Merge 5888bba9e659795587555e56552e38ec8713211f into da7223d4aa42ff9fc680b0951d043dd03cec2d30

* improve extraction of global "useful data" Array from player JS
* also handle tv-player and add tests: thx seproDev (yt-dlp/yt-dlp#12684)

Co-Authored-By: sepro <sepro@sepr0.com>

test/test_YoutubeDL.py

@@ -411,6 +411,15 @@ class TestFormatSelection(unittest.TestCase):
         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
         self.assertEqual(downloaded_ids, ['248+141'])
 
+        for f in ['248+141', '141+248']:
+            info_dict = _make_result(list(formats_order), extractor='youtube')
+            ydl = YDL({'format': f})
+            yie = YoutubeIE(ydl)
+            yie._sort_formats(info_dict['formats'])
+            ydl.process_ie_result(info_dict)
+            downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
+            self.assertEqual(downloaded_ids, ['248+141'])
+
         for f1, f2 in zip(formats_order, formats_order[1:]):
             info_dict = _make_result([f1, f2], extractor='youtube')
             ydl = YDL({'format': 'best/bestvideo'})

test/test_youtube_signature.py

@@ -232,8 +232,32 @@ _NSIG_TESTS = [
         'W9HJZKktxuYoDTqW', 'jHbbkcaxm54',
     ),
     (
-        'https://www.youtube.com/s/player/91201489/player_ias_tce.vflset/en_US/base.js',
-        'W9HJZKktxuYoDTqW', 'U48vOZHaeYS6vO',
+        'https://www.youtube.com/s/player/643afba4/player_ias.vflset/en_US/base.js',
+        'W9HJZKktxuYoDTqW', 'larxUlagTRAcSw',
+    ),
+    (
+        'https://www.youtube.com/s/player/e7567ecf/player_ias_tce.vflset/en_US/base.js',
+        'Sy4aDGc0VpYRR9ew_', '5UPOT1VhoZxNLQ',
+    ),
+    (
+        'https://www.youtube.com/s/player/d50f54ef/player_ias_tce.vflset/en_US/base.js',
+        'Ha7507LzRmH3Utygtj', 'XFTb2HoeOE5MHg',
+    ),
+    (
+        'https://www.youtube.com/s/player/074a8365/player_ias_tce.vflset/en_US/base.js',
+        'Ha7507LzRmH3Utygtj', 'ufTsrE0IVYrkl8v',
+    ),
+    (
+        'https://www.youtube.com/s/player/643afba4/player_ias.vflset/en_US/base.js',
+        'N5uAlLqm0eg1GyHO', 'dCBQOejdq5s-ww',
+    ),
+    (
+        'https://www.youtube.com/s/player/69f581a5/tv-player-ias.vflset/tv-player-ias.js',
+        '-qIP447rVlTTwaZjY', 'KNcGOksBAvwqQg',
+    ),
+    (
+        'https://www.youtube.com/s/player/643afba4/tv-player-ias.vflset/tv-player-ias.js',
+        'ir9-V6cdbCiyKxhr', '2PL7ZDYAALMfmA',
     ),
 ]
 

youtube_dl/YoutubeDL.py

@@ -1489,39 +1489,48 @@ class YoutubeDL(object):
                             yield matches[-1]
             elif selector.type == MERGE:
                 def _merge(formats_info):
-                    format_1, format_2 = [f['format_id'] for f in formats_info]
+                    format_1, format_2 = formats_info[0], formats_info[1]
+                    format_1_id, format_2_id = format_1['format_id'], format_2['format_id']
                     # The first format must contain the video and the
                     # second the audio
-                    if formats_info[0].get('vcodec') == 'none':
+
+                    # If the user swapped the two inputs, try swapping it for
+                    # them
+                    if format_1.get('acodec') != 'none' and format_2.get('vcodec') != 'none':
+                        temp = format_1
+                        format_1 = format_2
+                        format_2 = temp
+
+                    if format_1.get('vcodec') == 'none':
                         self.report_error('The first format must '
                                           'contain the video, try using '
-                                          '"-f %s+%s"' % (format_2, format_1))
-                        return
+                                          '"-f %s+%s"' % (format_1_id, format_2_id))
+
                     # Formats must be opposite (video+audio)
-                    if formats_info[0].get('acodec') == 'none' and formats_info[1].get('acodec') == 'none':
+                    if format_1.get('acodec') == 'none' and format_2.get('acodec') == 'none':
                         self.report_error(
                             'Both formats %s and %s are video-only, you must specify "-f video+audio"'
-                            % (format_1, format_2))
+                            % (format_1_id, format_2_id))
                         return
                     output_ext = (
-                        formats_info[0]['ext']
+                        format_1['ext']
                         if self.params.get('merge_output_format') is None
                         else self.params['merge_output_format'])
                     return {
                         'requested_formats': formats_info,
-                        'format': '%s+%s' % (formats_info[0].get('format'),
-                                             formats_info[1].get('format')),
-                        'format_id': '%s+%s' % (formats_info[0].get('format_id'),
-                                                formats_info[1].get('format_id')),
-                        'width': formats_info[0].get('width'),
-                        'height': formats_info[0].get('height'),
-                        'resolution': formats_info[0].get('resolution'),
-                        'fps': formats_info[0].get('fps'),
-                        'vcodec': formats_info[0].get('vcodec'),
-                        'vbr': formats_info[0].get('vbr'),
-                        'stretched_ratio': formats_info[0].get('stretched_ratio'),
-                        'acodec': formats_info[1].get('acodec'),
-                        'abr': formats_info[1].get('abr'),
+                        'format': '%s+%s' % (format_1.get('format'),
+                                             format_2.get('format')),
+                        'format_id': '%s+%s' % (format_1.get('format_id'),
+                                                format_2.get('format_id')),
+                        'width': format_1.get('width'),
+                        'height': format_1.get('height'),
+                        'resolution': format_1.get('resolution'),
+                        'fps': format_1.get('fps'),
+                        'vcodec': format_1.get('vcodec'),
+                        'vbr': format_1.get('vbr'),
+                        'stretched_ratio': format_1.get('stretched_ratio'),
+                        'acodec': format_2.get('acodec'),
+                        'abr': format_2.get('abr'),
                         'ext': output_ext,
                     }
 

youtube_dl/extractor/youtube.py

@@ -91,12 +91,12 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
             'INNERTUBE_CONTEXT': {
                 'client': {
                     'clientName': 'IOS',
-                    'clientVersion': '19.45.4',
+                    'clientVersion': '20.10.4',
                     'deviceMake': 'Apple',
                     'deviceModel': 'iPhone16,2',
-                    'userAgent': 'com.google.ios.youtube/19.45.4 (iPhone16,2; U; CPU iOS 18_1_0 like Mac OS X;)',
+                    'userAgent': 'com.google.ios.youtube/20.10.4 (iPhone16,2; U; CPU iOS 18_3_2 like Mac OS X;)',
                     'osName': 'iPhone',
-                    'osVersion': '18.1.0.22B83',
+                    'osVersion': '18.3.2.22D82',
                 },
             },
             'INNERTUBE_CONTEXT_CLIENT_NAME': 5,
@@ -109,7 +109,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
             'INNERTUBE_CONTEXT': {
                 'client': {
                     'clientName': 'MWEB',
-                    'clientVersion': '2.20241202.07.00',
+                    'clientVersion': '2.20250311.03.00',
                     # mweb previously did not require PO Token with this UA
                     'userAgent': 'Mozilla/5.0 (iPad; CPU OS 16_7_10 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.6 Mobile/15E148 Safari/604.1,gzip(gfe)',
                 },
@@ -122,7 +122,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
             'INNERTUBE_CONTEXT': {
                 'client': {
                     'clientName': 'TVHTML5',
-                    'clientVersion': '7.20250120.19.00',
+                    'clientVersion': '7.20250312.16.00',
                     'userAgent': 'Mozilla/5.0 (ChromiumStylePlatform) Cobalt/Version',
                 },
             },
@@ -133,7 +133,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
             'INNERTUBE_CONTEXT': {
                 'client': {
                     'clientName': 'WEB',
-                    'clientVersion': '2.20241126.01.00',
+                    'clientVersion': '2.20250312.04.00',
                 },
             },
             'INNERTUBE_CONTEXT_CLIENT_NAME': 1,
@@ -692,7 +692,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         'invidious': '|'.join(_INVIDIOUS_SITES),
     }
     _PLAYER_INFO_RE = (
-        r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})/player',
+        r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})//(?:tv-)?player',
         r'/(?P<id>[a-zA-Z0-9_-]{8,})/player(?:_ias\.vflset(?:/[a-zA-Z]{2,3}_[a-zA-Z]{2,3})?|-plasma-ias-(?:phone|tablet)-[a-z]{2}_[A-Z]{2}\.vflset)/base\.js$',
         r'\b(?P<id>vfl[a-zA-Z0-9_-]+)\b.*?\.js$',
     )
@@ -1857,7 +1857,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
     def _extract_n_function_code_jsi(self, video_id, jsi, player_id=None):
 
         var_ay = self._search_regex(
-            r'(?:[;\s]|^)\s*(var\s*[\w$]+\s*=\s*"[^"]+"\s*\.\s*split\("\{"\))(?=\s*[,;])',
+            r'(?:[;\s]|^)\s*(var\s*[\w$]+\s*=\s*"(?:\\"|[^"])+"\s*\.\s*split\("\W+"\))(?=\s*[,;])',
             jsi.code, 'useful values', default='')
 
         func_name = self._extract_n_function_name(jsi.code)