[youtube] Fix extraction.
[youtube-dl] / youtube_dl / extractor / sky.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import (
6     extract_attributes,
7     smuggle_url,
8     strip_or_none,
9     urljoin,
10 )
11
12
13 class SkyBaseIE(InfoExtractor):
14     def _real_extract(self, url):
15         video_id = self._match_id(url)
16         webpage = self._download_webpage(url, video_id)
17         video_data = extract_attributes(self._search_regex(
18             r'(<div.+?class="[^"]*sdc-article-video__media-ooyala[^"]*"[^>]+>)',
19             webpage, 'video data'))
20
21         video_url = 'ooyala:%s' % video_data['data-video-id']
22         if video_data.get('data-token-required') == 'true':
23             token_fetch_options = self._parse_json(video_data.get(
24                 'data-token-fetch-options', '{}'), video_id, fatal=False) or {}
25             token_fetch_url = token_fetch_options.get('url')
26             if token_fetch_url:
27                 embed_token = self._download_webpage(urljoin(
28                     url, token_fetch_url), video_id, fatal=False)
29                 if embed_token:
30                     video_url = smuggle_url(
31                         video_url, {'embed_token': embed_token.strip('"')})
32
33         return {
34             '_type': 'url_transparent',
35             'id': video_id,
36             'url': video_url,
37             'title': self._og_search_title(webpage),
38             'description': strip_or_none(self._og_search_description(webpage)),
39             'ie_key': 'Ooyala',
40         }
41
42
43 class SkySportsIE(SkyBaseIE):
44     _VALID_URL = r'https?://(?:www\.)?skysports\.com/watch/video/(?P<id>[0-9]+)'
45     _TEST = {
46         'url': 'http://www.skysports.com/watch/video/10328419/bale-its-our-time-to-shine',
47         'md5': '77d59166cddc8d3cb7b13e35eaf0f5ec',
48         'info_dict': {
49             'id': 'o3eWJnNDE6l7kfNO8BOoBlRxXRQ4ANNQ',
50             'ext': 'mp4',
51             'title': 'Bale: It\'s our time to shine',
52             'description': 'md5:e88bda94ae15f7720c5cb467e777bb6d',
53         },
54         'add_ie': ['Ooyala'],
55     }
56
57
58 class SkyNewsIE(SkyBaseIE):
59     _VALID_URL = r'https?://news\.sky\.com/video/[0-9a-z-]+-(?P<id>[0-9]+)'
60     _TEST = {
61         'url': 'https://news.sky.com/video/russian-plane-inspected-after-deadly-fire-11712962',
62         'md5': 'd6327e581473cea9976a3236ded370cd',
63         'info_dict': {
64             'id': '1ua21xaDE6lCtZDmbYfl8kwsKLooJbNM',
65             'ext': 'mp4',
66             'title': 'Russian plane inspected after deadly fire',
67             'description': 'The Russian Investigative Committee has released video of the wreckage of a passenger plane which caught fire near Moscow.',
68         },
69         'add_ie': ['Ooyala'],
70     }