[myvi:embed] Add extractor (closes #15521)
[youtube-dl] / youtube_dl / extractor / myvi.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5
6 from .common import InfoExtractor
7 from .vimple import SprutoBaseIE
8
9
10 class MyviIE(SprutoBaseIE):
11     _VALID_URL = r'''(?x)
12                         (?:
13                             https?://
14                                 (?:www\.)?
15                                 myvi\.(?:ru/player|tv)/
16                                 (?:
17                                     (?:
18                                         embed/html|
19                                         flash|
20                                         api/Video/Get
21                                     )/|
22                                     content/preloader\.swf\?.*\bid=
23                                 )|
24                             myvi:
25                         )
26                         (?P<id>[\da-zA-Z_-]+)
27                     '''
28     _TESTS = [{
29         'url': 'http://myvi.ru/player/embed/html/oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wObeRTZaCATzucDQIDph8hQU0',
30         'md5': '571bbdfba9f9ed229dc6d34cc0f335bf',
31         'info_dict': {
32             'id': 'f16b2bbd-cde8-481c-a981-7cd48605df43',
33             'ext': 'mp4',
34             'title': 'хозяин жизни',
35             'thumbnail': r're:^https?://.*\.jpg$',
36             'duration': 25,
37         },
38     }, {
39         'url': 'http://myvi.ru/player/content/preloader.swf?id=oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wOYf1WFpPfc_bWTKGVf_Zafr0',
40         'only_matching': True,
41     }, {
42         'url': 'http://myvi.ru/player/api/Video/Get/oOy4euHA6LVwNNAjhD9_Jq5Ha2Qf0rtVMVFMAZav8wObeRTZaCATzucDQIDph8hQU0',
43         'only_matching': True,
44     }, {
45         'url': 'http://myvi.tv/embed/html/oTGTNWdyz4Zwy_u1nraolwZ1odenTd9WkTnRfIL9y8VOgHYqOHApE575x4_xxS9Vn0?ap=0',
46         'only_matching': True,
47     }, {
48         'url': 'http://myvi.ru/player/flash/ocp2qZrHI-eZnHKQBK4cZV60hslH8LALnk0uBfKsB-Q4WnY26SeGoYPi8HWHxu0O30',
49         'only_matching': True,
50     }, {
51         'url': 'myvi:YwbqszQynUaHPn_s82sx0Q2',
52         'only_matching': True,
53     }]
54
55     @classmethod
56     def _extract_url(cls, webpage):
57         mobj = re.search(
58             r'<iframe[^>]+?src=(["\'])(?P<url>(?:https?:)?//myvi\.(?:ru/player|tv)/(?:embed/html|flash)/[^"]+)\1', webpage)
59         if mobj:
60             return mobj.group('url')
61
62     def _real_extract(self, url):
63         video_id = self._match_id(url)
64
65         spruto = self._download_json(
66             'http://myvi.ru/player/api/Video/Get/%s?sig' % video_id, video_id)['sprutoData']
67
68         return self._extract_spruto(spruto, video_id)
69
70
71 class MyviEmbedIE(InfoExtractor):
72     _VALID_URL = r'https?://(?:www\.)?myvi\.tv/(?:[^?]+\?.*?\bv=|embed/)(?P<id>[\da-z]+)'
73     _TESTS = [{
74         'url': 'https://www.myvi.tv/embed/ccdqic3wgkqwpb36x9sxg43t4r',
75         'info_dict': {
76             'id': 'b3ea0663-3234-469d-873e-7fecf36b31d1',
77             'ext': 'mp4',
78             'title': 'Твоя (original song).mp4',
79             'thumbnail': r're:^https?://.*\.jpg$',
80             'duration': 277,
81         },
82         'params': {
83             'skip_download': True,
84         },
85     }, {
86         'url': 'https://www.myvi.tv/idmi6o?v=ccdqic3wgkqwpb36x9sxg43t4r#watch',
87         'only_matching': True,
88     }]
89
90     @classmethod
91     def suitable(cls, url):
92         return False if MyviIE.suitable(url) else super(MyviEmbedIE, cls).suitable(url)
93
94     def _real_extract(self, url):
95         video_id = self._match_id(url)
96
97         webpage = self._download_webpage(
98             'https://www.myvi.tv/embed/%s' % video_id, video_id)
99
100         myvi_id = self._search_regex(
101             r'CreatePlayer\s*\(\s*["\'].*?\bv=([\da-zA-Z_]+)',
102             webpage, 'video id')
103
104         return self.url_result('myvi:%s' % myvi_id, ie=MyviIE.ie_key())