aboutsummaryrefslogtreecommitdiffhomepage
diff options
context:
space:
mode:
authorAiur Adept <[email protected]>2024-08-01 14:18:34 -0400
committerGitHub <[email protected]>2024-08-01 19:18:34 +0100
commit71223bff39551a11b6959a3de2dd9e2f070f3c4f (patch)
tree2c82ef18f1d4bdf73b7fad98637f7d2019c49090
parente1b3fa242cb94eb9dc949ab3f2cace91c46f11bf (diff)
downloadyoutube-dl-71223bff39551a11b6959a3de2dd9e2f070f3c4f.tar.gz
youtube-dl-71223bff39551a11b6959a3de2dd9e2f070f3c4f.zip
[Youtube] Fix nsig extraction for player 20dfca59 (#32891)
* dirkf's patch for nsig extraction * add generic search per yt-dlp/yt-dlp/pull/10611 - thx bashonly --------- Co-authored-by: dirkf <[email protected]>
-rw-r--r--test/test_youtube_signature.py4
-rw-r--r--youtube_dl/extractor/youtube.py32
2 files changed, 30 insertions, 6 deletions
diff --git a/test/test_youtube_signature.py b/test/test_youtube_signature.py
index 5b4aa3aa0..1c5f667f5 100644
--- a/test/test_youtube_signature.py
+++ b/test/test_youtube_signature.py
@@ -174,6 +174,10 @@ _NSIG_TESTS = [
'https://www.youtube.com/s/player/5604538d/player_ias.vflset/en_US/base.js',
'7X-he4jjvMx7BCX', 'sViSydX8IHtdWA',
),
+ (
+ 'https://www.youtube.com/s/player/20dfca59/player_ias.vflset/en_US/base.js',
+ '-fLCxedkAk4LUTK2', 'O8kfRq1y1eyHGw',
+ ),
]
diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 84371ff06..509e374a4 100644
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -1659,18 +1659,38 @@ 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
+ # or: (b="nn"[+a.D],c=a.get(b))&&(c=nfunc[idx](c)
+ # or: (PL(a),b=a.j.n||null)&&(b=nfunc[idx](b)
# old: .get("n"))&&(b=nfunc[idx](b)
# older: .get("n"))&&(b=nfunc(b)
r'''(?x)
- (?:\(\s*(?P<b>[a-z])\s*=\s*(?:
+ (?:\((?:[\w$()\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*
+ "n+"\[\s*\+?s*[\w$.]+\s*]|
+ (?P<b1>(?:[\w$]+\s*\.\s*)+n\b(?:(?!&&).)+\))
+ )\s*
+ (?(b1)
+ &&\s*\(\s*(?P=b)|
+ (?:
+ ,(?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)|(?P=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'))
+ ''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'),
+ default=(None, None))
+ # thx bashonly: yt-dlp/yt-dlp/pull/10611
+ if not func_name:
+ self.report_warning('Falling back to generic n function search')
+ return self._search_regex(
+ r'''(?xs)
+ (?:(?<=[^\w$])|^) # instead of \b, which ignores $
+ (?P<name>(?!\d)[a-zA-Z\d_$]+)\s*=\s*function\((?!\d)[a-zA-Z\d_$]+\)
+ \s*\{(?:(?!};).)+?["']enhanced_except_
+ ''', jscode, 'Initial JS player n function name', group='name')
if not idx:
+ self.report_warning('Falling back to generic n function search')
return func_name
return self._parse_json(self._search_regex(