]>
Commit | Line | Data |
---|---|---|
1 | import re | |
2 | ||
3 | from .common import InfoExtractor | |
4 | from ..compat import compat_str | |
5 | from ..utils import ( | |
6 | orderedSet, | |
7 | parse_duration, | |
8 | try_get, | |
9 | ) | |
10 | ||
11 | ||
12 | class MarkizaIE(InfoExtractor): | |
13 | _VALID_URL = r'https?://(?:www\.)?videoarchiv\.markiza\.sk/(?:video/(?:[^/]+/)*|embed/)(?P<id>\d+)(?:[_/]|$)' | |
14 | _TESTS = [{ | |
15 | 'url': 'http://videoarchiv.markiza.sk/video/oteckovia/84723_oteckovia-109', | |
16 | 'md5': 'ada4e9fad038abeed971843aa028c7b0', | |
17 | 'info_dict': { | |
18 | 'id': '139078', | |
19 | 'ext': 'mp4', | |
20 | 'title': 'Oteckovia 109', | |
21 | 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e', | |
22 | 'thumbnail': r're:^https?://.*\.jpg$', | |
23 | 'duration': 2760, | |
24 | }, | |
25 | }, { | |
26 | 'url': 'http://videoarchiv.markiza.sk/video/televizne-noviny/televizne-noviny/85430_televizne-noviny', | |
27 | 'info_dict': { | |
28 | 'id': '85430', | |
29 | 'title': 'Televízne noviny', | |
30 | }, | |
31 | 'playlist_count': 23, | |
32 | }, { | |
33 | 'url': 'http://videoarchiv.markiza.sk/video/oteckovia/84723', | |
34 | 'only_matching': True, | |
35 | }, { | |
36 | 'url': 'http://videoarchiv.markiza.sk/video/84723', | |
37 | 'only_matching': True, | |
38 | }, { | |
39 | 'url': 'http://videoarchiv.markiza.sk/video/filmy/85190_kamenak', | |
40 | 'only_matching': True, | |
41 | }, { | |
42 | 'url': 'http://videoarchiv.markiza.sk/video/reflex/zo-zakulisia/84651_pribeh-alzbetky', | |
43 | 'only_matching': True, | |
44 | }, { | |
45 | 'url': 'http://videoarchiv.markiza.sk/embed/85295', | |
46 | 'only_matching': True, | |
47 | }] | |
48 | ||
49 | def _real_extract(self, url): | |
50 | video_id = self._match_id(url) | |
51 | ||
52 | data = self._download_json( | |
53 | 'http://videoarchiv.markiza.sk/json/video_jwplayer7.json', | |
54 | video_id, query={'id': video_id}) | |
55 | ||
56 | info = self._parse_jwplayer_data(data, m3u8_id='hls', mpd_id='dash') | |
57 | ||
58 | if info.get('_type') == 'playlist': | |
59 | info.update({ | |
60 | 'id': video_id, | |
61 | 'title': try_get( | |
62 | data, lambda x: x['details']['name'], compat_str), | |
63 | }) | |
64 | else: | |
65 | info['duration'] = parse_duration( | |
66 | try_get(data, lambda x: x['details']['duration'], compat_str)) | |
67 | return info | |
68 | ||
69 | ||
70 | class MarkizaPageIE(InfoExtractor): | |
71 | _VALID_URL = r'https?://(?:www\.)?(?:(?:[^/]+\.)?markiza|tvnoviny)\.sk/(?:[^/]+/)*(?P<id>\d+)_' | |
72 | _TESTS = [{ | |
73 | 'url': 'http://www.markiza.sk/soubiz/zahranicny/1923705_oteckovia-maju-svoj-den-ti-slavni-nie-su-o-nic-menej-rozkosni', | |
74 | 'md5': 'ada4e9fad038abeed971843aa028c7b0', | |
75 | 'info_dict': { | |
76 | 'id': '139355', | |
77 | 'ext': 'mp4', | |
78 | 'title': 'Oteckovia 110', | |
79 | 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e', | |
80 | 'thumbnail': r're:^https?://.*\.jpg$', | |
81 | 'duration': 2604, | |
82 | }, | |
83 | 'params': { | |
84 | 'skip_download': True, | |
85 | }, | |
86 | }, { | |
87 | 'url': 'http://dajto.markiza.sk/filmy-a-serialy/1774695_frajeri-vo-vegas', | |
88 | 'only_matching': True, | |
89 | }, { | |
90 | 'url': 'http://superstar.markiza.sk/aktualne/1923870_to-je-ale-telo-spevacka-ukazala-sexy-postavicku-v-bikinach', | |
91 | 'only_matching': True, | |
92 | }, { | |
93 | 'url': 'http://hybsa.markiza.sk/aktualne/1923790_uzasna-atmosfera-na-hybsa-v-poprade-superstaristi-si-prve-koncerty-pred-davom-ludi-poriadne-uzili', | |
94 | 'only_matching': True, | |
95 | }, { | |
96 | 'url': 'http://doma.markiza.sk/filmy/1885250_moja-vysnivana-svadba', | |
97 | 'only_matching': True, | |
98 | }, { | |
99 | 'url': 'http://www.tvnoviny.sk/domace/1923887_po-smrti-manzela-ju-cakalo-poriadne-prekvapenie', | |
100 | 'only_matching': True, | |
101 | }] | |
102 | ||
103 | @classmethod | |
104 | def suitable(cls, url): | |
105 | return False if MarkizaIE.suitable(url) else super(MarkizaPageIE, cls).suitable(url) | |
106 | ||
107 | def _real_extract(self, url): | |
108 | playlist_id = self._match_id(url) | |
109 | ||
110 | webpage = self._download_webpage( | |
111 | # Downloading for some hosts (e.g. dajto, doma) fails with 500 | |
112 | # although everything seems to be OK, so considering 500 | |
113 | # status code to be expected. | |
114 | url, playlist_id, expected_status=500) | |
115 | ||
116 | entries = [ | |
117 | self.url_result('http://videoarchiv.markiza.sk/video/%s' % video_id) | |
118 | for video_id in orderedSet(re.findall( | |
119 | r'(?:initPlayer_|data-entity=["\']|id=["\']player_)(\d+)', | |
120 | webpage))] | |
121 | ||
122 | return self.playlist_result(entries, playlist_id) |