2 from __future__ import unicode_literals
4 from .common import InfoExtractor
7 class SportschauIE(InfoExtractor):
9 _VALID_URL = r'https?://(?:www\.)?sportschau\.de/\w+(?:/\w+)?/video(?P<id>\w+)\.html'
11 'url': 'http://www.sportschau.de/tourdefrance/videoseppeltkokainhatnichtsmitklassischemdopingzutun100.html',
12 'md5': 'a6ef460ab9f4089b079832e06d554cec',
14 'id': 'seppeltkokainhatnichtsmitklassischemdopingzutun100',
16 'title': 'Seppelt: "Kokain hat nichts mit klassischem Doping zu tun" - Tour de France - sportschau.de',
17 'thumbnail': 're:^https?://.*\.jpg$',
18 'description': 'Der ARD-Doping Experte Hajo Seppelt gibt seine Einschätzung zum ersten Dopingfall der diesjährigen Tour de France um den Italiener Luca Paolini ab.',
22 def _real_extract(self, url):
23 video_id = self._match_id(url)
24 ext = '-mc_defaultQuality-h.json'
25 json_url = url[:-5] + ext
27 json = self._download_json(json_url, video_id)
28 thumb_url = json['_previewImage']
30 m3u8_url = json['_mediaArray'][1]['_mediaStreamArray'][0]['_stream'][0]
31 m3u8_formats = self._extract_m3u8_formats(m3u8_url, video_id, ext="mp4")
33 webpage = self._download_webpage(url, video_id)
34 title = self._html_search_regex(r'<title>(.*?)</title>', webpage, 'title')
35 desc = self._html_search_meta('description', webpage)
40 'formats': m3u8_formats,
42 'thumbnail': thumb_url,