[youporn] Fix title and description extraction (closes #18748)

This commit is contained in:
Sergey M․ 2019-01-09 00:37:01 +07:00
parent 2543938bbe
commit 6089ff40e7
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -68,11 +68,9 @@ class YouPornIE(InfoExtractor):
request.add_header('Cookie', 'age_verified=1') request.add_header('Cookie', 'age_verified=1')
webpage = self._download_webpage(request, display_id) webpage = self._download_webpage(request, display_id)
title = self._search_regex( title = self._html_search_regex(
[r'(?:video_titles|videoTitle)\s*[:=]\s*(["\'])(?P<title>(?:(?!\1).)+)\1', r'(?s)<div[^>]+class=["\']watchVideoTitle[^>]+>(.+?)</div>',
r'<h1[^>]+class=["\']heading\d?["\'][^>]*>(?P<title>[^<]+)<'], webpage, 'title', default=None) or self._og_search_title(
webpage, 'title', group='title',
default=None) or self._og_search_title(
webpage, default=None) or self._html_search_meta( webpage, default=None) or self._html_search_meta(
'title', webpage, fatal=True) 'title', webpage, fatal=True)
@ -134,7 +132,11 @@ class YouPornIE(InfoExtractor):
formats.append(f) formats.append(f)
self._sort_formats(formats) self._sort_formats(formats)
description = self._og_search_description(webpage, default=None) description = self._html_search_regex(
r'(?s)<div[^>]+\bid=["\']description["\'][^>]*>(.+?)</div>',
webpage, 'description',
default=None) or self._og_search_description(
webpage, default=None)
thumbnail = self._search_regex( thumbnail = self._search_regex(
r'(?:imageurl\s*=|poster\s*:)\s*(["\'])(?P<thumbnail>.+?)\1', r'(?:imageurl\s*=|poster\s*:)\s*(["\'])(?P<thumbnail>.+?)\1',
webpage, 'thumbnail', fatal=False, group='thumbnail') webpage, 'thumbnail', fatal=False, group='thumbnail')