1 # -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
7 from .common import InfoExtractor
11 compat_urllib_parse_urlparse,
16 class CeskaTelevizeIE(InfoExtractor):
17 _VALID_URL = r'https?://www\.ceskatelevize\.cz/(porady|ivysilani)/(.+/)?(?P<id>[^?#]+)'
21 'url': 'http://www.ceskatelevize.cz/ivysilani/10532695142-prvni-republika/213512120230004-spanelska-chripka',
23 'id': '213512120230004',
25 'title': 'První republika: Španělská chřipka',
29 'skip_download': True, # requires rtmpdump
31 'skip': 'Works only from Czech Republic.',
34 'url': 'http://www.ceskatelevize.cz/ivysilani/1030584952-tsatsiki-maminka-a-policajt',
38 'title': 'Tsatsiki, maminka a policajt',
42 'skip_download': True, # requires rtmpdump
44 'skip': 'Works only from Czech Republic.',
47 'url': 'http://www.ceskatelevize.cz/ivysilani/10532695142-prvni-republika/bonus/14716-zpevacka-z-duparny-bobina',
51 'title': 'První republika: Zpěvačka z Dupárny Bobina',
55 'skip_download': True, # requires rtmpdump
60 def _real_extract(self, url):
61 url = url.replace('/porady/', '/ivysilani/').replace('/video/', '')
63 mobj = re.match(self._VALID_URL, url)
64 video_id = mobj.group('id')
66 webpage = self._download_webpage(url, video_id)
68 NOT_AVAILABLE_STRING = 'This content is not available at your territory due to limited copyright.'
69 if '%s</p>' % NOT_AVAILABLE_STRING in webpage:
70 raise ExtractorError(NOT_AVAILABLE_STRING, expected=True)
72 typ = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":"(.+?)","id":".+?"\}\],', webpage, 'type')
73 episode_id = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":".+?","id":"(.+?)"\}\],', webpage, 'episode_id')
76 'playlist[0][type]': typ,
77 'playlist[0][id]': episode_id,
78 'requestUrl': compat_urllib_parse_urlparse(url).path,
79 'requestSource': 'iVysilani',
82 req = compat_urllib_request.Request('http://www.ceskatelevize.cz/ivysilani/ajax/get-playlist-url',
83 data=compat_urllib_parse.urlencode(data))
85 req.add_header('Content-type', 'application/x-www-form-urlencoded')
86 req.add_header('x-addr', '127.0.0.1')
87 req.add_header('X-Requested-With', 'XMLHttpRequest')
88 req.add_header('Referer', url)
90 playlistpage = self._download_webpage(req, video_id)
92 req = compat_urllib_request.Request(compat_urllib_parse.unquote(json.loads(playlistpage)['url']))
93 req.add_header('Referer', url)
95 playlist = self._download_xml(req, video_id)
98 for i in playlist.find('smilRoot/body'):
99 if 'AD' not in i.attrib['id']:
100 base_url = i.attrib['base']
101 parsedurl = compat_urllib_parse_urlparse(base_url)
102 duration = i.attrib['duration']
104 for video in i.findall('video'):
105 if video.attrib['label'] != 'AD':
106 format_id = video.attrib['label']
107 play_path = video.attrib['src']
108 vbr = int(video.attrib['system-bitrate'])
111 'format_id': format_id,
114 'play_path': play_path,
115 'app': parsedurl.path[1:] + '?' + parsedurl.query,
120 self._sort_formats(formats)
124 'title': self._html_search_regex(r'<title>(.+?) — iVysílání — Česká televize</title>', webpage, 'title'),
125 'duration': float(duration),