[pornhub] Improve title extraction (closes #24184)

pull/8/head
Sergey M․ 5 years ago
parent 1e1c1960aa
commit 46cc54ca8f
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

@ -189,10 +189,10 @@ class PornHubIE(PornHubBaseIE):
# http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying # http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying
# on that anymore. # on that anymore.
title = self._html_search_meta( title = self._html_search_meta(
'twitter:title', webpage, default=None) or self._search_regex( 'twitter:title', webpage, default=None) or self._html_search_regex(
(r'<h1[^>]+class=["\']title["\'][^>]*>(?P<title>[^<]+)', (r'(?s)<h1[^>]+class=["\']title["\'][^>]*>(?P<title>.+?)</h1>',
r'<div[^>]+data-video-title=(["\'])(?P<title>.+?)\1', r'<div[^>]+data-video-title=(["\'])(?P<title>(?:(?!\1).)+)\1',
r'shareTitle\s*=\s*(["\'])(?P<title>.+?)\1'), r'shareTitle["\']\s*[=:]\s*(["\'])(?P<title>(?:(?!\1).)+)\1'),
webpage, 'title', group='title') webpage, 'title', group='title')
video_urls = [] video_urls = []

Loading…
Cancel
Save