yt_dlp/extractor/rmcdecouverte.py

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