[rmcdecouverte] Fix extraction
[youtube-dl] / youtube_dl / extractor / rmcdecouverte.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from .brightcove import BrightcoveLegacyIE
6 from ..compat import (
7     compat_parse_qs,
8     compat_urlparse,
9 )
10
11
12 class RMCDecouverteIE(InfoExtractor):
13     _VALID_URL = r'https?://rmcdecouverte\.bfmtv\.com/mediaplayer-replay.*?\bid=(?P<id>\d+)'
14
15     _TEST = {
16         'url': 'http://rmcdecouverte.bfmtv.com/mediaplayer-replay/?id=16548',
17         'info_dict': {
18             'id': '5411254766001',
19             'ext': 'mp4',
20             'title': '39/45:LE RESEAU DES FAUX BILLETS',
21             'description': 'ic Brunet propose un nouvel \u00e9pisode des Grains de sable de l\'Histoire sur la plus grosse affaire de contrefa\u00e7on de la Seconde Guerre mondiale.',
22             'uploader_id': '1969646226001',
23             'upload_date': '20170426',
24             'timestamp': 1493166610,
25         },
26         'params': {
27             # rtmp download
28             'skip_download': True,
29         },
30         'skip': 'Only works from France',
31     }
32     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1969646226001/default_default/index.html?videoId=%s'
33
34     def _real_extract(self, url):
35         video_id = self._match_id(url)
36         webpage = self._download_webpage(url, video_id)
37         brightcove_legacy_url = BrightcoveLegacyIE._extract_brightcove_url(webpage)
38         if brightcove_legacy_url:
39             brightcove_id = compat_parse_qs(compat_urlparse.urlparse(brightcove_legacy_url).query)['@videoPlayer'][0]
40         else:
41             brightcove_id = self._search_regex(r'data-video-id="(.*?)"', webpage, 'brightcove_id')
42         return self.url_result(self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id, 'BrightcoveNew', brightcove_id)