1 # -*- coding: utf-8 -*-
2 from __future__ import unicode_literals
6 from .subtitles import SubtitlesInfoExtractor
10 compat_urllib_parse_urlparse,
19 class CeskaTelevizeIE(SubtitlesInfoExtractor):
20 _VALID_URL = r'https?://www\.ceskatelevize\.cz/(porady|ivysilani)/(.+/)?(?P<id>[^?#]+)'
24 'url': 'http://www.ceskatelevize.cz/ivysilani/ivysilani/10441294653-hyde-park-civilizace/214411058091220',
26 'id': '214411058091220',
28 'title': 'Hyde Park Civilizace',
29 'description': 'Věda a současná civilizace. Interaktivní pořad - prostor pro vaše otázky a komentáře',
30 'thumbnail': 're:^https?://.*\.jpg',
35 'skip_download': True,
39 'url': 'http://www.ceskatelevize.cz/ivysilani/10532695142-prvni-republika/bonus/14716-zpevacka-z-duparny-bobina',
43 'title': 'První republika: Zpěvačka z Dupárny Bobina',
44 'description': 'Sága mapující atmosféru první republiky od r. 1918 do r. 1945.',
45 'thumbnail': 're:^https?://.*\.jpg',
50 'skip_download': True,
55 def _real_extract(self, url):
56 url = url.replace('/porady/', '/ivysilani/').replace('/video/', '')
58 mobj = re.match(self._VALID_URL, url)
59 video_id = mobj.group('id')
61 webpage = self._download_webpage(url, video_id)
63 NOT_AVAILABLE_STRING = 'This content is not available at your territory due to limited copyright.'
64 if '%s</p>' % NOT_AVAILABLE_STRING in webpage:
65 raise ExtractorError(NOT_AVAILABLE_STRING, expected=True)
67 typ = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":"(.+?)","id":".+?"\}\],', webpage, 'type')
68 episode_id = self._html_search_regex(r'getPlaylistUrl\(\[\{"type":".+?","id":"(.+?)"\}\],', webpage, 'episode_id')
71 'playlist[0][type]': typ,
72 'playlist[0][id]': episode_id,
73 'requestUrl': compat_urllib_parse_urlparse(url).path,
74 'requestSource': 'iVysilani',
77 req = compat_urllib_request.Request(
78 'http://www.ceskatelevize.cz/ivysilani/ajax/get-client-playlist',
79 data=compat_urllib_parse.urlencode(data))
81 req.add_header('Content-type', 'application/x-www-form-urlencoded')
82 req.add_header('x-addr', '127.0.0.1')
83 req.add_header('X-Requested-With', 'XMLHttpRequest')
84 req.add_header('Referer', url)
86 playlistpage = self._download_json(req, video_id)
88 playlist_url = playlistpage['url']
89 if playlist_url == 'error_region':
90 raise ExtractorError(NOT_AVAILABLE_STRING, expected=True)
92 req = compat_urllib_request.Request(compat_urllib_parse.unquote(playlist_url))
93 req.add_header('Referer', url)
95 playlist = self._download_json(req, video_id)
97 item = playlist['playlist'][0]
99 for format_id, stream_url in item['streamUrls'].items():
100 formats.extend(self._extract_m3u8_formats(stream_url, video_id, 'mp4'))
101 self._sort_formats(formats)
103 title = self._og_search_title(webpage)
104 description = self._og_search_description(webpage)
105 duration = float_or_none(item.get('duration'))
106 thumbnail = item.get('previewImageUrl')
109 subtitles = self.extract_subtitles(video_id, webpage)
111 if self._downloader.params.get('listsubtitles', False):
112 self._list_available_subtitles(video_id, webpage)
115 subtitles = self._fix_subtitles(self.extract_subtitles(video_id, webpage))
120 'description': description,
121 'thumbnail': thumbnail,
122 'duration': duration,
124 'subtitles': subtitles,
127 def _fix_subtitles(self, subtitles):
128 """ Convert milisecond-based subtitles to SRT """
129 if subtitles is None:
130 return subtitles # subtitles not requested
132 def _msectotimecode(msec):
133 """ Helper utility to convert miliseconds to timecode """
135 for divider in [1000, 60, 60, 100]:
136 components.append(msec % divider)
138 return "{3:02}:{2:02}:{1:02},{0:03}".format(*components)
140 def _fix_subtitle(subtitle):
141 for line in subtitle.splitlines():
142 m = re.match(r"^ *([0-9]+); *([0-9]+) +([0-9]+) *$", line)
145 start, stop = (_msectotimecode(int(t)) for t in m.groups()[1:])
146 yield "{} --> {}".format(start, stop)
151 for k, v in subtitles.items():
152 fixed_subtitles[k] = "\r\n".join(_fix_subtitle(v))
153 return fixed_subtitles
155 def _get_available_subtitles(self, video_id, webpage):
156 video_id = video_id.partition('-')[0]
157 url = 'http://imgct.ceskatelevize.cz/cache/data/ivysilani/' \
158 'subtitles/{}/{}/sub.txt'.format(video_id[:3], video_id)
159 req = HEADRequest(url)
160 sub = self._request_webpage(
162 note="Checking subtitles",
163 errnote="No subtitles found",
165 return {'cs': url} if sub else {}