1
0
mirror of https://codeberg.org/polarisfm/youtube-dl synced 2024-11-16 06:34:31 +01:00

Using _og_search_description() as a Fallback for Description-Regex

This commit is contained in:
Thomas Christlieb 2017-02-01 10:54:29 +01:00
parent 94eb1f9229
commit b6449e98fe

View File

@ -342,15 +342,6 @@ class ProSiebenSat1IE(ProSiebenSat1BaseIE):
r'<h2 class="video-title" itemprop="name">\s*(.+?)</h2>', r'<h2 class="video-title" itemprop="name">\s*(.+?)</h2>',
r'<div[^>]+id="veeseoTitle"[^>]*>(.+?)</div>', r'<div[^>]+id="veeseoTitle"[^>]*>(.+?)</div>',
] ]
_DESCRIPTION_REGEXES = [
r'<p itemprop="description">\s*(.+?)</p>',
r'<div class="videoDecription">\s*<p><strong>Beschreibung</strong>: (.+?)</p>',
r'<div class="g-plusone" data-size="medium"></div>\s*</div>\s*</header>\s*(.+?)\s*<footer>',
r'<p class="att-description">\s*(.+?)\s*</p>',
r'<p class="video-description" itemprop="description">\s*(.+?)</p>',
r'<div[^>]+id="veeseoDescription"[^>]*>(.+?)</div>',
r'<meta property="og:description" content="(.*?)">',
]
_UPLOAD_DATE_REGEXES = [ _UPLOAD_DATE_REGEXES = [
r'<meta property="og:published_time" content="(.+?)">', r'<meta property="og:published_time" content="(.+?)">',
r'<span>\s*(\d{2}\.\d{2}\.\d{4} \d{2}:\d{2}) \|\s*<span itemprop="duration"', r'<span>\s*(\d{2}\.\d{2}\.\d{4} \d{2}:\d{2}) \|\s*<span itemprop="duration"',
@ -375,8 +366,7 @@ class ProSiebenSat1IE(ProSiebenSat1BaseIE):
self._CLIPID_REGEXES, webpage, 'clip id') self._CLIPID_REGEXES, webpage, 'clip id')
title = self._html_search_regex(self._TITLE_REGEXES, webpage, 'title') title = self._html_search_regex(self._TITLE_REGEXES, webpage, 'title')
info = self._extract_video_info(url, clip_id) info = self._extract_video_info(url, clip_id)
description = self._html_search_regex( description = self._og_search_description(webpage, default=None)
self._DESCRIPTION_REGEXES, webpage, 'description', fatal=False)
thumbnail = self._og_search_thumbnail(webpage) thumbnail = self._og_search_thumbnail(webpage)
upload_date = unified_strdate(self._html_search_regex( upload_date = unified_strdate(self._html_search_regex(
self._UPLOAD_DATE_REGEXES, webpage, 'upload date', default=None)) self._UPLOAD_DATE_REGEXES, webpage, 'upload date', default=None))