Merge 8b11dbba00 into c5098961b0

Now also succeeds with player b12cc44b

test/test_jsinterp.py

@@ -425,6 +425,34 @@ class TestJSInterpreter(unittest.TestCase):
             self._test(jsi, [''], args=['', '-'])
             self._test(jsi, [], args=['', ''])
 
+    def test_slice(self):
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice()}', [0, 1, 2, 3, 4, 5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(0)}', [0, 1, 2, 3, 4, 5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(5)}', [5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(99)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-2)}', [7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-99)}', [0, 1, 2, 3, 4, 5, 6, 7, 8])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(0, 0)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(1, 0)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(0, 1)}', [0])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(3, 6)}', [3, 4, 5])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(1, -1)}', [1, 2, 3, 4, 5, 6, 7])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-1, 1)}', [])
+        self._test('function f(){return [0, 1, 2, 3, 4, 5, 6, 7, 8].slice(-3, -1)}', [6, 7])
+        self._test('function f(){return "012345678".slice()}', '012345678')
+        self._test('function f(){return "012345678".slice(0)}', '012345678')
+        self._test('function f(){return "012345678".slice(5)}', '5678')
+        self._test('function f(){return "012345678".slice(99)}', '')
+        self._test('function f(){return "012345678".slice(-2)}', '78')
+        self._test('function f(){return "012345678".slice(-99)}', '012345678')
+        self._test('function f(){return "012345678".slice(0, 0)}', '')
+        self._test('function f(){return "012345678".slice(1, 0)}', '')
+        self._test('function f(){return "012345678".slice(0, 1)}', '0')
+        self._test('function f(){return "012345678".slice(3, 6)}', '345')
+        self._test('function f(){return "012345678".slice(1, -1)}', '1234567')
+        self._test('function f(){return "012345678".slice(-1, 1)}', '')
+        self._test('function f(){return "012345678".slice(-3, -1)}', '67')
+
 
 if __name__ == '__main__':
     unittest.main()

test/test_youtube_signature.py

@@ -174,6 +174,14 @@ _NSIG_TESTS = [
         'https://www.youtube.com/s/player/5604538d/player_ias.vflset/en_US/base.js',
         '7X-he4jjvMx7BCX', 'sViSydX8IHtdWA',
     ),
+    (
+        'https://www.youtube.com/s/player/20dfca59/player_ias.vflset/en_US/base.js',
+        '-fLCxedkAk4LUTK2', 'O8kfRq1y1eyHGw',
+    ),
+    (
+        'https://www.youtube.com/s/player/b12cc44b/player_ias.vflset/en_US/base.js',
+        'keLa5R2U00sR9SQK', 'N1OGyujjEwMnLw',
+    ),
 ]
 
 

youtube_dl/extractor/tenplay.py

@@ -3,68 +3,179 @@ from __future__ import unicode_literals
 
 from .common import InfoExtractor
 from ..utils import (
-    HEADRequest,
     parse_age_limit,
-    parse_iso8601,
+    unescapeHTML,
-    # smuggle_url,
+    str_to_int,
+    urljoin
 )
 
+from datetime import datetime
+import base64
+import json
+
 
 class TenPlayIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?10play\.com\.au/(?:[^/]+/)+(?P<id>tpv\d{6}[a-z]{5})'
+    _VALID_URL = r'''(?x)^
+        https?://(?:www\.)?10play\.com\.au/(?:
+            (?:[^/]+/)+(?P<id>tpv\d{6}[a-z]{5})| (?# Individual show id)
+            (?P<show>[^/]+)/episodes/(?P<season>[^/]+)(?# Entire season playlist)
+        )'''
     _TESTS = [{
-        'url': 'https://10play.com.au/masterchef/episodes/season-1/masterchef-s1-ep-1/tpv190718kwzga',
+        'url': 'https://10play.com.au/masterchef/episodes/season-1/episode-1/tpv220408msjpb',
         'info_dict': {
-            'id': '6060533435001',
+            'id': 'tpv220408msjpb',
             'ext': 'mp4',
             'title': 'MasterChef - S1 Ep. 1',
             'description': 'md5:4fe7b78e28af8f2d900cd20d900ef95c',
             'age_limit': 10,
             'timestamp': 1240828200,
             'upload_date': '20090427',
-            'uploader_id': '2199827728001',
         },
         'params': {
-            # 'format': 'bestvideo',
             'skip_download': True,
+            'usenetrc': True,
         }
     }, {
         'url': 'https://10play.com.au/how-to-stay-married/web-extras/season-1/terrys-talks-ep-1-embracing-change/tpv190915ylupc',
         'only_matching': True,
+    }, {
+        'info_dict': {
+            'title': 'Season 2022'
+        },
+        'url': 'https://10play.com.au/the-bold-and-the-beautiful-fast-tracked/episodes/season-2022',
+        'playlist_count': 256,
+        'params': {
+            'skip_download': True,
+            'usenetrc': True,
+        }
     }]
-    # BRIGHTCOVE_URL_TEMPLATE = 'https://players.brightcove.net/2199827728001/cN6vRtRQt_default/index.html?videoId=%s'
+
+    _NETRC_MACHINE = '10play.com.au'
     _GEO_BYPASS = False
-    _FASTLY_URL_TEMPL = 'https://10-selector.global.ssl.fastly.net/s/kYEXFC/media/%s?mbr=true&manifest=m3u&format=redirect'
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+        self._access_token = None
+
+    def get_access_token(self, content_id):
+        if self._access_token is None:
+            # log in with username and password
+            username, password = self._get_login_info()
+
+            if username is None or password is None:
+                self.raise_login_required()
+
+            ten_auth_header = base64.b64encode(datetime.utcnow().strftime("%Y%m%d%H%M%S").encode())
+
+            auth_request_data = json.dumps({"email": username, "password": password}).encode()
+            token_data = self._download_json(
+                'https://10play.com.au/api/user/auth', content_id,
+                note='Logging in to 10play',
+                data=auth_request_data,
+                headers={
+                    'Content-Type': 'application/json;charset=utf-8',
+                    'X-Network-Ten-Auth': ten_auth_header
+                })
+
+            self._access_token = token_data['jwt']['accessToken']
+
+        return self._access_token
+
+    def extract_playlist(self, url):
+        matches = self._VALID_URL_RE.match(url)
+        show = matches.group('show')
+        season = matches.group('season')
+
+        # The api/v1 endpoint is throwing up 403 Forbidden, so we need to use the old API
+        season_info = self._download_json(
+            f'https://10play.com.au/api/shows/{show}/episodes/{season}', f"{show}/{season}",
+            note='Fetching playlist info')
+
+        # Try to find a carousel with the title "episodes", otherwise default to the top one
+        episodes_carousel = next((c for c in season_info['content'][0]['components'] if c['title'].lower() == 'episodes'),
+                                 season_info['content'][0]['components'][0])
+
+        episodes = episodes_carousel['slides']
+
+        load_more_url = urljoin(url, episodes_carousel['loadMoreUrl'])
+
+        while episodes_carousel['hasMore']:
+            skip_ids = [ep['id'] for ep in episodes]
+
+            episodes_carousel = self._download_json(
+                load_more_url, f"{show}/{season}",
+                note=f'Fetching episodes {len(skip_ids)}+',
+                query={'skipIds[]': skip_ids})
+
+            episodes += episodes_carousel['items']
+
+        episodes_urls = [urljoin(url, ep['cardLink']) for ep in episodes]
+
+        return self.playlist_from_matches(episodes_urls, playlist_title=season_info['content'][0].get('title'))
+
+    # Altered version to check for geoblocking without extraneous HEAD request
+    def _extract_m3u8_formats(self, m3u8_url, video_id, ext=None,
+                              entry_protocol='m3u8', preference=None,
+                              m3u8_id=None, note=None, errnote=None,
+                              fatal=True, live=False, data=None, headers={},
+                              query={}):
+        res = self._download_webpage_handle(
+            m3u8_url, video_id,
+            note=note or 'Downloading m3u8 information',
+            errnote=errnote or 'Failed to download m3u8 information',
+            fatal=fatal, data=data, headers=headers, query=query)
+
+        if res is False:
+            return []
+
+        m3u8_doc, urlh = res
+        m3u8_url = urlh.geturl()
+
+        if '10play-not-in-oz' in m3u8_url:
+            self.raise_geo_restricted(countries=['AU'])
+
+        return self._parse_m3u8_formats(
+            m3u8_doc, m3u8_url, ext=ext, entry_protocol=entry_protocol,
+            preference=preference, m3u8_id=m3u8_id, live=live)
 
     def _real_extract(self, url):
         content_id = self._match_id(url)
-        data = self._download_json(
+
-            'https://10play.com.au/api/video/' + content_id, content_id)
+        if content_id is None or content_id == 'None':
-        video = data.get('video') or {}
+            return self.extract_playlist(url)
-        metadata = data.get('metaData') or {}
+
-        brightcove_id = video.get('videoId') or metadata['showContentVideoId']
+        video_info = self._download_json(
-        # brightcove_url = smuggle_url(
+            'https://10play.com.au/api/v1/videos/' + content_id, content_id, note='Fetching video info')
-        #     self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
+
-        #     {'geo_countries': ['AU']})
+        playback_url = video_info['playbackApiEndpoint']
-        m3u8_url = self._request_webpage(HEADRequest(
+
-            self._FASTLY_URL_TEMPL % brightcove_id), brightcove_id).geturl()
+        headers = {}
-        if '10play-not-in-oz' in m3u8_url:
+
-            self.raise_geo_restricted(countries=['AU'])
+        # Handle member-gated videos
-        formats = self._extract_m3u8_formats(m3u8_url, brightcove_id, 'mp4')
+        if video_info.get('memberGated'):
+            access_token = self.get_access_token(content_id)
+            headers.update(Authorization=f"Bearer {access_token}")
+
+        playback_info = self._download_json(
+            playback_url, content_id,
+            note='Fetching playback info',
+            headers=headers)
+
+        formats = self._extract_m3u8_formats(playback_info['source'], content_id, 'mp4')
         self._sort_formats(formats)
 
         return {
-            # '_type': 'url_transparent',
-            # 'url': brightcove_url,
             'formats': formats,
-            'id': brightcove_id,
+            'id': content_id,
-            'title': video.get('title') or metadata.get('pageContentName') or metadata['showContentName'],
+            'title': unescapeHTML(video_info.get('subtitle')) or unescapeHTML(video_info.get('title')),
-            'description': video.get('description'),
+            'description': unescapeHTML(video_info.get('description')),
-            'age_limit': parse_age_limit(video.get('showRatingClassification') or metadata.get('showProgramClassification')),
+            'age_limit': parse_age_limit(video_info.get('classification')),
-            'series': metadata.get('showName'),
+            'series': video_info.get('tvShow'),
-            'season': metadata.get('showContentSeason'),
+            'season': video_info.get('season'),
-            'timestamp': parse_iso8601(metadata.get('contentPublishDate') or metadata.get('pageContentPublishDate')),
+            'season_number': str_to_int(video_info.get('season')),
-            'thumbnail': video.get('poster'),
+            'episode': unescapeHTML(video_info.get('title')),
-            'uploader_id': '2199827728001',
+            'episode_number': str_to_int(video_info.get('episode')),
-            # 'ie_key': 'BrightcoveNew',
+            'timestamp': video_info.get('published'),
+            'duration': video_info.get('duration'),
+            'thumbnail': video_info.get('posterImageUrl'),
         }

youtube_dl/extractor/youtube.py

@@ -1659,17 +1659,46 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
     def _extract_n_function_name(self, jscode):
         func_name, idx = self._search_regex(
             # new: (b=String.fromCharCode(110),c=a.get(b))&&c=nfunc[idx](c)
-            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)s
+            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)
-            # old: .get("n"))&&(b=nfunc[idx](b)
+            # or:  (PL(a),b=a.j.n||null)&&(b=nfunc[idx](b)
-            # older: .get("n"))&&(b=nfunc(b)
+            # or:  (b="nn"[+a.D],vL(a),c=a.j[b]||null)&&(c=narray[idx](c),a.set(b,c),narray.length||nfunc("")
+            # old: (b=a.get("n"))&&(b=nfunc[idx](b)(?P<c>[a-z])\s*=\s*[a-z]\s*
+            # older: (b=a.get("n"))&&(b=nfunc(b)
             r'''(?x)
-                (?:\(\s*(?P<b>[a-z])\s*=\s*(?:
+                \((?:[\w$()\s]+,)*?\s*      # (
-                    String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
+                (?P<b>[a-z])\s*=\s*         # b=
-                    "n+"\[\s*\+?s*[\w$.]+\s*]
+                (?:
-                )\s*,(?P<c>[a-z])\s*=\s*[a-z]\s*)?
+                    (?:                     # expect ,c=a.get(b) (etc)
-                \.\s*get\s*\(\s*(?(b)(?P=b)|"n{1,2}")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\s*
+                        String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
-                (?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
+                        "n+"\[\s*\+?s*[\w$.]+\s*]
-            ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'))
+                    )\s*(?:,[\w$()\s]+(?=,))*|
+                       (?P<old>[\w$]+)      # a (old[er])
+                   )\s*
+                   (?(old)
+                                            # b.get("n")
+                       (?:\.\s*[\w$]+\s*|\[\s*[\w$]+\s*]\s*)*?
+                       (?:\.\s*n|\[\s*"n"\s*]|\.\s*get\s*\(\s*"n"\s*\))
+                       |                    # ,c=a.get(b)
+                       ,\s*(?P<c>[a-z])\s*=\s*[a-z]\s*
+                       (?:\.\s*[\w$]+\s*|\[\s*[\w$]+\s*]\s*)*?
+                       (?:\[\s*(?P=b)\s*]|\.\s*get\s*\(\s*(?P=b)\s*\))
+                   )
+                                            # interstitial junk
+                   \s*(?:\|\|\s*null\s*)?(?:\)\s*)?&&\s*(?:\(\s*)?
+               (?(c)(?P=c)|(?P=b))\s*=\s*   # [c|b]=
+                                            # nfunc|nfunc[idx]
+                   (?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
+            ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'),
+            default=(None, None))
+        # thx bashonly: yt-dlp/yt-dlp/pull/10611
+        if not func_name:
+            self.report_warning('Falling back to generic n function search')
+            return self._search_regex(
+                r'''(?xs)
+                    (?:(?<=[^\w$])|^)       # instead of \b, which ignores $
+                    (?P<name>(?!\d)[a-zA-Z\d_$]+)\s*=\s*function\((?!\d)[a-zA-Z\d_$]+\)
+                    \s*\{(?:(?!};).)+?["']enhanced_except_
+                ''', jscode, 'Initial JS player n function name', group='name')
         if not idx:
             return func_name
 

youtube_dl/jsinterp.py

@@ -925,9 +925,16 @@ class JSInterpreter(object):
                     obj.reverse()
                     return obj
                 elif member == 'slice':
-                    assertion(isinstance(obj, list), 'must be applied on a list')
+                    assertion(isinstance(obj, (list, compat_str)), 'must be applied on a list or string')
-                    assertion(len(argvals) == 1, 'takes exactly one argument')
+                    # From [1]:
-                    return obj[argvals[0]:]
+                    # .slice() - like [:]
+                    # .slice(n) - like [n:] (not [slice(n)]
+                    # .slice(m, n) - like [m:n] or [slice(m, n)]
+                    # [1] https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/slice
+                    assertion(len(argvals) <= 2, 'takes between 0 and 2 arguments')
+                    if len(argvals) < 2:
+                        argvals += (None,)
+                    return obj[slice(*argvals)]
                 elif member == 'splice':
                     assertion(isinstance(obj, list), 'must be applied on a list')
                     assertion(argvals, 'takes one or more arguments')