1 from __future__ import unicode_literals
3 from .common import InfoExtractor
6 class KarriereVideosIE(InfoExtractor):
7 _VALID_URL = r'http://(?:www\.)?karrierevideos\.at/berufsvideos/([a-z-]+)/(?P<id>[a-z-]+)'
9 'url': 'http://www.karrierevideos.at/berufsvideos/mittlere-hoehere-schulen/altenpflegerin',
11 'id': 'altenpflegerin',
13 'title': 'AltenpflegerIn',
14 'thumbnail': 're:^http://.*\.png\?v=[0-9]+',
15 'description': 'md5:dbadd1259fde2159a9b28667cb664ae2'
18 'skip_download': 'requires rtmpdump'
22 def _real_extract(self, url):
23 video_id = self._match_id(url)
24 webpage = self._download_webpage(url, video_id)
26 description = self._html_search_regex(
27 r'<div class="leadtext">\n{0,}?\s{0,}<p>(.*?)</p>',
28 webpage, 'description')
30 playlist = self._html_search_regex(r'/config/video/(.*?)\.xml', webpage, 'playlist')
31 playlist = self._download_xml(
32 'http://www.karrierevideos.at/player-playlist.xml.php?p=' + playlist,
35 namespace = 'http://developer.longtailvideo.com/trac/wiki/FlashFormats'
37 item = playlist.find('tracklist/item')
38 streamer = item.find('{%s}streamer' % namespace).text
42 'title': self._html_search_meta('title', webpage),
43 'description': description,
44 'thumbnail': 'http://www.karrierevideos.at' + self._html_search_meta('thumbnail', webpage),
46 'url': streamer.replace('rtmpt', 'http'),
47 'play_path': 'mp4:' + item.find('{%s}file' % namespace).text,