Merge 74bbdc079b1dae209faebb8790282d55a6c88e89 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_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/__init__.py

@@ -282,6 +282,7 @@ def _real_main(argv=None):
         postprocessors.append({
             'key': 'FFmpegSubtitlesConvertor',
             'format': opts.convertsubtitles,
+            'ttml_convert_style': opts.ttmlconvertstyle,
         })
     if opts.embedsubtitles:
         postprocessors.append({

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)

youtube_dl/options.py

@@ -870,6 +870,10 @@ def parseOpts(overrideArguments=None):
         '--convert-subs', '--convert-subtitles',
         metavar='FORMAT', dest='convertsubtitles', default=None,
         help='Convert the subtitles to other format (currently supported: srt|ass|vtt|lrc)')
+    postproc.add_option(
+        '--ttml-convert-no-style',
+        action='store_false', dest='ttmlconvertstyle', default=True,
+        help='Discard all style information when converting subtitles from ttml/dfxp')
 
     parser.add_option_group(general)
     parser.add_option_group(network)

youtube_dl/postprocessor/ffmpeg.py

@@ -591,9 +591,10 @@ class FFmpegFixupM3u8PP(FFmpegPostProcessor):
 
 
 class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
-    def __init__(self, downloader=None, format=None):
+    def __init__(self, downloader=None, format=None, ttml_convert_style=True):
         super(FFmpegSubtitlesConvertorPP, self).__init__(downloader)
         self.format = format
+        self.ttml_convert_style = ttml_convert_style
 
     def run(self, info):
         subs = info.get('requested_subtitles')
@@ -626,7 +627,7 @@ class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor):
                 srt_file = subtitles_filename(filename, lang, 'srt', info.get('ext'))
 
                 with open(dfxp_file, 'rb') as f:
-                    srt_data = dfxp2srt(f.read())
+                    srt_data = dfxp2srt(f.read(), self.ttml_convert_style)
 
                 with open(srt_file, 'w', encoding='utf-8') as f:
                     f.write(srt_data)

youtube_dl/utils.py

@@ -4902,7 +4902,7 @@ def srt_subtitles_timecode(seconds):
     return '%02d:%02d:%02d,%03d' % (seconds / 3600, (seconds % 3600) / 60, seconds % 60, (seconds % 1) * 1000)
 
 
-def dfxp2srt(dfxp_data):
+def dfxp2srt(dfxp_data, convert_style=True):
     '''
     @param dfxp_data A bytes-like object containing DFXP data
     @returns A unicode object containing converted SRT data
@@ -4944,7 +4944,7 @@ def dfxp2srt(dfxp_data):
         def start(self, tag, attrib):
             if tag in (_x('ttml:br'), 'br'):
                 self._out += '\n'
-            else:
+            elif convert_style:
                 unclosed_elements = []
                 style = {}
                 element_style_id = attrib.get('style')
@@ -4987,7 +4987,7 @@ def dfxp2srt(dfxp_data):
                 self._unclosed_elements.append(unclosed_elements)
 
         def end(self, tag):
-            if tag not in (_x('ttml:br'), 'br'):
+            if convert_style and tag not in (_x('ttml:br'), 'br'):
                 unclosed_elements = self._unclosed_elements.pop()
                 for element in reversed(unclosed_elements):
                     self._out += '</%s>' % element