2 from __future__ import unicode_literals
4 from .common import InfoExtractor
11 class TVCIE(InfoExtractor):
12 _VALID_URL = r'http://(?:www\.)?tvc\.ru/.*/show/.*id/(?P<id>\d+)'
15 'url': 'http://www.tvc.ru/channel/brand/id/29/show/episodes/episode_id/39702/',
16 'md5': 'aa6fb3cf384e18a0ad3b30ee2898beba',
19 'display_id': '39702',
21 'title': 'События. "События". Эфир от 22.05.2015 14:30',
22 'description': 'md5:ad7aa7db22903f983e687b8a3e98c6dd',
23 'thumbnail': 're:^https?://.*\.jpg$',
28 'url': 'http://www.tvc.ru/news/show/id/69944',
29 'md5': 'b173128ee7b88b5b06c84e5f7880909f',
32 'display_id': '69944',
34 'title': 'Эксперты: в столице встал вопрос о максимально безопасных остановках',
35 'description': 'md5:f675c8eaf23aab9df542d31773ed6518',
36 'thumbnail': 're:^https?://.*\.jpg$',
42 def _real_extract(self, url):
43 display_id = self._match_id(url)
45 webpage = self._download_webpage(url, display_id)
47 video_url = self._og_search_video_url(webpage)
49 video_id = self._search_regex(
50 r'video/iframe/id/(\d+)/', video_url, 'video id')
52 video_json_url = 'http://www.tvc.ru/video/json/id/%s' % (video_id)
54 video_json = self._download_json(video_json_url, video_id)
57 for info in video_json.get('path', {}).get('quality', []):
58 format_id = self._search_regex(
59 r'cdnvideo/([^-]+)-[^/]+/', info.get('url'), 'format id',
62 'format_id': str_or_none(format_id),
63 'url': info.get('url'),
64 'width': int_or_none(info.get('width')),
65 'height': int_or_none(info.get('height')),
66 'tbr': int_or_none(info.get('bitrate')),
69 self._sort_formats(formats)
73 'display_id': display_id,
74 'title': self._og_search_title(webpage),
75 'description': self._og_search_description(webpage),
76 'thumbnail': self._og_search_thumbnail(webpage),
77 'duration': int_or_none(video_json.get('duration')),