youtube_dl/extractor/rmcdecouverte.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from .brightcove import BrightcoveLegacyIE
   8 from ..compat import (
   9     compat_parse_qs,
  10     compat_urlparse,
  11 )
  12 from ..utils import smuggle_url
  13
  14
  15 class RMCDecouverteIE(InfoExtractor):
  16     _VALID_URL = r'https?://rmcdecouverte\.bfmtv\.com/(?:(?:[^/]+/)*program_(?P<id>\d+)|(?P<live_id>mediaplayer-direct))'
  17
  18     _TESTS = [{
  19         'url': 'https://rmcdecouverte.bfmtv.com/wheeler-dealers-occasions-a-saisir/program_2566/',
  20         'info_dict': {
  21             'id': '5983675500001',
  22             'ext': 'mp4',
  23             'title': 'CORVETTE',
  24             'description': 'md5:c1e8295521e45ffebf635d6a7658f506',
  25             'uploader_id': '1969646226001',
  26             'upload_date': '20181226',
  27             'timestamp': 1545861635,
  28         },
  29         'params': {
  30             'skip_download': True,
  31         },
  32         'skip': 'only available for a week',
  33     }, {
  34         # live, geo restricted, bypassable
  35         'url': 'https://rmcdecouverte.bfmtv.com/mediaplayer-direct/',
  36         'only_matching': True,
  37     }]
  38     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/1969646226001/default_default/index.html?videoId=%s'
  39
  40     def _real_extract(self, url):
  41         mobj = re.match(self._VALID_URL, url)
  42         display_id = mobj.group('id') or mobj.group('live_id')
  43         webpage = self._download_webpage(url, display_id)
  44         brightcove_legacy_url = BrightcoveLegacyIE._extract_brightcove_url(webpage)
  45         if brightcove_legacy_url:
  46             brightcove_id = compat_parse_qs(compat_urlparse.urlparse(
  47                 brightcove_legacy_url).query)['@videoPlayer'][0]
  48         else:
  49             brightcove_id = self._search_regex(
  50                 r'data-video-id=["\'](\d+)', webpage, 'brightcove id')
  51         return self.url_result(
  52             smuggle_url(
  53                 self.BRIGHTCOVE_URL_TEMPLATE % brightcove_id,
  54                 {'geo_countries': ['FR']}),
  55             'BrightcoveNew', brightcove_id)