[porncom] Fix metadata extraction (closes #14911)

This commit is contained in:
Sergey M․ 2017-12-05 22:42:02 +07:00
parent c94427dd60
commit 61d18c8a4b
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -77,12 +77,14 @@ class PornComIE(InfoExtractor):
self._sort_formats(formats) self._sort_formats(formats)
view_count = str_to_int(self._search_regex( view_count = str_to_int(self._search_regex(
r'class=["\']views["\'][^>]*><p>([\d,.]+)', webpage, (r'Views:\s*</span>\s*<span>\s*([\d,.]+)',
r'class=["\']views["\'][^>]*><p>([\d,.]+)'), webpage,
'view count', fatal=False)) 'view count', fatal=False))
def extract_list(kind): def extract_list(kind):
s = self._search_regex( s = self._search_regex(
r'(?s)<p[^>]*>%s:(.+?)</p>' % kind.capitalize(), (r'(?s)%s:\s*</span>\s*<span>(.+?)</span>' % kind.capitalize(),
r'(?s)<p[^>]*>%s:(.+?)</p>' % kind.capitalize()),
webpage, kind, fatal=False) webpage, kind, fatal=False)
return re.findall(r'<a[^>]+>([^<]+)</a>', s or '') return re.findall(r'<a[^>]+>([^<]+)</a>', s or '')