Merge 5831f6feef00d58334f144f9cf5d0b65c5502a43 into da7223d4aa42ff9fc680b0951d043dd03cec2d30

[YouTube] Improve support for tce-style player JS
* 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>
2025-07-13 23:14:13 +09:00 · 2025-03-22 07:20:23 +08:00 · 2025-03-21 16:26:25 +00:00 · 2025-03-21 16:13:24 +00:00 · 2022-05-29 00:26:26 -07:00
3 changed files with 45 additions and 13 deletions
--- a/test/test_youtube_signature.py
+++ b/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',
    ),
 ]

--- a/youtube_dl/extractor/youtube.py
+++ b/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)
--- a/youtube_dl/postprocessor/ffmpeg.py
+++ b/youtube_dl/postprocessor/ffmpeg.py
@ -22,6 +22,7 @@ from ..utils import (
    dfxp2srt,
    ISO639Utils,
    replace_extension,
+    compat_str,
 )


@ -300,9 +301,16 @@ class FFmpegExtractAudioPP(FFmpegPostProcessor):
            extension = self._preferredcodec
            more_opts = []
            if self._preferredquality is not None:
-                # The opus codec doesn't support the -aq option
-                if int(self._preferredquality) < 10 and extension != 'opus':
-                    more_opts += ['-q:a', self._preferredquality]
+                if int(self._preferredquality) < 10:
+                    if extension == 'opus':
+                        # The opus codec doesn't support the -q:a option
+                        # Magic number of 24 used to give quality numbers 0-9 *some* meaning (as suggested by "knarrff" on Github)
+                        # 0 (best quality) gives 210 kb/s
+                        # 5 (default) gives 96 kb/s
+                        # 9 (worst quality) gives 6 kb/s, which is the opus minimum (and 0 kb/s wouldn't make any sense)
+                        more_opts += ['-b:a', compat_str(max(6, int((9 - int(self._preferredquality)) * 24))) + 'k']
+                    else:
+                        more_opts += ['-q:a', self._preferredquality]
                else:
                    more_opts += ['-b:a', self._preferredquality + 'k']
            if self._preferredcodec == 'aac':
Author	SHA1	Message	Date
d3fault	2b8165a1b5	Merge 5831f6feef00d58334f144f9cf5d0b65c5502a43 into da7223d4aa42ff9fc680b0951d043dd03cec2d30	2025-03-22 07:20:23 +08:00
dirkf	da7223d4aa	[YouTube] Improve support for tce-style player JS * 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>	2025-03-21 16:26:25 +00:00
dirkf	37c2440d6a	[YouTube] Update player client data thx seproDev (yt-dlp/yt-dlp#12603) Co-authored-by: sepro <sepro@sepr0.com>	2025-03-21 16:13:24 +00:00
d3fault	5831f6feef	[postproc/ffmpeg] fix opus using default quality of 5 as bitrate (closes #28389 ) Fix #28389. When extracting audio, and the containing format isn't 'opus', but the requested format is, then youtube-dl will call ffmpeg to do the conversion. Without specifying the preferred quality, youtube-dl defaults to '5'. Integers 0-9 are currently sent to ffmpeg as the bitrate for 'opus', which is clearly an error and this commit fixes that. It uses an arbitrary (specific to opus) scale to convert 0-9 into reasonable bitrates between 6 kb/s to 210 kb/s.... with 5 (the default) giving 96 kb/s	2022-05-29 00:26:26 -07:00