1
0
mirror of https://codeberg.org/polarisfm/youtube-dl synced 2024-11-30 03:58:24 +01:00

[4tube] Fix metadata extraction (Closes #10321)

This commit is contained in:
Sergey M․ 2016-08-13 19:55:09 +07:00
parent db535435b3
commit 77afa008dd
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -43,14 +43,14 @@ class FourTubeIE(InfoExtractor):
'uploadDate', webpage)) 'uploadDate', webpage))
thumbnail = self._html_search_meta('thumbnailUrl', webpage) thumbnail = self._html_search_meta('thumbnailUrl', webpage)
uploader_id = self._html_search_regex( uploader_id = self._html_search_regex(
r'<a class="img-avatar" href="[^"]+/channels/([^/"]+)" title="Go to [^"]+ page">', r'<a class="item-to-subscribe" href="[^"]+/channels/([^/"]+)" title="Go to [^"]+ page">',
webpage, 'uploader id', fatal=False) webpage, 'uploader id', fatal=False)
uploader = self._html_search_regex( uploader = self._html_search_regex(
r'<a class="img-avatar" href="[^"]+/channels/[^/"]+" title="Go to ([^"]+) page">', r'<a class="item-to-subscribe" href="[^"]+/channels/[^/"]+" title="Go to ([^"]+) page">',
webpage, 'uploader', fatal=False) webpage, 'uploader', fatal=False)
categories_html = self._search_regex( categories_html = self._search_regex(
r'(?s)><i class="icon icon-tag"></i>\s*Categories / Tags\s*.*?<ul class="list">(.*?)</ul>', r'(?s)><i class="icon icon-tag"></i>\s*Categories / Tags\s*.*?<ul class="[^"]*?list[^"]*?">(.*?)</ul>',
webpage, 'categories', fatal=False) webpage, 'categories', fatal=False)
categories = None categories = None
if categories_html: if categories_html:
@ -59,10 +59,10 @@ class FourTubeIE(InfoExtractor):
r'(?s)<li><a.*?>(.*?)</a>', categories_html)] r'(?s)<li><a.*?>(.*?)</a>', categories_html)]
view_count = str_to_int(self._search_regex( view_count = str_to_int(self._search_regex(
r'<meta itemprop="interactionCount" content="UserPlays:([0-9,]+)">', r'<meta[^>]+itemprop="interactionCount"[^>]+content="UserPlays:([0-9,]+)">',
webpage, 'view count', fatal=False)) webpage, 'view count', fatal=False))
like_count = str_to_int(self._search_regex( like_count = str_to_int(self._search_regex(
r'<meta itemprop="interactionCount" content="UserLikes:([0-9,]+)">', r'<meta[^>]+itemprop="interactionCount"[^>]+content="UserLikes:([0-9,]+)">',
webpage, 'like count', fatal=False)) webpage, 'like count', fatal=False))
duration = parse_duration(self._html_search_meta('duration', webpage)) duration = parse_duration(self._html_search_meta('duration', webpage))