[youtube] Fix extraction.
[youtube-dl] / youtube_dl / extractor / karaoketv.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5
6
7 class KaraoketvIE(InfoExtractor):
8     _VALID_URL = r'https?://(?:www\.)?karaoketv\.co\.il/[^/]+/(?P<id>\d+)'
9     _TEST = {
10         'url': 'http://www.karaoketv.co.il/%D7%A9%D7%99%D7%A8%D7%99_%D7%A7%D7%A8%D7%99%D7%95%D7%A7%D7%99/58356/%D7%90%D7%99%D7%96%D7%95%D7%9F',
11         'info_dict': {
12             'id': '58356',
13             'ext': 'flv',
14             'title': 'קריוקי של איזון',
15         },
16         'params': {
17             # rtmp download
18             'skip_download': True,
19         }
20     }
21
22     def _real_extract(self, url):
23         video_id = self._match_id(url)
24
25         webpage = self._download_webpage(url, video_id)
26         api_page_url = self._search_regex(
27             r'<iframe[^>]+src=(["\'])(?P<url>https?://www\.karaoke\.co\.il/api_play\.php\?.+?)\1',
28             webpage, 'API play URL', group='url')
29
30         api_page = self._download_webpage(api_page_url, video_id)
31         video_cdn_url = self._search_regex(
32             r'<iframe[^>]+src=(["\'])(?P<url>https?://www\.video-cdn\.com/embed/iframe/.+?)\1',
33             api_page, 'video cdn URL', group='url')
34
35         video_cdn = self._download_webpage(video_cdn_url, video_id)
36         play_path = self._parse_json(
37             self._search_regex(
38                 r'var\s+options\s*=\s*({.+?});', video_cdn, 'options'),
39             video_id)['clip']['url']
40
41         settings = self._parse_json(
42             self._search_regex(
43                 r'var\s+settings\s*=\s*({.+?});', video_cdn, 'servers', default='{}'),
44             video_id, fatal=False) or {}
45
46         servers = settings.get('servers')
47         if not servers or not isinstance(servers, list):
48             servers = ('wowzail.video-cdn.com:80/vodcdn', )
49
50         formats = [{
51             'url': 'rtmp://%s' % server if not server.startswith('rtmp') else server,
52             'play_path': play_path,
53             'app': 'vodcdn',
54             'page_url': video_cdn_url,
55             'player_url': 'http://www.video-cdn.com/assets/flowplayer/flowplayer.commercial-3.2.18.swf',
56             'rtmp_real_time': True,
57             'ext': 'flv',
58         } for server in servers]
59
60         return {
61             'id': video_id,
62             'title': self._og_search_title(webpage),
63             'formats': formats,
64         }