2 from __future__ import unicode_literals
6 from .common import InfoExtractor
9 class Min20IE(InfoExtractor):
10 _VALID_URL = r'http://www\.20min\.ch/(videotv/\?vid=(?P<video_id>[0-9]+)|.+?-(?P<page_id>[0-9]+)$)'
12 'url': 'http://www.20min.ch/schweiz/news/story/-Wir-muessen-mutig-nach-vorne-schauen--22050469',
13 'md5': 'cd4cbb99b94130cff423e967cd275e5e',
17 'title': '«Wir müssen mutig nach vorne schauen»',
18 'description': 'Kein Land sei innovativer als die Schweiz, sagte Johann Schneider-Ammann in seiner Neujahrsansprache. Das Land müsse aber seine Hausaufgaben machen.',
19 'thumbnail': 'http://www.20min.ch/images/content/2/2/0/22050469/10/teaserbreit.jpg'
22 'url': 'http://www.20min.ch/videotv/?vid=469148&cid=2',
23 'md5': 'b52d6bc6ea6398e6a38f12cfd418149c',
27 'title': '85 000 Franken für 15 perfekte Minuten',
28 'description': 'Was die Besucher vom Silvesterzauber erwarten können. (Video: Alice Grosjean/Murat Temel)',
29 'thumbnail': 'http://thumbnails.20min-tv.ch/server063/469148/frame-72-469148.jpg'
33 # location of the flv videos, can't be extracted from the web page
34 _BASE_URL = "http://flv-rr.20min-tv.ch/videos/"
36 def _real_extract(self, url):
37 mobj = re.match(self._VALID_URL, url)
38 video_id = mobj.group('page_id')
40 # URL from the videoportal
41 video_id = mobj.group('video_id')
42 webpage = self._download_webpage(url, video_id)
43 title = self._html_search_regex(r'<h1>.*<span>(.+?)</span></h1>', webpage, 'title')
44 flash_id = self._search_regex(r"so\.addVariable\(\"file1\",\"([0-9]+)\"\)", webpage, 'flash_id')
46 description = self._html_search_regex(r'<meta name="description" content="(.+?)" />', webpage, 'description')
47 thumbnail = self._html_search_regex(r'<meta property="og:image" content="(.+?)" />', webpage, 'thumbnail')
48 url = self._BASE_URL + flash_id + "m.flv"
54 'description': description,
55 'thumbnail': thumbnail