[YouTube] Improve nsig function name extraction

Fixes player b7910ca8, using `,` vs `;`
See https://github.com/ytdl-org/youtube-dl/issues/32292#issuecomment-1602231170

Co-authored-by: dirkf
This commit is contained in:
pukkandan 2023-06-22 13:23:31 +05:30 committed by dirkf
parent 07af47960f
commit 9112e668a5
2 changed files with 16 additions and 14 deletions

View file

@ -1623,15 +1623,22 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
nfunc, idx = re.match(target, nfunc_and_idx).group('nfunc', 'idx')
if not idx:
return nfunc
VAR_RE_TMPL = r'var\s+%s\s*=\s*(?P<name>\[(?P<alias>%s)\])[;,]'
note = 'Initial JS player n function {0} (%s[%s])' % (nfunc, idx)
def search_function_code(needle, group):
return self._search_regex(
VAR_RE_TMPL % (re.escape(nfunc), needle), jscode,
note.format(group), group=group)
if int_or_none(idx) == 0:
real_nfunc = self._search_regex(
r'var %s\s*=\s*\[([a-zA-Z_$][\w$]*)\];' % (re.escape(nfunc), ), jscode,
'Initial JS player n function alias ({nfunc}[{idx}])'.format(**locals()))
real_nfunc = search_function_code(r'[a-zA-Z_$][\w$]*', group='alias')
if real_nfunc:
return real_nfunc
return self._parse_json(self._search_regex(
r'var %s\s*=\s*(\[.+?\]);' % (re.escape(nfunc), ), jscode,
'Initial JS player n function name ({nfunc}[{idx}])'.format(**locals())), nfunc, transform_source=js_to_json)[int(idx)]
return self._parse_json(
search_function_code('.+?', group='name'),
nfunc, transform_source=js_to_json)[int(idx)]
def _extract_n_function(self, video_id, player_url):
player_id = self._extract_player_info(player_url)