2 from __future__ import unicode_literals
6 from .common import InfoExtractor
13 class SportBoxIE(InfoExtractor):
14 _VALID_URL = r'https?://news\.sportbox\.ru/(?:[^/]+/)+spbvideo_NI\d+_(?P<display_id>.+)'
17 'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S',
18 'md5': 'ff56a598c2cf411a9a38a69709e97079',
22 'title': 'Гонка 2 заезд ««Объединенный 2000»: классы Туринг и Супер-продакшн',
23 'description': 'md5:81715fa9c4ea3d9e7915dc8180c778ed',
24 'thumbnail': 're:^https?://.*\.jpg$',
25 'timestamp': 1411896237,
26 'upload_date': '20140928',
31 'skip_download': True,
35 'url': 'http://news.sportbox.ru/video/no_ads/spbvideo_NI536574_V_Novorossijske_proshel_detskij_turnir_Pole_slavy_bojevoj?ci=211355',
36 'md5': 'ff56a598c2cf411a9a38a69709e97079',
40 'title': 'В Новороссийске прошел детский турнир «Поле славы боевой»',
41 'description': '16 детских коллективов приняли участие в суперфинале турнира «Поле славы боевой».',
42 'thumbnail': 're:^https?://.*\.jpg$',
43 'timestamp': 1426237001,
44 'upload_date': '20150313',
49 'skip_download': True,
53 'url': 'http://news.sportbox.ru/Vidy_sporta/billiard/spbvideo_NI486287_CHempionat-mira-po-dinamichnoy-piramide-4',
54 'only_matching': True,
58 def _real_extract(self, url):
59 mobj = re.match(self._VALID_URL, url)
60 display_id = mobj.group('display_id')
62 webpage = self._download_webpage(url, display_id)
64 sobj = re.search(r'src="/vdl/player/(?P<media_type>\w+)/(?P<video_id>\d+)"', webpage)
66 video_id = sobj.group('video_id')
67 media_type = sobj.group('media_type')
69 raise RegexNotFoundError('Unable to extract video_id')
71 player = self._download_webpage(
72 'http://news.sportbox.ru/vdl/player/%s/%s' % (media_type, video_id),
73 display_id, 'Downloading player webpage')
75 hls = self._search_regex(
76 r"sportboxPlayer\.jwplayer_common_params\.file\s*=\s*['\"]+([^\"]+)['\"]+", player, 'hls file')
78 formats = self._extract_m3u8_formats(hls, display_id, 'mp4')
80 title = self._html_search_regex(
81 r'<h1 itemprop="name">([^<]+)</h1>', webpage, 'title')
82 description = self._html_search_regex(
83 r'(?s)<div itemprop="description">(.+?)</div>', webpage, 'description', fatal=False)
84 thumbnail = self._og_search_thumbnail(webpage)
85 timestamp = parse_iso8601(self._search_regex(
86 r'<span itemprop="uploadDate">([^<]+)</span>', webpage, 'timestamp', fatal=False))
87 duration = parse_duration(self._html_search_regex(
88 r'<meta itemprop="duration" content="PT([^"]+)">', webpage, 'duration', fatal=False))
92 'display_id': display_id,
94 'description': description,
95 'thumbnail': thumbnail,
96 'timestamp': timestamp,