youtube_dlc/extractor/sky.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     extract_attributes,
   9     smuggle_url,
  10     strip_or_none,
  11     urljoin,
  12 )
  13
  14
  15 class SkyBaseIE(InfoExtractor):
  16     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_default/index.html?videoId=%s'
  17     _SDC_EL_REGEX = r'(?s)(<div[^>]+data-(?:component-name|fn)="sdc-(?:articl|sit)e-video"[^>]*>)'
  18
  19     def _process_ooyala_element(self, webpage, sdc_el, url):
  20         sdc = extract_attributes(sdc_el)
  21         provider = sdc.get('data-provider')
  22         if provider == 'ooyala':
  23             video_id = sdc['data-sdc-video-id']
  24             video_url = 'ooyala:%s' % video_id
  25             ie_key = 'Ooyala'
  26             ooyala_el = self._search_regex(
  27                 r'(<div[^>]+class="[^"]*\bsdc-article-video__media-ooyala\b[^"]*"[^>]+data-video-id="%s"[^>]*>)' % video_id,
  28                 webpage, 'video data', fatal=False)
  29             if ooyala_el:
  30                 ooyala_attrs = extract_attributes(ooyala_el) or {}
  31                 if ooyala_attrs.get('data-token-required') == 'true':
  32                     token_fetch_url = (self._parse_json(ooyala_attrs.get(
  33                         'data-token-fetch-options', '{}'),
  34                         video_id, fatal=False) or {}).get('url')
  35                     if token_fetch_url:
  36                         embed_token = self._download_json(urljoin(
  37                             url, token_fetch_url), video_id, fatal=False)
  38                         if embed_token:
  39                             video_url = smuggle_url(
  40                                 video_url, {'embed_token': embed_token})
  41         elif provider == 'brightcove':
  42             video_id = sdc['data-video-id']
  43             account_id = sdc.get('data-account-id') or '6058004172001'
  44             player_id = sdc.get('data-player-id') or 'RC9PQUaJ6'
  45             video_url = self.BRIGHTCOVE_URL_TEMPLATE % (account_id, player_id, video_id)
  46             ie_key = 'BrightcoveNew'
  47
  48         return {
  49             '_type': 'url_transparent',
  50             'id': video_id,
  51             'url': video_url,
  52             'ie_key': ie_key,
  53         }
  54
  55     def _real_extract(self, url):
  56         video_id = self._match_id(url)
  57         webpage = self._download_webpage(url, video_id)
  58         info = self._process_ooyala_element(webpage, self._search_regex(
  59             self._SDC_EL_REGEX, webpage, 'sdc element'), url)
  60         info.update({
  61             'title': self._og_search_title(webpage),
  62             'description': strip_or_none(self._og_search_description(webpage)),
  63         })
  64         return info
  65
  66
  67 class SkySportsIE(SkyBaseIE):
  68     IE_NAME = 'sky:sports'
  69     _VALID_URL = r'https?://(?:www\.)?skysports\.com/watch/video/([^/]+/)*(?P<id>[0-9]+)'
  70     _TESTS = [{
  71         'url': 'http://www.skysports.com/watch/video/10328419/bale-its-our-time-to-shine',
  72         'md5': '77d59166cddc8d3cb7b13e35eaf0f5ec',
  73         'info_dict': {
  74             'id': 'o3eWJnNDE6l7kfNO8BOoBlRxXRQ4ANNQ',
  75             'ext': 'mp4',
  76             'title': 'Bale: It\'s our time to shine',
  77             'description': 'md5:e88bda94ae15f7720c5cb467e777bb6d',
  78         },
  79         'add_ie': ['Ooyala'],
  80     }, {
  81         'url': 'https://www.skysports.com/watch/video/sports/f1/12160544/abu-dhabi-gp-the-notebook',
  82         'only_matching': True,
  83     }, {
  84         'url': 'https://www.skysports.com/watch/video/tv-shows/12118508/rainford-brent-how-ace-programme-helps',
  85         'only_matching': True,
  86     }]
  87
  88
  89 class SkyNewsIE(SkyBaseIE):
  90     IE_NAME = 'sky:news'
  91     _VALID_URL = r'https?://news\.sky\.com/video/[0-9a-z-]+-(?P<id>[0-9]+)'
  92     _TEST = {
  93         'url': 'https://news.sky.com/video/russian-plane-inspected-after-deadly-fire-11712962',
  94         'md5': '411e8893fd216c75eaf7e4c65d364115',
  95         'info_dict': {
  96             'id': 'ref:1ua21xaDE6lCtZDmbYfl8kwsKLooJbNM',
  97             'ext': 'mp4',
  98             'title': 'Russian plane inspected after deadly fire',
  99             'description': 'The Russian Investigative Committee has released video of the wreckage of a passenger plane which caught fire near Moscow.',
 100             'uploader_id': '6058004172001',
 101             'timestamp': 1567112345,
 102             'upload_date': '20190829',
 103         },
 104         'add_ie': ['BrightcoveNew'],
 105     }
 106
 107
 108 class SkySportsNewsIE(SkyBaseIE):
 109     IE_NAME = 'sky:sports:news'
 110     _VALID_URL = r'https?://(?:www\.)?skysports\.com/([^/]+/)*news/\d+/(?P<id>\d+)'
 111     _TEST = {
 112         'url': 'http://www.skysports.com/golf/news/12176/10871916/dustin-johnson-ready-to-conquer-players-championship-at-tpc-sawgrass',
 113         'info_dict': {
 114             'id': '10871916',
 115             'title': 'Dustin Johnson ready to conquer Players Championship at TPC Sawgrass',
 116             'description': 'Dustin Johnson is confident he can continue his dominant form in 2017 by adding the Players Championship to his list of victories.',
 117         },
 118         'playlist_count': 2,
 119     }
 120
 121     def _real_extract(self, url):
 122         article_id = self._match_id(url)
 123         webpage = self._download_webpage(url, article_id)
 124
 125         entries = []
 126         for sdc_el in re.findall(self._SDC_EL_REGEX, webpage):
 127             entries.append(self._process_ooyala_element(webpage, sdc_el, url))
 128
 129         return self.playlist_result(
 130             entries, article_id, self._og_search_title(webpage),
 131             self._html_search_meta(['og:description', 'description'], webpage))