yt_dlp/extractor/rmcdecouverte.py

   1 from .brightcove import BrightcoveLegacyIE
   2 from .common import InfoExtractor
   3 from ..compat import (
   4     compat_parse_qs,
   5     compat_urlparse,
   6 )
   7 from ..utils import smuggle_url
   8
   9
  10 class RMCDecouverteIE(InfoExtractor):
  11     _VALID_URL = r'https?://rmcdecouverte\.bfmtv\.com/(?:[^?#]*_(?P<id>\d+)|mediaplayer-direct)/?(?:[#?]|$)'
  12
  13     _TESTS = [{
  14         'url': 'https://rmcdecouverte.bfmtv.com/vestiges-de-guerre_22240/les-bunkers-secrets-domaha-beach_25303/',
  15         'info_dict': {
  16             'id': '6250879771001',
  17             'ext': 'mp4',
  18             'title': 'LES BUNKERS SECRETS D´OMAHA BEACH',
  19             'uploader_id': '1969646226001',
  20             'description': 'md5:aed573ca24abde62a148e0eba909657d',
  21             'timestamp': 1619622984,
  22             'upload_date': '20210428',
  23         },
  24         'params': {
  25             'skip_download': True,
  26         },
  27     }, {
  28         'url': 'https://rmcdecouverte.bfmtv.com/wheeler-dealers-occasions-a-saisir/program_2566/',
  29         'info_dict': {
  30             'id': '5983675500001',
  31             'ext': 'mp4',
  32             'title': 'CORVETTE',
  33             'description': 'md5:c1e8295521e45ffebf635d6a7658f506',
  34             'uploader_id': '1969646226001',
  35             'upload_date': '20181226',
  36             'timestamp': 1545861635,
  37         },
  38         'params': {
  39             'skip_download': True,
  40         },
  41         'skip': 'only available for a week',
  42     }, {
  43         'url': 'https://rmcdecouverte.bfmtv.com/avions-furtifs-la-technologie-de-lextreme_10598',
  44         'only_matching': True,
  45     }, {
  46         # The website accepts any URL as long as it has _\d+ at the end
  47         'url': 'https://rmcdecouverte.bfmtv.com/any/thing/can/go/here/_10598',
  48         'only_matching': True,
  49     }, {
  50         # live, geo restricted, bypassable
  51         'url': 'https://rmcdecouverte.bfmtv.com/mediaplayer-direct/',
  52         'only_matching': True,
  53     }]
  54     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1969646226001/default_default/index.html?videoId=%s'
  55
  56     def _real_extract(self, url):
  57         mobj = self._match_valid_url(url)
  58         display_id = mobj.group('id') or 'direct'
  59         webpage = self._download_webpage(url, display_id)
  60         brightcove_legacy_url = BrightcoveLegacyIE._extract_brightcove_url(webpage)
  61         if brightcove_legacy_url:
  62             brightcove_id = compat_parse_qs(compat_urlparse.urlparse(
  63                 brightcove_legacy_url).query)['@videoPlayer'][0]
  64         else:
  65             brightcove_id = self._search_regex(
  66                 r'data-video-id=["\'](\d+)', webpage, 'brightcove id')
  67         return self.url_result(
  68             smuggle_url(
  69                 self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
  70                 {'geo_countries': ['FR']}),
  71             'BrightcoveNew', brightcove_id)