yt_dlp/extractor/markiza.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     orderedSet,
   6     parse_duration,
   7     try_get,
   8 )
   9
  10
  11 class MarkizaIE(InfoExtractor):
  12     _WORKING = False
  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'], str),
  63             })
  64         else:
  65             info['duration'] = parse_duration(
  66                 try_get(data, lambda x: x['details']['duration'], str))
  67         return info
  68
  69
  70 class MarkizaPageIE(InfoExtractor):
  71     _WORKING = False
  72     _VALID_URL = r'https?://(?:www\.)?(?:(?:[^/]+\.)?markiza|tvnoviny)\.sk/(?:[^/]+/)*(?P<id>\d+)_'
  73     _TESTS = [{
  74         'url': 'http://www.markiza.sk/soubiz/zahranicny/1923705_oteckovia-maju-svoj-den-ti-slavni-nie-su-o-nic-menej-rozkosni',
  75         'md5': 'ada4e9fad038abeed971843aa028c7b0',
  76         'info_dict': {
  77             'id': '139355',
  78             'ext': 'mp4',
  79             'title': 'Oteckovia 110',
  80             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  81             'thumbnail': r're:^https?://.*\.jpg$',
  82             'duration': 2604,
  83         },
  84         'params': {
  85             'skip_download': True,
  86         },
  87     }, {
  88         'url': 'http://dajto.markiza.sk/filmy-a-serialy/1774695_frajeri-vo-vegas',
  89         'only_matching': True,
  90     }, {
  91         'url': 'http://superstar.markiza.sk/aktualne/1923870_to-je-ale-telo-spevacka-ukazala-sexy-postavicku-v-bikinach',
  92         'only_matching': True,
  93     }, {
  94         'url': 'http://hybsa.markiza.sk/aktualne/1923790_uzasna-atmosfera-na-hybsa-v-poprade-superstaristi-si-prve-koncerty-pred-davom-ludi-poriadne-uzili',
  95         'only_matching': True,
  96     }, {
  97         'url': 'http://doma.markiza.sk/filmy/1885250_moja-vysnivana-svadba',
  98         'only_matching': True,
  99     }, {
 100         'url': 'http://www.tvnoviny.sk/domace/1923887_po-smrti-manzela-ju-cakalo-poriadne-prekvapenie',
 101         'only_matching': True,
 102     }]
 103
 104     @classmethod
 105     def suitable(cls, url):
 106         return False if MarkizaIE.suitable(url) else super().suitable(url)
 107
 108     def _real_extract(self, url):
 109         playlist_id = self._match_id(url)
 110
 111         webpage = self._download_webpage(
 112             # Downloading for some hosts (e.g. dajto, doma) fails with 500
 113             # although everything seems to be OK, so considering 500
 114             # status code to be expected.
 115             url, playlist_id, expected_status=500)
 116
 117         entries = [
 118             self.url_result(f'http://videoarchiv.markiza.sk/video/{video_id}')
 119             for video_id in orderedSet(re.findall(
 120                 r'(?:initPlayer_|data-entity=["\']|id=["\']player_)(\d+)',
 121                 webpage))]
 122
 123         return self.playlist_result(entries, playlist_id)