[extractor/youtube] Fallback regex for nsig code extraction

This commit is contained in:
pukkandan 2022-09-01 13:38:25 +05:30
parent f2e9fa3ef7
commit b505e8517a
No known key found for this signature in database
GPG Key ID: 7EEE9E1E817D0A39
1 changed files with 18 additions and 2 deletions

View File

@ -2661,7 +2661,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
raise ExtractorError('Cannot decrypt nsig without player_url') raise ExtractorError('Cannot decrypt nsig without player_url')
player_url = urljoin('https://www.youtube.com', player_url) player_url = urljoin('https://www.youtube.com', player_url)
try:
jsi, player_id, func_code = self._extract_n_function_code(video_id, player_url) jsi, player_id, func_code = self._extract_n_function_code(video_id, player_url)
except ExtractorError as e:
raise ExtractorError('Unable to extract nsig function code', cause=e)
if self.get_param('youtube_print_sig_code'): if self.get_param('youtube_print_sig_code'):
self.to_screen(f'Extracted nsig function from {player_id}:\n{func_code[1]}\n') self.to_screen(f'Extracted nsig function from {player_id}:\n{func_code[1]}\n')
@ -2706,7 +2709,20 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
if func_code: if func_code:
return jsi, player_id, func_code return jsi, player_id, func_code
func_code = jsi.extract_function_code(self._extract_n_function_name(jscode)) func_name = self._extract_n_function_name(jscode)
# For redundancy
func_code = self._search_regex(
r'''(?xs)%s\s*=\s*function\s*\((?P<var>[\w$]+)\)\s*
# NB: The end of the regex is intentionally kept strict
{(?P<code>.+?}\s*return\ [\w$]+.join\(""\))};''' % func_name,
jscode, 'nsig function', group=('var', 'code'), default=None)
if func_code:
func_code = ([func_code[0]], func_code[1])
else:
self.write_debug('Extracting nsig function with jsinterp')
func_code = jsi.extract_function_code(func_name)
self.cache.store('youtube-nsig', player_id, func_code) self.cache.store('youtube-nsig', player_id, func_code)
return jsi, player_id, func_code return jsi, player_id, func_code