]> gitweb @ CieloNegro.org - youtube-dl.git/blobdiff - youtube_dl/extractor/youtube.py
Merge branch 'cracked' of https://github.com/hassaanaliw/youtube-dl
[youtube-dl.git] / youtube_dl / extractor / youtube.py
index f420b81482a6b0684cc44b42010b02769d869cc4..5449df8e0d6292a8fe45b504e8eea110b8e6a9ca 100644 (file)
@@ -347,8 +347,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
         self.to_screen(u'RTMP download detected')
 
     def _extract_signature_function(self, video_id, player_url, slen):
-        id_m = re.match(r'.*-(?P<id>[a-zA-Z0-9_-]+)\.(?P<ext>[a-z]+)$',
-                        player_url)
+        id_m = re.match(
+            r'.*-(?P<id>[a-zA-Z0-9_-]+)(?:/watch_as3)?\.(?P<ext>[a-z]+)$',
+            player_url)
         player_type = id_m.group('ext')
         player_id = id_m.group('id')
 
@@ -865,71 +866,26 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
     def _decrypt_signature(self, s, video_id, player_url, age_gate=False):
         """Turn the encrypted s field into a working signature"""
 
-        if player_url is not None:
-            if player_url.startswith(u'//'):
-                player_url = u'https:' + player_url
-            try:
-                player_id = (player_url, len(s))
-                if player_id not in self._player_cache:
-                    func = self._extract_signature_function(
-                        video_id, player_url, len(s)
-                    )
-                    self._player_cache[player_id] = func
-                func = self._player_cache[player_id]
-                if self._downloader.params.get('youtube_print_sig_code'):
-                    self._print_sig_code(func, len(s))
-                return func(s)
-            except Exception:
-                tb = traceback.format_exc()
-                self._downloader.report_warning(
-                    u'Automatic signature extraction failed: ' + tb)
-
-            self._downloader.report_warning(
-                u'Warning: Falling back to static signature algorithm')
-
-        return self._static_decrypt_signature(
-            s, video_id, player_url, age_gate)
-
-    def _static_decrypt_signature(self, s, video_id, player_url, age_gate):
-        if age_gate:
-            # The videos with age protection use another player, so the
-            # algorithms can be different.
-            if len(s) == 86:
-                return s[2:63] + s[82] + s[64:82] + s[63]
-
-        if len(s) == 93:
-            return s[86:29:-1] + s[88] + s[28:5:-1]
-        elif len(s) == 92:
-            return s[25] + s[3:25] + s[0] + s[26:42] + s[79] + s[43:79] + s[91] + s[80:83]
-        elif len(s) == 91:
-            return s[84:27:-1] + s[86] + s[26:5:-1]
-        elif len(s) == 90:
-            return s[25] + s[3:25] + s[2] + s[26:40] + s[77] + s[41:77] + s[89] + s[78:81]
-        elif len(s) == 89:
-            return s[84:78:-1] + s[87] + s[77:60:-1] + s[0] + s[59:3:-1]
-        elif len(s) == 88:
-            return s[7:28] + s[87] + s[29:45] + s[55] + s[46:55] + s[2] + s[56:87] + s[28]
-        elif len(s) == 87:
-            return s[6:27] + s[4] + s[28:39] + s[27] + s[40:59] + s[2] + s[60:]
-        elif len(s) == 86:
-            return s[80:72:-1] + s[16] + s[71:39:-1] + s[72] + s[38:16:-1] + s[82] + s[15::-1]
-        elif len(s) == 85:
-            return s[3:11] + s[0] + s[12:55] + s[84] + s[56:84]
-        elif len(s) == 84:
-            return s[78:70:-1] + s[14] + s[69:37:-1] + s[70] + s[36:14:-1] + s[80] + s[:14][::-1]
-        elif len(s) == 83:
-            return s[80:63:-1] + s[0] + s[62:0:-1] + s[63]
-        elif len(s) == 82:
-            return s[80:37:-1] + s[7] + s[36:7:-1] + s[0] + s[6:0:-1] + s[37]
-        elif len(s) == 81:
-            return s[56] + s[79:56:-1] + s[41] + s[55:41:-1] + s[80] + s[40:34:-1] + s[0] + s[33:29:-1] + s[34] + s[28:9:-1] + s[29] + s[8:0:-1] + s[9]
-        elif len(s) == 80:
-            return s[1:19] + s[0] + s[20:68] + s[19] + s[69:80]
-        elif len(s) == 79:
-            return s[54] + s[77:54:-1] + s[39] + s[53:39:-1] + s[78] + s[38:34:-1] + s[0] + s[33:29:-1] + s[34] + s[28:9:-1] + s[29] + s[8:0:-1] + s[9]
+        if player_url is None:
+            raise ExtractorError(u'Cannot decrypt signature without player_url')
 
-        else:
-            raise ExtractorError(u'Unable to decrypt signature, key length %d not supported; retrying might work' % (len(s)))
+        if player_url.startswith(u'//'):
+            player_url = u'https:' + player_url
+        try:
+            player_id = (player_url, len(s))
+            if player_id not in self._player_cache:
+                func = self._extract_signature_function(
+                    video_id, player_url, len(s)
+                )
+                self._player_cache[player_id] = func
+            func = self._player_cache[player_id]
+            if self._downloader.params.get('youtube_print_sig_code'):
+                self._print_sig_code(func, len(s))
+            return func(s)
+        except Exception as e:
+            tb = traceback.format_exc()
+            raise ExtractorError(
+                u'Automatic signature extraction failed: ' + tb, cause=e)
 
     def _get_available_subtitles(self, video_id, webpage):
         try:
@@ -1265,31 +1221,38 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
                         url += '&signature=' + url_data['sig'][0]
                     elif 's' in url_data:
                         encrypted_sig = url_data['s'][0]
+
+                        if not age_gate:
+                            jsplayer_url_json = self._search_regex(
+                                r'"assets":.+?"js":\s*("[^"]+")',
+                                video_webpage, u'JS player URL')
+                            player_url = json.loads(jsplayer_url_json)
+                        if player_url is None:
+                            player_url_json = self._search_regex(
+                                r'ytplayer\.config.*?"url"\s*:\s*("[^"]+")',
+                                video_webpage, u'age gate player URL')
+                            player_url = json.loads(player_url_json)
+
                         if self._downloader.params.get('verbose'):
-                            if age_gate:
-                                if player_url is None:
-                                    player_version = 'unknown'
-                                else:
+                            if player_url is None:
+                                player_version = 'unknown'
+                                player_desc = 'unknown'
+                            else:
+                                if player_url.endswith('swf'):
                                     player_version = self._search_regex(
                                         r'-(.+)\.swf$', player_url,
                                         u'flash player', fatal=False)
-                                player_desc = 'flash player %s' % player_version
-                            else:
-                                player_version = self._search_regex(
-                                    r'html5player-(.+?)\.js', video_webpage,
-                                    'html5 player', fatal=False)
-                                player_desc = u'html5 player %s' % player_version
+                                    player_desc = 'flash player %s' % player_version
+                                else:
+                                    player_version = self._search_regex(
+                                        r'html5player-(.+?)\.js', video_webpage,
+                                        'html5 player', fatal=False)
+                                    player_desc = u'html5 player %s' % player_version
 
                             parts_sizes = u'.'.join(compat_str(len(part)) for part in encrypted_sig.split('.'))
                             self.to_screen(u'encrypted signature length %d (%s), itag %s, %s' %
                                 (len(encrypted_sig), parts_sizes, url_data['itag'][0], player_desc))
 
-                        if not age_gate:
-                            jsplayer_url_json = self._search_regex(
-                                r'"assets":.+?"js":\s*("[^"]+")',
-                                video_webpage, u'JS player URL')
-                            player_url = json.loads(jsplayer_url_json)
-
                         signature = self._decrypt_signature(
                             encrypted_sig, video_id, player_url, age_gate)
                         url += '&signature=' + signature