from ..utils import (
ExtractorError,
int_or_none,
+ js_to_json,
orderedSet,
sanitized_Request,
str_to_int,
'dislike_count': int,
'comment_count': int,
'age_limit': 18,
+ 'tags': list,
+ 'categories': list,
},
}, {
# non-ASCII title
'dislike_count': int,
'comment_count': int,
'age_limit': 18,
+ 'tags': list,
+ 'categories': list,
},
'params': {
'skip_download': True,
webpage = self._download_webpage(req, video_id)
error_msg = self._html_search_regex(
- r'(?s)<div[^>]+class=(["\']).*?\b(?:removed|userMessageSection)\b.*?\1[^>]*>(?P<error>.+?)</div>',
+ r'(?s)<div[^>]+class=(["\'])(?:(?!\1).)*\b(?:removed|userMessageSection)\b(?:(?!\1).)*\1[^>]*>(?P<error>.+?)</div>',
webpage, 'error message', default=None, group='error')
if error_msg:
error_msg = re.sub(r'\s+', ' ', error_msg)
})
self._sort_formats(formats)
+ page_params = self._parse_json(self._search_regex(
+ r'page_params\.zoneDetails\[([\'"])[^\'"]+\1\]\s*=\s*(?P<data>{[^}]+})',
+ webpage, 'page parameters', group='data', default='{}'),
+ video_id, transform_source=js_to_json, fatal=False)
+ tags = categories = None
+ if page_params:
+ tags = page_params.get('tags', '').split(',')
+ categories = page_params.get('categories', '').split(',')
+
return {
'id': video_id,
'uploader': video_uploader,
'comment_count': comment_count,
'formats': formats,
'age_limit': 18,
+ 'tags': tags,
+ 'categories': categories,
}