Merge 5abcd988a2 into c5098961b0

[Youtube] Rework n function extraction pattern
Now also succeeds with player b12cc44b
2024-12-22 20:20:09 +09:00 · 2024-08-21 22:33:01 -04:00 · 2024-08-06 20:59:09 +01:00 · 2024-08-06 20:51:38 +01:00 · 2024-08-01 19:18:34 +01:00 · 2022-07-28 14:57:33 +02:00
6 changed files with 261 additions and 13 deletions
--- a/test/test_jsinterp.py
+++ b/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()
--- a/test/test_youtube_signature.py
+++ b/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',
+    ),
 ]


--- a/youtube_dl/extractor/cozytv.py
+++ b/youtube_dl/extractor/cozytv.py
@ -0,0 +1,172 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+import re
+
+from .common import InfoExtractor
+from ..utils import (
+    float_or_none,
+    int_or_none,
+    unified_timestamp,
+    ExtractorError,
+)
+
+
+class CozyTVBaseIE(InfoExtractor):
+    _API_BASE = 'https://api.cozy.tv'
+
+    def _join_url(self, *parts):
+        return '/'.join(parts).rstrip('/')
+
+    def _user_status(self, user, video_id):
+        url = self._join_url(self._API_BASE, 'cache', user, 'status')
+        return self._download_json(url, video_id)
+
+    def _extend_formats(self, formats, cdn_url, video_id, live=False):
+        formats.extend(self._extract_m3u8_formats(
+            cdn_url + '/index.m3u8',
+            video_id,
+            ext='mp4',
+            m3u8_id='hls',
+            preference=-1,
+            live=live))
+        formats.extend(self._extract_m3u8_formats(
+            cdn_url + '/index-ts-audio.m3u8',
+            video_id,
+            ext='mp4',
+            m3u8_id='hls-audio',
+            preference=-2,
+            fatal=False,
+            live=live))
+        self._sort_formats(formats)
+
+
+class CozyTVStreamIE(CozyTVBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?cozy\.tv/(?P<user>[^/?#]+)'
+    _TEST = {
+        'url': 'https://cozy.tv/alexjones',
+        'info_dict': {
+            'id': 'alexjones',
+            'title': 'The Alex Jones Show - Infowars.com',
+            'ext': 'mp4',
+            'uploader': 'AlexJones',
+            'timestamp': 1658422904,
+            'upload_date': '20220721',
+            'uploader_id': 'alexjones',
+            'uploader_url': 'https://cozy.tv/alexjones',
+            'view_count': int,
+            'is_live': True,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }
+
+    def _real_extract(self, url):
+        video_id, user = re.match(self._VALID_URL, url).group('user', 'user')
+        user_status = self._user_status(user, video_id)
+        is_live = user_status['isLive']
+
+        if is_live is None:
+            raise ExtractorError('%s is offline' % user, expected=True)
+
+        cdn_url = self._join_url(user_status['livecdns'][0], 'live', user)
+
+        info = {
+            'id': video_id,
+            'title': user_status['title'],
+        }
+
+        timestamp = unified_timestamp(is_live)
+        formats = []
+        self._extend_formats(formats, cdn_url, video_id, live=True)
+
+        info.update({
+            'display_id': video_id,
+            'formats': formats,
+            'timestamp': timestamp,
+            'uploader': user_status.get('displayName') or user,
+            'uploader_id': user,
+            'uploader_url': 'https://cozy.tv/' + user,
+            'view_count': int_or_none(user_status.get('viewers')),
+            'is_live': True,
+        })
+
+        return info
+
+
+class CozyTVReplayIE(CozyTVBaseIE):
+    _VALID_URL = r'https?://(?:www\.)?cozy\.tv/(?P<user>[^/]+)/replays/(?P<id>\d{4}-(?:0[1-9]|1[0-2])-(?:0[1-9]|[12][0-9]|3[01])(?:_\d+)?)'
+    _TESTS = [
+        {
+            'url': 'https://cozy.tv/althype/replays/2022-05-24_2',
+            'info_dict': {
+                'id': '628d070e3f2f1a99756b6c2f',
+                'title': 'Secret Stream: Making Video',
+                'ext': 'mp4',
+                'display_id': '2022-05-24_2',
+                'thumbnail': r're:^https?://.*/replays/althype/2022-05-24_2/thumb\.webp$',
+                'uploader': 'AltHype',
+                'timestamp': 1653409506,
+                'upload_date': '20220524',
+                'uploader_id': 'althype',
+                'uploader_url': 'https://cozy.tv/althype',
+                'duration': 9303.0,
+                'view_count': 217,
+            },
+            'params': {
+                'skip_download': True,
+            },
+        },
+        {
+            'url': 'https://cozy.tv/nick/replays/2022-07-28',
+            'info_dict': {
+                'id': '62e23d1c4a22de6fb37b83f5',
+                'title': 'GAY PANDEMIC: Biden Admin To Declare MONKEYPOX EMERGENCY | America First Ep. 1037',
+                'ext': 'mp4',
+                'display_id': '2022-07-28',
+                'thumbnail': r're:^https?://.*/replays/nick/2022-07-28/thumb\.webp$',
+                'uploader': 'Nick',
+                'timestamp': 1658993895,
+                'upload_date': '20220728',
+                'uploader_id': 'nick',
+                'uploader_url': 'https://cozy.tv/nick',
+                'duration': 16410.0,
+                'view_count': 6281,
+            },
+            'params': {
+                'skip_download': True,
+            },
+        },
+    ]
+
+    def _real_extract(self, url):
+        video_id, user = re.match(self._VALID_URL, url).group('id', 'user')
+        user_status = self._user_status(user, video_id)
+        data_url = self._join_url(self._API_BASE, 'cache', user, 'replay', video_id)
+        data = self._download_json(data_url, video_id)
+
+        cdn_url = self._join_url(data['cdns'][0], 'replays', user, video_id)
+
+        info = {
+            'id': data.get('_id') or video_id,
+            'title': data['title'],
+        }
+
+        timestamp = unified_timestamp(data.get('date'))
+        formats = []
+        self._extend_formats(formats, cdn_url, video_id)
+
+        info.update({
+            'display_id': video_id,
+            'duration': float_or_none(data.get('duration')),
+            'formats': formats,
+            'thumbnail': cdn_url + '/thumb.webp',
+            'timestamp': timestamp,
+            'uploader': user_status.get('displayName') or user,
+            'uploader_id': user,
+            'uploader_url': 'https://cozy.tv/' + user,
+            'view_count': int_or_none(data.get('peakViewers')),
+        })
+
+        return info
--- a/youtube_dl/extractor/extractors.py
+++ b/youtube_dl/extractor/extractors.py
@ -248,6 +248,10 @@ from .cnn import (
    CNNArticleIE,
 )
 from .coub import CoubIE
+from .cozytv import (
+    CozyTVReplayIE,
+    CozyTVStreamIE,
+)
 from .comedycentral import (
    ComedyCentralIE,
    ComedyCentralTVIE,
--- a/youtube_dl/extractor/youtube.py
+++ b/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
-            # old: .get("n"))&&(b=nfunc[idx](b)
-            # older: .get("n"))&&(b=nfunc(b)
+            # or:  (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)
+            # or:  (PL(a),b=a.j.n||null)&&(b=nfunc[idx](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*(?:
-                    String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
-                    "n+"\[\s*\+?s*[\w$.]+\s*]
-                )\s*,(?P<c>[a-z])\s*=\s*[a-z]\s*)?
-                \.\s*get\s*\(\s*(?(b)(?P=b)|"n{1,2}")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\s*
-                (?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
-            ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'))
+                \((?:[\w$()\s]+,)*?\s*      # (
+                (?P<b>[a-z])\s*=\s*         # b=
+                (?:
+                    (?:                     # expect ,c=a.get(b) (etc)
+                        String\s*\.\s*fromCharCode\s*\(\s*110\s*\)|
+                        "n+"\[\s*\+?s*[\w$.]+\s*]
+                    )\s*(?:,[\w$()\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

--- a/youtube_dl/jsinterp.py
+++ b/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(len(argvals) == 1, 'takes exactly one argument')
-                    return obj[argvals[0]:]
+                    assertion(isinstance(obj, (list, compat_str)), 'must be applied on a list or string')
+                    # From [1]:
+                    # .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')
Author	SHA1	Message	Date
Valeriy Litkovskyy	30bd1e3352	Merge `5abcd988a2` into `c5098961b0`	2024-08-21 22:33:01 -04:00
dirkf	c5098961b0	[Youtube] Rework n function extraction pattern Now also succeeds with player b12cc44b	2024-08-06 20:59:09 +01:00
dirkf	dbc08fba83	[jsinterp] Improve slice implementation for player b12cc44b Partly taken from yt-dlp/yt-dlp#10664, thx seproDev Fixes #32896	2024-08-06 20:51:38 +01:00
Aiur Adept	71223bff39	[Youtube] Fix nsig extraction for player 20dfca59 (#32891 ) * dirkf's patch for nsig extraction * add generic search per yt-dlp/yt-dlp/pull/10611 - thx bashonly --------- Co-authored-by: dirkf <fieldhouse@gmx.net>	2024-08-01 19:18:34 +01:00
Valeriy Litkovskyy	5abcd988a2	[cozytv] Add new extractor	2022-07-28 14:57:33 +02:00