[youtube] Fix extraction.
[youtube-dl] / youtube_dl / extractor / cwtv.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..utils import (
6     ExtractorError,
7     int_or_none,
8     parse_age_limit,
9     parse_iso8601,
10     smuggle_url,
11     str_or_none,
12 )
13
14
15 class CWTVIE(InfoExtractor):
16     _VALID_URL = r'https?://(?:www\.)?cw(?:tv(?:pr)?|seed)\.com/(?:shows/)?(?:[^/]+/)+[^?]*\?.*\b(?:play|watch)=(?P<id>[a-z0-9]{8}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{12})'
17     _TESTS = [{
18         'url': 'http://cwtv.com/shows/arrow/legends-of-yesterday/?play=6b15e985-9345-4f60-baf8-56e96be57c63',
19         'info_dict': {
20             'id': '6b15e985-9345-4f60-baf8-56e96be57c63',
21             'ext': 'mp4',
22             'title': 'Legends of Yesterday',
23             'description': 'Oliver and Barry Allen take Kendra Saunders and Carter Hall to a remote location to keep them hidden from Vandal Savage while they figure out how to defeat him.',
24             'duration': 2665,
25             'series': 'Arrow',
26             'season_number': 4,
27             'season': '4',
28             'episode_number': 8,
29             'upload_date': '20151203',
30             'timestamp': 1449122100,
31         },
32         'params': {
33             # m3u8 download
34             'skip_download': True,
35         },
36         'skip': 'redirect to http://cwtv.com/shows/arrow/',
37     }, {
38         'url': 'http://www.cwseed.com/shows/whose-line-is-it-anyway/jeff-davis-4/?play=24282b12-ead2-42f2-95ad-26770c2c6088',
39         'info_dict': {
40             'id': '24282b12-ead2-42f2-95ad-26770c2c6088',
41             'ext': 'mp4',
42             'title': 'Jeff Davis 4',
43             'description': 'Jeff Davis is back to make you laugh.',
44             'duration': 1263,
45             'series': 'Whose Line Is It Anyway?',
46             'season_number': 11,
47             'episode_number': 20,
48             'upload_date': '20151006',
49             'timestamp': 1444107300,
50             'age_limit': 14,
51             'uploader': 'CWTV',
52         },
53         'params': {
54             # m3u8 download
55             'skip_download': True,
56         },
57     }, {
58         'url': 'http://cwtv.com/thecw/chroniclesofcisco/?play=8adebe35-f447-465f-ab52-e863506ff6d6',
59         'only_matching': True,
60     }, {
61         'url': 'http://cwtvpr.com/the-cw/video?watch=9eee3f60-ef4e-440b-b3b2-49428ac9c54e',
62         'only_matching': True,
63     }, {
64         'url': 'http://cwtv.com/shows/arrow/legends-of-yesterday/?watch=6b15e985-9345-4f60-baf8-56e96be57c63',
65         'only_matching': True,
66     }]
67
68     def _real_extract(self, url):
69         video_id = self._match_id(url)
70         data = self._download_json(
71             'http://images.cwtv.com/feed/mobileapp/video-meta/apiversion_8/guid_' + video_id,
72             video_id)
73         if data.get('result') != 'ok':
74             raise ExtractorError(data['msg'], expected=True)
75         video_data = data['video']
76         title = video_data['title']
77         mpx_url = video_data.get('mpx_url') or 'http://link.theplatform.com/s/cwtv/media/guid/2703454149/%s?formats=M3U' % video_id
78
79         season = str_or_none(video_data.get('season'))
80         episode = str_or_none(video_data.get('episode'))
81         if episode and season:
82             episode = episode[len(season):]
83
84         return {
85             '_type': 'url_transparent',
86             'id': video_id,
87             'title': title,
88             'url': smuggle_url(mpx_url, {'force_smil_url': True}),
89             'description': video_data.get('description_long'),
90             'duration': int_or_none(video_data.get('duration_secs')),
91             'series': video_data.get('series_name'),
92             'season_number': int_or_none(season),
93             'episode_number': int_or_none(episode),
94             'timestamp': parse_iso8601(video_data.get('start_time')),
95             'age_limit': parse_age_limit(video_data.get('rating')),
96             'ie_key': 'ThePlatform',
97         }