[ie] Make _search_nuxt_data more lenient

Authored by: std-move

Co-authored-by: std-move <26625259+std-move@users.noreply.github.com>
This commit is contained in:
bashonly 2023-09-21 16:54:57 -05:00
parent 52414d64ca
commit 904a19ee93
No known key found for this signature in database
GPG key ID: 783F096F253D15B0

View file

@ -1687,7 +1687,7 @@ class InfoExtractor:
def _search_nuxt_data(self, webpage, video_id, context_name='__NUXT__', *, fatal=True, traverse=('data', 0)): def _search_nuxt_data(self, webpage, video_id, context_name='__NUXT__', *, fatal=True, traverse=('data', 0)):
"""Parses Nuxt.js metadata. This works as long as the function __NUXT__ invokes is a pure function""" """Parses Nuxt.js metadata. This works as long as the function __NUXT__ invokes is a pure function"""
rectx = re.escape(context_name) rectx = re.escape(context_name)
FUNCTION_RE = r'\(function\((?P<arg_keys>.*?)\){return\s+(?P<js>{.*?})\s*;?\s*}\((?P<arg_vals>.*?)\)' FUNCTION_RE = r'\(function\((?P<arg_keys>.*?)\){(?:.*?)return\s+(?P<js>{.*?})\s*;?\s*}\((?P<arg_vals>.*?)\)'
js, arg_keys, arg_vals = self._search_regex( js, arg_keys, arg_vals = self._search_regex(
(rf'<script>\s*window\.{rectx}={FUNCTION_RE}\s*\)\s*;?\s*</script>', rf'{rectx}\(.*?{FUNCTION_RE}'), (rf'<script>\s*window\.{rectx}={FUNCTION_RE}\s*\)\s*;?\s*</script>', rf'{rectx}\(.*?{FUNCTION_RE}'),
webpage, context_name, group=('js', 'arg_keys', 'arg_vals'), webpage, context_name, group=('js', 'arg_keys', 'arg_vals'),