From: Sergey M․ Date: Thu, 30 Apr 2015 15:25:17 +0000 (+0600) Subject: [vporn] Fix metadata extraction (#5560) X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=commitdiff_plain;h=cd298882cd6f8ed2571dd372f684ec17e992fd9d;p=youtube-dl [vporn] Fix metadata extraction (#5560) --- diff --git a/youtube_dl/extractor/vporn.py b/youtube_dl/extractor/vporn.py index 2d23effcc..04ed0b381 100644 --- a/youtube_dl/extractor/vporn.py +++ b/youtube_dl/extractor/vporn.py @@ -64,29 +64,29 @@ class VpornIE(InfoExtractor): title = self._html_search_regex( r'videoname\s*=\s*\'([^\']+)\'', webpage, 'title').strip() description = self._html_search_regex( - r'
(.*?)
', webpage, 'description', fatal=False) + r'class="(?:descr|description_txt)">(.*?)', + webpage, 'description', fatal=False) thumbnail = self._html_search_regex( r'flashvars\.imageUrl\s*=\s*"([^"]+)"', webpage, 'description', fatal=False, default=None) if thumbnail: thumbnail = 'http://www.vporn.com' + thumbnail uploader = self._html_search_regex( - r'(?s)UPLOADED BY.*?([^<]+)', + r'(?s)Uploaded by:.*?([^<]+)', webpage, 'uploader', fatal=False) categories = re.findall(r'([^<]+)', webpage) duration = parse_duration(self._search_regex( - r'duration (\d+ min \d+ sec)', webpage, 'duration', fatal=False)) + r'Runtime:\s*\s*(\d+ min \d+ sec)', + webpage, 'duration', fatal=False)) - view_count = str_to_int(self._html_search_regex( - r'([\d,\.]+) VIEWS', webpage, 'view count', fatal=False)) - like_count = str_to_int(self._html_search_regex( - r'([\d,\.]+)', webpage, 'like count', fatal=False)) - dislike_count = str_to_int(self._html_search_regex( - r'([\d,\.]+)', webpage, 'dislike count', fatal=False)) + view_count = str_to_int(self._search_regex( + r'class="views">([\d,\.]+) [Vv]iews<', + webpage, 'view count', fatal=False)) comment_count = str_to_int(self._html_search_regex( - r'

Comments \(([\d,\.]+)\)

', webpage, 'comment count', fatal=False)) + r"'Comments \(([\d,\.]+)\)'", + webpage, 'comment count', default=None)) formats = [] @@ -117,8 +117,6 @@ class VpornIE(InfoExtractor): 'categories': categories, 'duration': duration, 'view_count': view_count, - 'like_count': like_count, - 'dislike_count': dislike_count, 'comment_count': comment_count, 'age_limit': 18, 'formats': formats,