]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/markiza.py
3 from .common
import InfoExtractor
4 from ..compat
import compat_str
12 class MarkizaIE(InfoExtractor
):
14 _VALID_URL
= r
'https?://(?:www\.)?videoarchiv\.markiza\.sk/(?:video/(?:[^/]+/)*|embed/)(?P<id>\d+)(?:[_/]|$)'
16 'url': 'http://videoarchiv.markiza.sk/video/oteckovia/84723_oteckovia-109',
17 'md5': 'ada4e9fad038abeed971843aa028c7b0',
21 'title': 'Oteckovia 109',
22 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
23 'thumbnail': r
're:^https?://.*\.jpg$',
27 'url': 'http://videoarchiv.markiza.sk/video/televizne-noviny/televizne-noviny/85430_televizne-noviny',
30 'title': 'Televízne noviny',
34 'url': 'http://videoarchiv.markiza.sk/video/oteckovia/84723',
35 'only_matching': True,
37 'url': 'http://videoarchiv.markiza.sk/video/84723',
38 'only_matching': True,
40 'url': 'http://videoarchiv.markiza.sk/video/filmy/85190_kamenak',
41 'only_matching': True,
43 'url': 'http://videoarchiv.markiza.sk/video/reflex/zo-zakulisia/84651_pribeh-alzbetky',
44 'only_matching': True,
46 'url': 'http://videoarchiv.markiza.sk/embed/85295',
47 'only_matching': True,
50 def _real_extract(self
, url
):
51 video_id
= self
._match
_id
(url
)
53 data
= self
._download
_json
(
54 'http://videoarchiv.markiza.sk/json/video_jwplayer7.json',
55 video_id
, query
={'id': video_id}
)
57 info
= self
._parse
_jwplayer
_data
(data
, m3u8_id
='hls', mpd_id
='dash')
59 if info
.get('_type') == 'playlist':
63 data
, lambda x
: x
['details']['name'], compat_str
),
66 info
['duration'] = parse_duration(
67 try_get(data
, lambda x
: x
['details']['duration'], compat_str
))
71 class MarkizaPageIE(InfoExtractor
):
73 _VALID_URL
= r
'https?://(?:www\.)?(?:(?:[^/]+\.)?markiza|tvnoviny)\.sk/(?:[^/]+/)*(?P<id>\d+)_'
75 'url': 'http://www.markiza.sk/soubiz/zahranicny/1923705_oteckovia-maju-svoj-den-ti-slavni-nie-su-o-nic-menej-rozkosni',
76 'md5': 'ada4e9fad038abeed971843aa028c7b0',
80 'title': 'Oteckovia 110',
81 'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
82 'thumbnail': r
're:^https?://.*\.jpg$',
86 'skip_download': True,
89 'url': 'http://dajto.markiza.sk/filmy-a-serialy/1774695_frajeri-vo-vegas',
90 'only_matching': True,
92 'url': 'http://superstar.markiza.sk/aktualne/1923870_to-je-ale-telo-spevacka-ukazala-sexy-postavicku-v-bikinach',
93 'only_matching': True,
95 'url': 'http://hybsa.markiza.sk/aktualne/1923790_uzasna-atmosfera-na-hybsa-v-poprade-superstaristi-si-prve-koncerty-pred-davom-ludi-poriadne-uzili',
96 'only_matching': True,
98 'url': 'http://doma.markiza.sk/filmy/1885250_moja-vysnivana-svadba',
99 'only_matching': True,
101 'url': 'http://www.tvnoviny.sk/domace/1923887_po-smrti-manzela-ju-cakalo-poriadne-prekvapenie',
102 'only_matching': True,
106 def suitable(cls
, url
):
107 return False if MarkizaIE
.suitable(url
) else super(MarkizaPageIE
, cls
).suitable(url
)
109 def _real_extract(self
, url
):
110 playlist_id
= self
._match
_id
(url
)
112 webpage
= self
._download
_webpage
(
113 # Downloading for some hosts (e.g. dajto, doma) fails with 500
114 # although everything seems to be OK, so considering 500
115 # status code to be expected.
116 url
, playlist_id
, expected_status
=500)
119 self
.url_result('http://videoarchiv.markiza.sk/video/%s' % video_id
)
120 for video_id
in orderedSet(re
.findall(
121 r
'(?:initPlayer_|data-entity=["\']|
id=["\']player_)(\d+)',
124 return self.playlist_result(entries, playlist_id)