]>
Commit | Line | Data |
---|---|---|
2cead7e7 | 1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
452908b2 S |
4 | import re |
5 | ||
2cead7e7 | 6 | from .common import InfoExtractor |
7 | ||
8 | ||
9 | class TeleBruxellesIE(InfoExtractor): | |
283d1c6a | 10 | _VALID_URL = r'https?://(?:www\.)?(?:telebruxelles|bx1)\.be/(news|sport|dernier-jt|emission)/?(?P<id>[^/#?]+)' |
2cead7e7 | 11 | _TESTS = [{ |
283d1c6a RA |
12 | 'url': 'http://bx1.be/news/que-risque-lauteur-dune-fausse-alerte-a-la-bombe/', |
13 | 'md5': 'a2a67a5b1c3e8c9d33109b902f474fd9', | |
2cead7e7 | 14 | 'info_dict': { |
283d1c6a RA |
15 | 'id': '158856', |
16 | 'display_id': 'que-risque-lauteur-dune-fausse-alerte-a-la-bombe', | |
17 | 'ext': 'mp4', | |
18 | 'title': 'Que risque l’auteur d’une fausse alerte à la bombe ?', | |
19 | 'description': 'md5:3cf8df235d44ebc5426373050840e466', | |
dd60be2b | 20 | }, |
2cead7e7 | 21 | }, { |
283d1c6a RA |
22 | 'url': 'http://bx1.be/sport/futsal-schaerbeek-sincline-5-3-a-thulin/', |
23 | 'md5': 'dfe07ecc9c153ceba8582ac912687675', | |
2cead7e7 | 24 | 'info_dict': { |
283d1c6a RA |
25 | 'id': '158433', |
26 | 'display_id': 'futsal-schaerbeek-sincline-5-3-a-thulin', | |
27 | 'ext': 'mp4', | |
28 | 'title': 'Futsal : Schaerbeek s’incline 5-3 à Thulin', | |
29 | 'description': 'md5:fd013f1488d5e2dceb9cebe39e2d569b', | |
dd60be2b | 30 | }, |
283d1c6a RA |
31 | }, { |
32 | 'url': 'http://bx1.be/emission/bxenf1-gastronomie/', | |
33 | 'only_matching': True, | |
dd60be2b | 34 | }] |
2cead7e7 | 35 | |
36 | def _real_extract(self, url): | |
dd60be2b PH |
37 | display_id = self._match_id(url) |
38 | webpage = self._download_webpage(url, display_id) | |
2cead7e7 | 39 | |
dd60be2b | 40 | article_id = self._html_search_regex( |
452908b2 | 41 | r"<article id=\"post-(\d+)\"", webpage, 'article ID', default=None) |
dd60be2b PH |
42 | title = self._html_search_regex( |
43 | r'<h1 class=\"entry-title\">(.*?)</h1>', webpage, 'title') | |
452908b2 | 44 | description = self._og_search_description(webpage, default=None) |
2cead7e7 | 45 | |
dd60be2b | 46 | rtmp_url = self._html_search_regex( |
452908b2 | 47 | r'file\s*:\s*"(rtmp://[^/]+/vod/mp4:"\s*\+\s*"[^"]+"\s*\+\s*".mp4)"', |
dd60be2b | 48 | webpage, 'RTMP url') |
452908b2 | 49 | rtmp_url = re.sub(r'"\s*\+\s*"', '', rtmp_url) |
283d1c6a RA |
50 | formats = self._extract_wowza_formats(rtmp_url, article_id or display_id) |
51 | self._sort_formats(formats) | |
dd60be2b PH |
52 | |
53 | return { | |
452908b2 | 54 | 'id': article_id or display_id, |
dd60be2b PH |
55 | 'display_id': display_id, |
56 | 'title': title, | |
57 | 'description': description, | |
283d1c6a | 58 | 'formats': formats, |
dd60be2b | 59 | } |