yt_dlp/extractor/rmcdecouverte.py

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