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             'format': 'bestvideo',
  30             'skip_download': True,
  31         },
  32     }, {
  33         'url': 'https://rmcdecouverte.bfmtv.com/wheeler-dealers-occasions-a-saisir/program_2566/',
  34         'info_dict': {
  35             'id': '5983675500001',
  36             'ext': 'mp4',
  37             'title': 'CORVETTE',
  38             'description': 'md5:c1e8295521e45ffebf635d6a7658f506',
  39             'uploader_id': '1969646226001',
  40             'upload_date': '20181226',
  41             'timestamp': 1545861635,
  42         },
  43         'params': {
  44             'skip_download': True,
  45         },
  46         'skip': 'only available for a week',
  47     }, {
  48         'url': 'https://rmcdecouverte.bfmtv.com/avions-furtifs-la-technologie-de-lextreme_10598',
  49         'only_matching': True,
  50     }, {
  51         # The website accepts any URL as long as it has _\d+ at the end
  52         'url': 'https://rmcdecouverte.bfmtv.com/any/thing/can/go/here/_10598',
  53         'only_matching': True,
  54     }, {
  55         # live, geo restricted, bypassable
  56         'url': 'https://rmcdecouverte.bfmtv.com/mediaplayer-direct/',
  57         'only_matching': True,
  58     }]
  59     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1969646226001/default_default/index.html?videoId=%s'
  60
  61     def _real_extract(self, url):
  62         mobj = self._match_valid_url(url)
  63         display_id = mobj.group('id') or 'direct'
  64         webpage = self._download_webpage(url, display_id)
  65         brightcove_legacy_url = BrightcoveLegacyIE._extract_brightcove_url(webpage)
  66         if brightcove_legacy_url:
  67             brightcove_id = compat_parse_qs(compat_urlparse.urlparse(
  68                 brightcove_legacy_url).query)['@videoPlayer'][0]
  69         else:
  70             brightcove_id = self._search_regex(
  71                 r'data-video-id=["\'](\d+)', webpage, 'brightcove id')
  72         return self.url_result(
  73             smuggle_url(
  74                 self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
  75                 {'geo_countries': ['FR']}),
  76             'BrightcoveNew', brightcove_id)