[cwtv] fix extraction(closes #17256)
[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     int_or_none,
7     parse_age_limit,
8     parse_iso8601,
9     smuggle_url,
10     str_or_none,
11 )
12
13
14 class CWTVIE(InfoExtractor):
15     _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})'
16     _TESTS = [{
17         'url': 'http://cwtv.com/shows/arrow/legends-of-yesterday/?play=6b15e985-9345-4f60-baf8-56e96be57c63',
18         'info_dict': {
19             'id': '6b15e985-9345-4f60-baf8-56e96be57c63',
20             'ext': 'mp4',
21             'title': 'Legends of Yesterday',
22             '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.',
23             'duration': 2665,
24             'series': 'Arrow',
25             'season_number': 4,
26             'season': '4',
27             'episode_number': 8,
28             'upload_date': '20151203',
29             'timestamp': 1449122100,
30         },
31         'params': {
32             # m3u8 download
33             'skip_download': True,
34         },
35         'skip': 'redirect to http://cwtv.com/shows/arrow/',
36     }, {
37         'url': 'http://www.cwseed.com/shows/whose-line-is-it-anyway/jeff-davis-4/?play=24282b12-ead2-42f2-95ad-26770c2c6088',
38         'info_dict': {
39             'id': '24282b12-ead2-42f2-95ad-26770c2c6088',
40             'ext': 'mp4',
41             'title': 'Jeff Davis 4',
42             'description': 'Jeff Davis is back to make you laugh.',
43             'duration': 1263,
44             'series': 'Whose Line Is It Anyway?',
45             'season_number': 11,
46             'episode_number': 20,
47             'upload_date': '20151006',
48             'timestamp': 1444107300,
49             'age_limit': 14,
50             'uploader': 'CWTV',
51         },
52         'params': {
53             # m3u8 download
54             'skip_download': True,
55         },
56     }, {
57         'url': 'http://cwtv.com/thecw/chroniclesofcisco/?play=8adebe35-f447-465f-ab52-e863506ff6d6',
58         'only_matching': True,
59     }, {
60         'url': 'http://cwtvpr.com/the-cw/video?watch=9eee3f60-ef4e-440b-b3b2-49428ac9c54e',
61         'only_matching': True,
62     }, {
63         'url': 'http://cwtv.com/shows/arrow/legends-of-yesterday/?watch=6b15e985-9345-4f60-baf8-56e96be57c63',
64         'only_matching': True,
65     }]
66
67     def _real_extract(self, url):
68         video_id = self._match_id(url)
69         video_data = self._download_json(
70             'http://images.cwtv.com/feed/mobileapp/video-meta/apiversion_8/guid_' + video_id,
71             video_id)['video']
72         title = video_data['title']
73         mpx_url = video_data.get('mpx_url') or 'http://link.theplatform.com/s/cwtv/media/guid/2703454149/%s?formats=M3U' % video_id
74
75         season = str_or_none(video_data.get('season'))
76         episode = str_or_none(video_data.get('episode'))
77         if episode and season:
78             episode = episode.lstrip(season)
79
80         return {
81             '_type': 'url_transparent',
82             'id': video_id,
83             'title': title,
84             'url': smuggle_url(mpx_url, {'force_smil_url': True}),
85             'description': video_data.get('description_long'),
86             'duration': int_or_none(video_data.get('duration_secs')),
87             'series': video_data.get('series_name'),
88             'season_number': int_or_none(season),
89             'episode_number': int_or_none(episode),
90             'timestamp': parse_iso8601(video_data.get('start_time')),
91             'age_limit': parse_age_limit(video_data.get('rating')),
92             'ie_key': 'ThePlatform',
93         }