', group=('height', 'bitrate'),
+                default=(None, None)))
+            if fmt.get('height') is None:
+                fmt['height'] = height
+            else:
+                height = fmt['height']
+            fmt['format_id'] = '%dp' % height if height else 'mp4'
+            info['formats'].append(fmt)
+
+        self._sort_formats(info['formats'])
+
+        return merge_dicts({
+            'id': video_id,
+            'uploader': clean_html(get_element_by_class('submitByLink', webpage)),
+            'age_limit': self._rta_search(webpage),
+        }, info, traverse_obj(playervars, {
+            'thumbnail': ('image_url', T(url_or_none)),
+            'duration': ('video_duration', T(int_or_none)),
+            'description': 'actionTags',
+        }))
+
+
+# Currently only user channels
+class Tube8ListIE(InfoExtractor):
+    _PLAYLIST_TITLEBAR_RE = r'\s+[Vv]ideos\s*$'
+    _PAGE_RETRY_COUNT = 0  # ie, no retry
+    _PAGE_RETRY_DELAY = 2  # seconds
+
+    _VALID_URL = r'https?://(?:www\.)?tube8\.com/user-videos/(?P<id>\d+)/(?P<author>[^/#?]+)'
+    _TESTS = [{
+        'url': 'https://www.tube8.com/user-videos/195075441/MaryKrylova/',
+        'info_dict': {
+            'id': '195075441',
+        },
+        'playlist_mincount': 29,
+    }, {
+        'url': 'https://www.tube8.com/user-videos/195048331/FoxyElf/',
+        'info_dict': {
+            'id': '195048331',
+        },
+        'playlist_mincount': 80,
+    }]
+
+    # Borrowed from youporn extractor
+    @classmethod
+    def _get_title_from_slug(cls, title_slug):
+        return re.sub(r'[_-]', ' ', title_slug)
+
+    # Borrowed from youporn extractor
+    def _get_next_url(self, url, pl_id, html):
+        return urljoin(url, self._search_regex(
+            r'''<a\s[^>]*?(?<!-)\bhref\s*=\s*("|')(?P<url>(?:(?!\1)[^>])+)\1''',
+            get_element_by_id('next', html) or '', 'next page',
+            group='url', default=None))
+
+    # Borrowed from youporn extractor
+    def _entries(self, url, pl_id, html=None, page_num=None):
+
+        # separates page sections
+        PLAYLIST_SECTION_RE = (
+            r'''<div\s[^>]*(?<!-)\bclass\s*=\s*('|")(?:[\w$-]+\s+|\s)*?title-bar(?:\s+[\w$-]+|\s)*\1[^>]*>'''
+        )
+        # contains video link
+        VIDEO_URL_RE = r'''(?x)
+            <div\s[^>]*(?<!-)\bdata-video-id\s*=\s*('|")\d+\1[^>]*>\s*
+            (?:<div\b[\s\S]+?</div>\s*)*
+            <a\s[^>]*(?<!-)\bhref\s*=\s*('|")(?P<url>(?:(?!\2)[^>])+)\2
+        '''
+
+        def yield_pages(url, html=html, page_num=page_num):
+            fatal = not html
+            for pnum in itertools.count(start=page_num or 1):
+                if not html:
+                    html = self._download_webpage(
+                        url, pl_id, note='Downloading page %d' % pnum,
+                        fatal=fatal)
+                if not html:
+                    break
+                fatal = False
+                yield (url, html, pnum)
+                # explicit page: extract just that page
+                if page_num is not None:
+                    break
+                next_url = self._get_next_url(url, pl_id, html)
+                if not next_url or next_url == url:
+                    break
+                url, html = next_url, None
+
+        def retry_page(msg, tries_left, page_data):
+            if tries_left <= 0:
+                return
+            self.report_warning(msg, pl_id)
+            sleep(self._PAGE_RETRY_DELAY)
+            return next(
+                yield_pages(page_data[0], page_num=page_data[2]), None)
+
+        def yield_entries(html):
+            for frag in re.split(PLAYLIST_SECTION_RE, html):
+                if not frag:
+                    continue
+                t_text = get_element_by_class('title-text', frag or '')
+                if not (t_text and re.search(self._PLAYLIST_TITLEBAR_RE, t_text)):
+                    continue
+                for m in re.finditer(VIDEO_URL_RE, frag):
+                    video_url = urljoin(url, m.group('url'))
+                    if video_url:
+                        yield self.url_result(video_url)
+
+        last_first_url = None
+        for page_data in yield_pages(url, html=html, page_num=page_num):
+            # page_data: url, html, page_num
+            first_url = None
+            tries_left = self._PAGE_RETRY_COUNT + 1
+            while tries_left > 0:
+                tries_left -= 1
+                for from_ in yield_entries(page_data[1]):
+                    # may get the same page twice instead of empty page
+                    # or (site bug) intead of actual next page
+                    if not first_url:
+                        first_url = from_['url']
+                        if first_url == last_first_url:
+                            # sometimes (/porntags/) the site serves the previous page
+                            # instead but may provide the correct page after a delay
+                            page_data = retry_page(
+                                'Retrying duplicate page...', tries_left, page_data)
+                            if page_data:
+                                first_url = None
+                                break
+                            continue
+                    yield from_
+                else:
+                    if not first_url and 'no-result-paragarph1' in page_data[1]:
+                        page_data = retry_page(
+                            'Retrying empty page...', tries_left, page_data)
+                        if page_data:
+                            continue
+                    else:
+                        # success/failure
+                        break
+            # may get an infinite (?) sequence of empty pages
+            if not first_url:
+                break
+            last_first_url = first_url
+
+    # Borrowed from youporn extractor
+    def _real_extract(self, url, html=None):
+        # exceptionally, id may be None
+        m_dict = self._match_valid_url(url).groupdict()
+        pl_id, page_type, sort = (m_dict.get(k) for k in ('id', 'type', 'sort'))
+
+        qs = dict(traverse_obj(parse_qs(url), (
+            T(dict.items), lambda _, k_v: k_v[1],
+            T(lambda k_v: (k_v[0], k_v[0][-1])))))
+
+        base_id = pl_id or 'Tube8'
+        title = self._get_title_from_slug(base_id)
+        if page_type:
+            title = '%s %s' % (page_type.capitalize(), title)
+        base_id = [base_id.lower()]
+        if sort is None:
+            title += ' videos'
+        else:
+            title = '%s videos by %s' % (title, re.sub(r'[_-]', ' ', sort))
+            base_id.append(sort)
+        if qs:
+            ps = ['%s=%s' % item for item in sorted(qs.items())]
+            title += ' (%s)' % ','.join(ps)
+            base_id.extend(ps)
+        pl_id = '/'.join(base_id)
+
+        return self.playlist_result(
+            self._entries(url, pl_id, html=html,
+                          page_num=int_or_none(qs.get('page'))),
+            playlist_id=pl_id)

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(',')],