[karrierevideos] add support for www.karrierevideos.at (closes #5354)
[youtube-dl] / youtube_dl / extractor / karrierevideos.py
1 from __future__ import unicode_literals
2
3 from .common import InfoExtractor
4
5
6 class KarriereVideosIE(InfoExtractor):
7     _VALID_URL = r'http://(?:www\.)?karrierevideos\.at/berufsvideos/([a-z-]+)/(?P<id>[a-z-]+)'
8     _TEST = {
9         'url': 'http://www.karrierevideos.at/berufsvideos/mittlere-hoehere-schulen/altenpflegerin',
10         'info_dict': {
11             'id': 'altenpflegerin',
12             'ext': 'mp4',
13             'title': 'AltenpflegerIn',
14             'thumbnail': 're:^http://.*\.png\?v=[0-9]+',
15             'description': 'md5:dbadd1259fde2159a9b28667cb664ae2'
16         },
17         'params': {
18             'skip_download': 'requires rtmpdump'
19         }
20     }
21
22     def _real_extract(self, url):
23         video_id = self._match_id(url)
24         webpage = self._download_webpage(url, video_id)
25
26         description = self._html_search_regex(
27             r'<div class="leadtext">\n{0,}?\s{0,}<p>(.*?)</p>',
28             webpage, 'description')
29
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=%s' % playlist,
33             video_id)
34
35         namespace = {
36             'jwplayer': 'http://developer.longtailvideo.com/trac/wiki/FlashFormats'
37         }
38
39         item = playlist.find('tracklist/item')
40         streamer = item.find('jwplayer:streamer', namespace).text
41
42         return {
43             'id': video_id,
44             'title': self._html_search_meta('title', webpage),
45             'description': description,
46             'thumbnail': 'http://www.karrierevideos.at' + self._html_search_meta('thumbnail', webpage),
47             'protocol': 'rtmp',
48             'url': streamer.replace('rtmpt', 'http'),
49             'play_path': 'mp4:' + item.find('jwplayer:file', namespace).text,
50             'tc_url': streamer,
51             'ext': 'mp4'
52         }