]>
Commit | Line | Data |
---|---|---|
133b1886 S |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
4 | import re | |
5 | ||
6 | from .common import InfoExtractor | |
7 | from ..utils import remove_end | |
8 | ||
9 | ||
10 | class TwentyMinutenIE(InfoExtractor): | |
11 | IE_NAME = '20min' | |
12 | _VALID_URL = r'https?://(?:www\.)?20min\.ch/(?:videotv/*\?.*\bvid=(?P<id>\d+)|(?:[^/]+/)*(?P<display_id>[^/#?]+))' | |
13 | _TESTS = [{ | |
14 | # regular video | |
15 | 'url': 'http://www.20min.ch/videotv/?vid=469148&cid=2', | |
16 | 'md5': 'b52d6bc6ea6398e6a38f12cfd418149c', | |
17 | 'info_dict': { | |
18 | 'id': '469148', | |
19 | 'ext': 'flv', | |
20 | 'title': '85 000 Franken für 15 perfekte Minuten', | |
21 | 'description': 'Was die Besucher vom Silvesterzauber erwarten können. (Video: Alice Grosjean/Murat Temel)', | |
22 | 'thumbnail': 'http://thumbnails.20min-tv.ch/server063/469148/frame-72-469148.jpg' | |
23 | } | |
24 | }, { | |
25 | # news article with video | |
26 | 'url': 'http://www.20min.ch/schweiz/news/story/-Wir-muessen-mutig-nach-vorne-schauen--22050469', | |
27 | 'md5': 'cd4cbb99b94130cff423e967cd275e5e', | |
28 | 'info_dict': { | |
29 | 'id': '469408', | |
30 | 'display_id': '-Wir-muessen-mutig-nach-vorne-schauen--22050469', | |
31 | 'ext': 'flv', | |
32 | 'title': '«Wir müssen mutig nach vorne schauen»', | |
33 | 'description': 'Kein Land sei innovativer als die Schweiz, sagte Johann Schneider-Ammann in seiner Neujahrsansprache. Das Land müsse aber seine Hausaufgaben machen.', | |
34 | 'thumbnail': 'http://www.20min.ch/images/content/2/2/0/22050469/10/teaserbreit.jpg' | |
35 | } | |
36 | }, { | |
37 | 'url': 'http://www.20min.ch/videotv/?cid=44&vid=468738', | |
38 | 'only_matching': True, | |
39 | }, { | |
40 | 'url': 'http://www.20min.ch/ro/sortir/cinema/story/Grandir-au-bahut--c-est-dur-18927411', | |
41 | 'only_matching': True, | |
42 | }] | |
43 | ||
44 | def _real_extract(self, url): | |
45 | mobj = re.match(self._VALID_URL, url) | |
46 | video_id = mobj.group('id') | |
47 | display_id = mobj.group('display_id') or video_id | |
48 | ||
49 | webpage = self._download_webpage(url, display_id) | |
50 | ||
51 | title = self._html_search_regex( | |
52 | r'<h1>.*?<span>(.+?)</span></h1>', | |
53 | webpage, 'title', default=None) | |
54 | if not title: | |
55 | title = remove_end(re.sub( | |
56 | r'^20 [Mm]inuten.*? -', '', self._og_search_title(webpage)), ' - News') | |
57 | ||
58 | if not video_id: | |
59 | video_id = self._search_regex( | |
60 | r'"file\d?"\s*,\s*\"(\d+)', webpage, 'video id') | |
61 | ||
62 | description = self._html_search_meta( | |
63 | 'description', webpage, 'description') | |
64 | thumbnail = self._og_search_thumbnail(webpage) | |
65 | ||
66 | return { | |
67 | 'id': video_id, | |
68 | 'display_id': display_id, | |
69 | 'url': 'http://speed.20min-tv.ch/%sm.flv' % video_id, | |
70 | 'title': title, | |
71 | 'description': description, | |
72 | 'thumbnail': thumbnail, | |
73 | } |