Compare commits

..

3 Commits

Author SHA1 Message Date
cdarlint
c8ea79651a
Merge 2d900facc4 into e1b3fa242c 2024-07-25 17:41:58 +07:00
dirkf
e1b3fa242c [Youtube] Find n function name in player 3400486c
Fixes #32877
2024-07-25 00:16:00 +01:00
dirkf
451046d62a [Youtube] Make n-sig throttling diagnostic up-to-date 2024-07-24 14:33:34 +01:00
2 changed files with 16 additions and 4 deletions

View File

@ -166,6 +166,14 @@ _NSIG_TESTS = [
'https://www.youtube.com/s/player/b22ef6e7/player_ias.vflset/en_US/base.js',
'b6HcntHGkvBLk_FRf', 'kNPW6A7FyP2l8A',
),
(
'https://www.youtube.com/s/player/3400486c/player_ias.vflset/en_US/base.js',
'lL46g3XifCKUZn1Xfw', 'z767lhet6V2Skl',
),
(
'https://www.youtube.com/s/player/5604538d/player_ias.vflset/en_US/base.js',
'7X-he4jjvMx7BCX', 'sViSydX8IHtdWA',
),
]

View File

@ -1647,7 +1647,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
except JSInterpreter.Exception as e:
self.report_warning(
'%s (%s %s)' % (
'Unable to decode n-parameter: download likely to be throttled',
'Unable to decode n-parameter: expect download to be blocked or throttled',
error_to_compat_str(e),
traceback.format_exc()),
video_id=video_id)
@ -1659,18 +1659,22 @@ 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
# old: .get("n"))&&(b=nfunc[idx](b)
# older: .get("n"))&&(b=nfunc(b)
r'''(?x)
(?:\(\s*(?P<b>[a-z])\s*=\s*String\s*\.\s*fromCharCode\s*\(\s*110\s*\)\s*,(?P<c>[a-z])\s*=\s*[a-z]\s*)?
\.\s*get\s*\(\s*(?(b)(?P=b)|"n")(?:\s*\)){2}\s*&&\s*\(\s*(?(c)(?P=c)|b)\s*=\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*
(?P<nfunc>[a-zA-Z_$][\w$]*)(?:\s*\[(?P<idx>\d+)\])?\s*\(\s*[\w$]+\s*\)
''', jscode, 'Initial JS player n function name', group=('nfunc', 'idx'))
if not idx:
return func_name
return self._parse_json(self._search_regex(
r'var {0}\s*=\s*(\[.+?\])\s*[,;]'.format(re.escape(func_name)), jscode,
r'var\s+{0}\s*=\s*(\[.+?\])\s*[,;]'.format(re.escape(func_name)), jscode,
'Initial JS player n function list ({0}.{1})'.format(func_name, idx)),
func_name, transform_source=js_to_json)[int(idx)]