2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..utils import merge_dicts
10 class MallTVIE(InfoExtractor):
11 _VALID_URL = r'https?://(?:(?:www|sk)\.)?mall\.tv/(?:[^/]+/)*(?P<id>[^/?#&]+)'
13 'url': 'https://www.mall.tv/18-miliard-pro-neziskovky-opravdu-jsou-sportovci-nebo-clovek-v-tisni-pijavice',
14 'md5': '1c4a37f080e1f3023103a7b43458e518',
17 'display_id': '18-miliard-pro-neziskovky-opravdu-jsou-sportovci-nebo-clovek-v-tisni-pijavice',
19 'title': '18 miliard pro neziskovky. Opravdu jsou sportovci nebo Člověk v tísni pijavice?',
20 'description': 'md5:25fc0ec42a72ba602b602c683fa29deb',
22 'timestamp': 1538870400,
23 'upload_date': '20181007',
27 'url': 'https://www.mall.tv/kdo-to-plati/18-miliard-pro-neziskovky-opravdu-jsou-sportovci-nebo-clovek-v-tisni-pijavice',
28 'only_matching': True,
30 'url': 'https://sk.mall.tv/gejmhaus/reklamacia-nehreje-vyrobnik-tepla-alebo-spekacka',
31 'only_matching': True,
34 def _real_extract(self, url):
35 display_id = self._match_id(url)
37 webpage = self._download_webpage(
38 url, display_id, headers=self.geo_verification_headers())
40 SOURCE_RE = r'(<source[^>]+\bsrc=(?:(["\'])(?:(?!\2).)+|[^\s]+)/(?P<id>[\da-z]+)/index)\b'
41 video_id = self._search_regex(
42 SOURCE_RE, webpage, 'video id', group='id')
44 media = self._parse_html5_media_entries(
45 url, re.sub(SOURCE_RE, r'\1.m3u8', webpage), video_id,
46 m3u8_id='hls', m3u8_entry_protocol='m3u8_native')[0]
48 info = self._search_json_ld(webpage, video_id, default={})
50 return merge_dicts(media, info, {
52 'display_id': display_id,
53 'title': self._og_search_title(webpage, default=None) or display_id,
54 'description': self._og_search_description(webpage, default=None),
55 'thumbnail': self._og_search_thumbnail(webpage, default=None),