1 # -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
7 from .common import InfoExtractor
11 compat_urllib_parse_urlparse
15 class CeskaTelevizeIE(InfoExtractor):
16 _VALID_URL = r'https?://www\.ceskatelevize\.cz/(porady|ivysilani)/(.+/)?(?P<id>[^?#]+)'
19 'url': 'http://www.ceskatelevize.cz/ivysilani/10532695142-prvni-republika/213512120230004-spanelska-chripka',
21 'id': '213512120230004',
23 'title': 'První republika: Španělská chřipka',
27 'skip_download': True, # requires rtmpdump
29 'skip': 'Works only from Czech Republic.',
31 'url': 'http://www.ceskatelevize.cz/ivysilani/1030584952-tsatsiki-maminka-a-policajt',
35 'title': 'Tsatsiki, maminka a policajt',
39 'skip_download': True, # requires rtmpdump
41 'skip': 'Works only from Czech Republic.',
44 def _real_extract(self, url):
45 url = url.replace('/porady/', '/ivysilani/').replace('/video/', '')
47 mobj = re.match(self._VALID_URL, url)
48 video_id = mobj.group('id')
50 webpage = self._download_webpage(url, video_id)
52 typ = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":"(.+?)","id":".+?"\}\],', webpage, 'type')
53 episode_id = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":".+?","id":"(.+?)"\}\],', webpage, 'episode_id')
56 'playlist[0][type]': typ,
57 'playlist[0][id]': episode_id,
58 'requestUrl': compat_urllib_parse_urlparse(url).path,
59 'requestSource': 'iVysilani',
62 req = compat_urllib_request.Request('http://www.ceskatelevize.cz/ivysilani/ajax/get-playlist-url',
63 data=compat_urllib_parse.urlencode(data))
65 req.add_header('Content-type', 'application/x-www-form-urlencoded')
66 req.add_header('x-addr', '127.0.0.1')
67 req.add_header('X-Requested-With', 'XMLHttpRequest')
68 req.add_header('Referer', url)
70 playlistpage = self._download_webpage(req, video_id)
72 req = compat_urllib_request.Request(compat_urllib_parse.unquote(json.loads(playlistpage)['url']))
73 req.add_header('Referer', url)
75 playlist = self._download_xml(req, video_id)
78 for i in playlist.find('smilRoot/body'):
79 if 'AD' not in i.attrib['id']:
80 base_url = i.attrib['base']
81 parsedurl = compat_urllib_parse_urlparse(base_url)
82 duration = i.attrib['duration']
84 for video in i.findall('video'):
85 if video.attrib['label'] != 'AD':
86 format_id = video.attrib['label']
87 play_path = video.attrib['src']
88 vbr = int(video.attrib['system-bitrate'])
91 'format_id': format_id,
94 'play_path': play_path,
95 'app': parsedurl.path[1:] + '?' + parsedurl.query,
100 self._sort_formats(formats)
104 'title': self._html_search_regex(r'<title>(.+?) — iVysílání — Česká televize</title>', webpage, 'title'),
105 'duration': float(duration),