Merge 67d578b076 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/extractors.py

@@ -1172,6 +1172,7 @@ from .soundgasm import (
 from .southpark import (
     SouthParkIE,
     SouthParkDeIE,
+    SouthParkDeEnIE,
     SouthParkDkIE,
     SouthParkEsIE,
     SouthParkNlIE

youtube_dl/extractor/mtv.py

@@ -140,6 +140,9 @@ class MTVServicesInfoExtractor(InfoExtractor):
             mediagen_url += '&' if '?' in mediagen_url else '?'
             mediagen_url += 'acceptMethods='
             mediagen_url += 'hls' if use_hls else 'fms'
+        if (self._LANG):
+            mediagen_url += '&' if '?' in mediagen_url else '?'
+            mediagen_url += 'lang=' + self._LANG
 
         mediagen_doc = self._download_xml(
             mediagen_url, video_id, 'Downloading video urls', fatal=False)

youtube_dl/extractor/southpark.py

@@ -54,42 +54,77 @@ class SouthParkEsIE(SouthParkIE):
     }]
 
 
-class SouthParkDeIE(SouthParkIE):
+class SouthParkDeIE(MTVServicesInfoExtractor):
     IE_NAME = 'southpark.de'
-    _VALID_URL = r'https?://(?:www\.)?(?P<url>southpark\.de/(?:clips|alle-episoden|collections)/(?P<id>.+?)(\?|#|$))'
+    _VALID_URL = r'https?://(?:www\.)?(?P<url>southpark\.de/(?:videoclip|folgen|collections)/(?P<id>.+?)(\?|#|$))'
-    _FEED_URL = 'http://www.southpark.de/feeds/video-player/mrss/'
+    _FEED_URL = 'http://feeds.mtvnservices.com/od/feed/intl-mrss-player-feed'
+    _LANG = 'de'
 
     _TESTS = [{
-        'url': 'http://www.southpark.de/clips/uygssh/the-government-wont-respect-my-privacy#tab=featured',
+        # clip
+        'url': 'https://www.southpark.de/videoclip/ct46op/south-park-zahnfee-cartman',
         'info_dict': {
-            'id': '85487c96-b3b9-4e39-9127-ad88583d9bf2',
+            'id': 'e99d45ea-ed00-11e0-aca6-0026b9414f30',
             'ext': 'mp4',
-            'title': 'South Park|The Government Won\'t Respect My Privacy',
+            'title': 'Zahnfee Cartman',
-            'description': 'Cartman explains the benefits of "Shitter" to Stan, Kyle and Craig.',
+            'description': 'Cartman verkleidet sich als Zahnfee, um Butters unter dem Kissen liegenden Zahn zu stehlen. Cartman bekommt 4$ für diesen Zahn, was das Streben nach mehr Zähnen nährt'
-            'timestamp': 1380160800,
-            'upload_date': '20130926',
         },
     }, {
-        # non-ASCII characters in initial URL
+        # episode
-        'url': 'http://www.southpark.de/alle-episoden/s18e09-hashtag-aufwärmen',
+        'url': 'https://www.southpark.de/folgen/242csn/south-park-her-mit-dem-hirn-staffel-1-ep-7',
         'info_dict': {
-            'title': 'Hashtag „Aufwärmen“',
+            'id': '607115f3-496f-40c3-8647-2b0bcff486c0',
-            'description': 'Kyle will mit seinem kleinen Bruder Ike Videospiele spielen. Als der nicht mehr mit ihm spielen will, hat Kyle Angst, dass er die Kids von heute nicht mehr versteht.',
+            'ext': 'mp4',
+            'title': 'South Park | Pink Eye | E 0107 | HDSS0107X deu | Version: 634312 | Comedy Central S1',
         },
-        'playlist_count': 3,
-    }, {
-        # non-ASCII characters in redirect URL
-        'url': 'http://www.southpark.de/alle-episoden/s18e09',
-        'info_dict': {
-            'title': 'Hashtag „Aufwärmen“',
-            'description': 'Kyle will mit seinem kleinen Bruder Ike Videospiele spielen. Als der nicht mehr mit ihm spielen will, hat Kyle Angst, dass er die Kids von heute nicht mehr versteht.',
-        },
-        'playlist_count': 3,
-    }, {
-        'url': 'http://www.southpark.de/collections/2476/superhero-showdown/1',
-        'only_matching': True,
     }]
 
+    def _get_feed_query(self, uri):
+        return {
+            'accountOverride': 'intl.mtvi.com',
+            'arcEp': 'shared.southpark.gsa.de',
+            'ep': '50c78199',
+            'imageEp': 'shared.southpark.gsa.de',
+            'clusterName': 'EMEAA',
+            'mgid': uri,
+        }
+
+
+class SouthParkDeEnIE(SouthParkIE):
+    IE_NAME = 'southpark.de:en'
+    _VALID_URL = r'https?://(?:www\.)?(?P<url>southpark\.de/en/(?:video-clips|episodes|collections)/(?P<id>.+?)(\?|#|$))'
+    _FEED_URL = 'http://feeds.mtvnservices.com/od/feed/intl-mrss-player-feed'
+
+    _TESTS = [{
+        # clip
+        'url': 'https://www.southpark.de/en/video-clips/ct46op/south-park-tooth-fairy-cartman',
+        'info_dict': {
+            'id': 'e99d45ea-ed00-11e0-aca6-0026b9414f30',
+            'ext': 'mp4',
+            'title': 'Tooth Fairy Cartman',
+            'description': 'Cartman dresses up as the Tooth Fairy to steal Butters\' tooth from underneath his pillow. Cartman gets $4 for this tooth, feeding the drive for more teeth.',
+        },
+    }, {
+        # episode
+        'url': 'https://www.southpark.de/en/episodes/yy0vjs/south-park-the-pandemic-special-season-24-ep-1',
+        'info_dict': {
+            'id': 'f5fbd823-04bc-11eb-9b1b-0e40cf2fc285',
+            'ext': 'mp4',
+            'title': 'South Park',
+            'description': 'Randy comes to terms with his role in the COVID-19 outbreak as the on-going pandemic presents continued challenges to the citizens of South Park.',
+        },
+    }]
+
+    def _get_feed_query(self, uri):
+        return {
+            'accountOverride': 'intl.mtvi.com',
+            'arcEp': 'shared.southpark.gsa.en',
+            'ep': '20476225',
+            'imageEp': 'shared.southpark.gsa.en',
+            'clusterName': 'EMEAA',
+            'mgid': uri,
+        }
+
 
 class SouthParkNlIE(SouthParkIE):
     IE_NAME = 'southpark.nl'

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*      # (
+                (?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$.]+\s*]
-                )\s*,(?P<c>[a-z])\s*=\s*[a-z]\s*)?
+                    )\s*(?:,[\w$()\s]+(?=,))*|
-                \.\s*get\s*\(\s*(?(b)(?P=b)|"n{1,2}")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\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'))
+            ''', 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')