Merge 206796d6b3 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

@@ -669,6 +669,10 @@ from .markiza import (
     MarkizaPageIE,
 )
 from .massengeschmacktv import MassengeschmackTVIE
+from .masterclass import (
+    MasterClassIE,
+    MasterClassCourseIE,
+)
 from .matchtv import MatchTVIE
 from .mdr import MDRIE
 from .medaltv import MedalTVIE

youtube_dl/extractor/masterclass.py

@@ -0,0 +1,137 @@
+# coding: utf-8
+from __future__ import unicode_literals
+import re
+
+from .brightcove import BrightcoveNewIE
+from .common import InfoExtractor
+from ..utils import (
+    int_or_none,
+    url_or_none,
+    parse_duration,
+    unified_strdate
+)
+
+
+class MasterClassBaseIE(InfoExtractor):
+    BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/default_default/index.html?videoId=%s'
+    BRIGHTCOVE_ACCOUNT_ID = '5344802162001'
+
+    def _call_api(self, course_id, csrf, video_id):
+        headers = {"Content-Type": "application/vnd.api+json",
+                   "X-CSRF-Token": csrf,
+                   "Accept": "application/vnd.api+json"}
+
+        return self._download_json('https://www.masterclass.com/jsonapi/v1/courses/%s?include=instructors%%2Cchapters.annotations' % course_id,
+                                   video_id, headers=headers, fatal=True)
+
+
+class MasterClassIE(MasterClassBaseIE):
+    IE_NAME = 'masterclass'
+    _VALID_URL = r'https?://(?:www\.)?masterclass\.com/classes/(?P<course>[a-z0-9\-]+)/chapters/(?P<id>[a-z0-9\-]+)'
+    _TESTS = [{
+        'url': 'https://www.masterclass.com/classes/gordon-ramsay-teaches-cooking-restaurant-recipes-at-home/chapters/pomme-puree',
+        'md5': '5fbe8736e761c2cc61ed0ee2841e8f47',
+        'info_dict': {
+            'id': '5834174464001',
+            'ext': 'mp4',
+            'title': 'GR2_Trailer_v9a_Brightcove_v2',
+            'timestamp': 1536792485,
+            'upload_date': '20180912',
+            'uploader_id': '5344802162001',
+            'duration': 93.44
+        }
+    }]
+
+    def _real_extract(self, url):
+        course_id, video_id = re.match(self._VALID_URL, url).groups()
+        webpage = self._download_webpage(url, video_id)
+
+        match_free = re.search(r'\"embedUrl\":\"(.+?)\"}</script>', webpage)
+        if match_free:  # match free trailer video
+            clip_url = match_free.groups()[0]
+            return self.url_result(clip_url)
+
+        csrf = self._html_search_regex(r'name=\"csrf-token\" content=\"(.+?)(?=\" \/)', webpage, 'csrf', fatal=True)
+        meta = self._call_api(course_id, csrf, video_id)
+
+        chapter = meta.get('data').get('attributes').get('title')
+        release_date = unified_strdate(meta.get('data').get('attributes').get('available_at'))
+        for json_data in meta['included']:
+            json_data = json_data['attributes']
+            if (json_data['slug'] == video_id):
+                brightcove_id = json_data['brightcove_video_id']
+                brightcove_url = self.BRIGHTCOVE_URL_TEMPLATE % (self.BRIGHTCOVE_ACCOUNT_ID, brightcove_id)
+
+                title = json_data['title']
+                chapter_number = int_or_none(json_data.get('number'))
+                duration = int_or_none(parse_duration(json_data.get('duration')))
+                description = json_data.get('abstract')
+                thumbnail = url_or_none(json_data.get('video_thumb_url'))
+
+                return ({
+                    '_type': 'url_transparent',
+                    'url': brightcove_url,
+                    'ie_key': BrightcoveNewIE.ie_key(),
+                    'id': video_id,
+                    'title': title,
+                    'chapter': chapter,
+                    'chapter_number': chapter_number,
+                    'release_date': release_date,
+                    'duration': duration,
+                    'description': description,
+                    'thumbnail': thumbnail
+                })
+
+
+class MasterClassCourseIE(MasterClassBaseIE):
+    IE_NAME = "masterclass:course"
+    _VALID_URL = r'https?://(?:www\.)?masterclass\.com/classes/(?P<id>[^/]+)'
+    _TESTS = [{
+        'url': 'https://www.masterclass.com/classes/alicia-keys-teaches-songwriting-and-producing',
+        'only_matching': True,
+    }, {
+        'url': 'https://www.masterclass.com/classes/matthew-walker-teaches-the-science-of-better-sleep',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        course_id = self._match_id(url)
+        webpage = self._download_webpage(url, course_id, fatal=True)
+
+        csrf = self._html_search_regex(r'name=\"csrf-token\" content=\"(.+?)(?=\" \/)', webpage, 'csrf', fatal=True)
+        meta = self._call_api(course_id, csrf, course_id)
+
+        chapter = meta.get('data').get('attributes').get('title')
+        release_date = unified_strdate(meta.get('data').get('attributes').get('available_at'))
+        entries = []
+        for json_data in meta['included']:
+            if (json_data['type'] == 'chapter'):
+                json_data = json_data['attributes']
+                brightcove_id = json_data['brightcove_video_id']
+                brightcove_url = self.BRIGHTCOVE_URL_TEMPLATE % (self.BRIGHTCOVE_ACCOUNT_ID, brightcove_id)
+
+                title = json_data['title']
+                video_id = json_data.get('slug')
+                chapter_number = int_or_none(json_data.get('number'))
+                duration = int_or_none(parse_duration(json_data.get('duration')))
+                description = json_data.get('abstract')
+                thumbnail = url_or_none(json_data.get('video_thumb_url'))
+
+                entries.append({
+                    '_type': 'url_transparent',
+                    'url': brightcove_url,
+                    'ie_key': BrightcoveNewIE.ie_key(),
+                    'id': video_id,
+                    'title': title,
+                    'chapter': chapter,
+                    'chapter_number': chapter_number,
+                    'release_date': release_date,
+                    'duration': duration,
+                    'description': description,
+                    'thumbnail': thumbnail
+                })
+
+        entries = sorted(entries, key=lambda i: int(i.get("chapter_number")))
+        title = meta['data']['attributes']['title']
+        description = meta.get('data').get('attributes').get('overview')
+        return self.playlist_result(entries, course_id, title, description)

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')