Merge ce604ae7dc into e1b3fa242c

Fixes #32877

test/test_youtube_signature.py

@@ -166,6 +166,14 @@ _NSIG_TESTS = [
         'https://www.youtube.com/s/player/b22ef6e7/player_ias.vflset/en_US/base.js',
         'b6HcntHGkvBLk_FRf', 'kNPW6A7FyP2l8A',
     ),
+    (
+        'https://www.youtube.com/s/player/3400486c/player_ias.vflset/en_US/base.js',
+        'lL46g3XifCKUZn1Xfw', 'z767lhet6V2Skl',
+    ),
+    (
+        'https://www.youtube.com/s/player/5604538d/player_ias.vflset/en_US/base.js',
+        '7X-he4jjvMx7BCX', 'sViSydX8IHtdWA',
+    ),
 ]
 
 

youtube_dl/downloader/external.py

@@ -393,6 +393,19 @@ class FFmpegFD(ExternalFD):
             # https://github.com/ytdl-org/youtube-dl/issues/11800#issuecomment-275037127
             # http://trac.ffmpeg.org/ticket/6125#comment:10
             args += ['-seekable', '1' if seekable else '0']
+        http_seekable = info_dict.get('_http_seekable')
+        if http_seekable is not None:
+            # setting -http_seekable prevents ffmpeg from guessing if the server
+            # supports seeking in other kinds of requests (by adding the same header
+            # as above: `Range: bytes=0-`)
+            args += ['-http_seekable', '1' if http_seekable else '0']
+        icy = info_dict.get('_icy')
+        if icy is not None:
+            # setting -icy 0 prevents ffmpeg from sending the header `Icy-Metadata: 1`,
+            # which can cause also problems
+            # https://github.com/ytdl-org/youtube-dl/pull/29688
+            # https://trac.ffmpeg.org/ticket/5460#comment:5
+            args += ['-icy', '1' if icy else '0']
 
         args += self._configuration_args()
 

youtube_dl/extractor/nbc.py

@@ -477,43 +477,94 @@ class NBCOlympicsIE(InfoExtractor):
 class NBCOlympicsStreamIE(AdobePassIE):
     IE_NAME = 'nbcolympics:stream'
     _VALID_URL = r'https?://stream\.nbcolympics\.com/(?P<id>[0-9a-z-]+)'
-    _TEST = {
-        'url': 'http://stream.nbcolympics.com/2018-winter-olympics-nbcsn-evening-feb-8',
-        'info_dict': {
-            'id': '203493',
-            'ext': 'mp4',
-            'title': 're:Curling, Alpine, Luge [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
+    _TESTS = [
+        # "Tokenized" .m3u8 source URL
+        {
+            'url': 'https://stream.nbcolympics.com/womens-soccer-group-round-11',
+            'info_dict': {
+                'id': '2019740',
+                'ext': 'mp4',
+                'title': r"re:Women's Group Stage - Netherlands vs\. Brazil [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$",
+            },
+            'params': {
+                # m3u8 download
+                'skip_download': True,
+            },
         },
-        'params': {
-            # m3u8 download
-            'skip_download': True,
+        # Plain .m3u8 source URL
+        {
+            'url': 'https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars',
+            'info_dict': {
+                'id': '2021729',
+                'ext': 'mp4',
+                'title': r're:Event Finals: M Floor, W Vault, M Pommel, W Uneven Bars [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
+            },
+            'params': {
+                # m3u8 download
+                'skip_download': True,
+            },
         },
-    }
+    ]
     _DATA_URL_TEMPLATE = 'http://stream.nbcolympics.com/data/%s_%s.json'
+    _LEAP_URL_TEMPLATE = 'https://api-leap.nbcsports.com/feeds/assets/%s?application=NBCOlympics&platform=%s&format=nbc-player&env=staging'
 
     def _real_extract(self, url):
         display_id = self._match_id(url)
         webpage = self._download_webpage(url, display_id)
         pid = self._search_regex(r'pid\s*=\s*(\d+);', webpage, 'pid')
-        resource = self._search_regex(
-            r"resource\s*=\s*'(.+)';", webpage,
-            'resource').replace("' + pid + '", pid)
+
         event_config = self._download_json(
             self._DATA_URL_TEMPLATE % ('event_config', pid),
-            pid)['eventConfig']
+            pid,
+            'Downloading event config',
+        )['eventConfig']
+        resource = event_config.get('resourceId', 'NBCOlympics')
         title = self._live_title(event_config['eventTitle'])
-        source_url = self._download_json(
-            self._DATA_URL_TEMPLATE % ('live_sources', pid),
-            pid)['videoSources'][0]['sourceUrl']
+
+        leap_config = self._download_json(
+            self._LEAP_URL_TEMPLATE % (pid, 'desktop'),
+            pid,
+            'Downloading leap config',
+        )
+        source_url = leap_config['videoSources'][0]['cdnSources']['primary'][0]['sourceUrl']
+
+        ap_resource = self._get_mvpd_resource(
+            resource,
+            re.sub(r'[^\w\d ]+', '', event_config['eventTitle']),
+            pid,
+            event_config.get('ratingId', 'NO VALUE'),
+        )
         media_token = self._extract_mvpd_auth(
-            url, pid, event_config.get('requestorId', 'NBCOlympics'), resource)
-        formats = self._extract_m3u8_formats(self._download_webpage(
-            'http://sp.auth.adobe.com/tvs/v1/sign', pid, query={
-                'cdn': 'akamai',
-                'mediaToken': base64.b64encode(media_token.encode()),
-                'resource': base64.b64encode(resource.encode()),
-                'url': source_url,
-            }), pid, 'mp4')
+            url, pid, event_config.get('requestorId', 'NBCOlympics'), ap_resource)
+
+        if event_config.get('cdnToken') is True:
+            source_url = self._download_json(
+                'https://tokens.playmakerservices.com/',
+                pid,
+                'Retrieving tokenized URL',
+                data=json.dumps({
+                    'application': 'NBCSports',
+                    'authentication-type': 'adobe-pass',
+                    'cdn': 'akamai',
+                    # Indicates that the player communicates its token not via the path
+                    # but via a cookie? NBC's player specifies `'false'` but field just
+                    # doesn't seem to have an effect.
+                    # 'inPath': 'false',
+                    'pid': pid,
+                    'platform': 'desktop',
+                    'requestorId': 'NBCOlympics',
+                    'resourceId': base64.b64encode(ap_resource.encode()).decode(),
+                    'token': base64.b64encode(media_token.encode()).decode(),
+                    'url': source_url,
+                    'version': 'v1',
+                }).encode(),
+            )['akamai'][0]['tokenizedUrl']
+
+        formats = self._extract_m3u8_formats(source_url, pid, 'mp4')
+        for f in formats:
+            f['_seekable'] = False
+            f['_http_seekable'] = False
+            f['_icy'] = False
         self._sort_formats(formats)
 
         return {

youtube_dl/extractor/youtube.py

@@ -1647,7 +1647,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         except JSInterpreter.Exception as e:
             self.report_warning(
                 '%s (%s %s)' % (
-                    'Unable to decode n-parameter: download likely to be throttled',
+                    'Unable to decode n-parameter: expect download to be blocked or throttled',
                     error_to_compat_str(e),
                     traceback.format_exc()),
                 video_id=video_id)
@@ -1659,18 +1659,22 @@ 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)
             r'''(?x)
-                (?:\(\s*(?P<b>[a-z])\s*=\s*String\s*\.\s*fromCharCode\s*\(\s*110\s*\)\s*,(?P<c>[a-z])\s*=\s*[a-z]\s*)?
-                \.\s*get\s*\(\s*(?(b)(?P=b)|"n")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\s*
+                (?:\(\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'))
         if not idx:
             return func_name
 
         return self._parse_json(self._search_regex(
-            r'var {0}\s*=\s*(\[.+?\])\s*[,;]'.format(re.escape(func_name)), jscode,
+            r'var\s+{0}\s*=\s*(\[.+?\])\s*[,;]'.format(re.escape(func_name)), jscode,
             'Initial JS player n function list ({0}.{1})'.format(func_name, idx)),
             func_name, transform_source=js_to_json)[int(idx)]