[screencast] Improve title regexes (Closes #9025)

This commit is contained in:
Sergey M․ 2016-04-01 23:01:55 +06:00
parent 791d6aaecc
commit 75d572e9fb

View File

@ -96,9 +96,9 @@ def _real_extract(self, url):
title = self._og_search_title(webpage, default=None) title = self._og_search_title(webpage, default=None)
if title is None: if title is None:
title = self._html_search_regex( title = self._html_search_regex(
[r'<b>Title:</b> ([^<]*)</div>', [r'<b>Title:</b> ([^<]+)</div>',
r'class="tabSeperator">></span><span class="tabText">(.*?)<', r'class="tabSeperator">></span><span class="tabText">(.+?)<',
r'<title>([^<]*)</title>'], r'<title>([^<]+)</title>'],
webpage, 'title') webpage, 'title')
thumbnail = self._og_search_thumbnail(webpage) thumbnail = self._og_search_thumbnail(webpage)
description = self._og_search_description(webpage, default=None) description = self._og_search_description(webpage, default=None)