From: Sergey M․ <dstftw@gmail.com>
Date: Mon, 2 Mar 2020 23:23:39 +0000 (+0700)
Subject: [pornhub] Improve title extraction (closes #24184)
X-Git-Url: http://git.bitcoin.ninja/?a=commitdiff_plain;h=46cc54ca8f13c7b823c1a12446cdd76d060c74b6;p=youtube-dl

[pornhub] Improve title extraction (closes #24184)
---

diff --git a/youtube_dl/extractor/pornhub.py b/youtube_dl/extractor/pornhub.py
index b3251ccd9..b8f65af7c 100644
--- a/youtube_dl/extractor/pornhub.py
+++ b/youtube_dl/extractor/pornhub.py
@@ -189,10 +189,10 @@ class PornHubIE(PornHubBaseIE):
         # http://www.pornhub.com/view_video.php?viewkey=1331683002), not relying
         # on that anymore.
         title = self._html_search_meta(
-            'twitter:title', webpage, default=None) or self._search_regex(
-            (r'<h1[^>]+class=["\']title["\'][^>]*>(?P<title>[^<]+)',
-             r'<div[^>]+data-video-title=(["\'])(?P<title>.+?)\1',
-             r'shareTitle\s*=\s*(["\'])(?P<title>.+?)\1'),
+            'twitter:title', webpage, default=None) or self._html_search_regex(
+            (r'(?s)<h1[^>]+class=["\']title["\'][^>]*>(?P<title>.+?)</h1>',
+             r'<div[^>]+data-video-title=(["\'])(?P<title>(?:(?!\1).)+)\1',
+             r'shareTitle["\']\s*[=:]\s*(["\'])(?P<title>(?:(?!\1).)+)\1'),
             webpage, 'title', group='title')
 
         video_urls = []