1 from __future__ import unicode_literals
5 from .common import InfoExtractor
6 from .kaltura import KalturaIE
13 class AZMedienBaseIE(InfoExtractor):
14 def _kaltura_video(self, partner_id, entry_id):
15 return self.url_result(
16 'kaltura:%s:%s' % (partner_id, entry_id), ie=KalturaIE.ie_key(),
20 class AZMedienIE(AZMedienBaseIE):
21 IE_DESC = 'AZ Medien videos'
32 /[0-9]+-episode-[^/\#]+
34 /[0-9]+-segment-(?:[^/\#]+\#)?|
44 'url': 'http://www.telezueri.ch/62-show-zuerinews/13772-episode-sonntag-18-dezember-2016/32419-segment-massenabweisungen-beim-hiltl-club-wegen-pelzboom',
48 'title': 'Massenabweisungen beim Hiltl Club wegen Pelzboom',
49 'description': 'md5:9ea9dd1b159ad65b36ddcf7f0d7c76a8',
50 'uploader_id': 'TeleZ?ri',
51 'upload_date': '20161218',
52 'timestamp': 1482084490,
55 'skip_download': True,
58 # URL with 'segment' and fragment:
59 'url': 'http://www.telebaern.tv/118-show-news/14240-episode-dienstag-17-januar-2017/33666-segment-achtung-gefahr#zu-wenig-pflegerinnen-und-pfleger',
62 # URL with 'episode' and fragment:
63 'url': 'http://www.telem1.ch/47-show-sonntalk/13986-episode-soldaten-fuer-grenzschutz-energiestrategie-obama-bilanz#soldaten-fuer-grenzschutz-energiestrategie-obama-bilanz',
66 # URL with 'show' and fragment:
67 'url': 'http://www.telezueri.ch/66-show-sonntalk#burka-plakate-trump-putin-china-besuch',
71 def _real_extract(self, url):
72 video_id = self._match_id(url)
74 webpage = self._download_webpage(url, video_id)
76 partner_id = self._search_regex(
77 r'<script[^>]+src=["\'](?:https?:)?//(?:[^/]+\.)?kaltura\.com(?:/[^/]+)*/(?:p|partner_id)/([0-9]+)',
78 webpage, 'kaltura partner id')
79 entry_id = self._html_search_regex(
80 r'<a[^>]+data-id=(["\'])(?P<id>(?:(?!\1).)+)\1[^>]+data-slug=["\']%s'
81 % re.escape(video_id), webpage, 'kaltura entry id', group='id')
83 return self._kaltura_video(partner_id, entry_id)
86 class AZMedienShowIE(AZMedienBaseIE):
87 IE_DESC = 'AZ Medien shows'
88 _VALID_URL = r'https?://(?:www\.)?(?:telezueri\.ch|telebaern\.tv|telem1\.ch)/(?P<id>[0-9]+-show-[^/#]+(?:/[0-9]+-episode-[^/#]+)?)$'
92 'url': 'http://www.telebaern.tv/118-show-news/13735-episode-donnerstag-15-dezember-2016',
94 'id': '118-show-news/13735-episode-donnerstag-15-dezember-2016',
95 'title': 'News - Donnerstag, 15. Dezember 2016',
99 # URL with 'show' only
100 'url': 'http://www.telezueri.ch/86-show-talktaeglich',
101 'only_matching': True
104 def _real_extract(self, url):
105 show_id = self._match_id(url)
106 webpage = self._download_webpage(url, show_id)
110 partner_id = self._search_regex(
111 r'src=["\'](?:https?:)?//(?:[^/]+\.)kaltura\.com/(?:[^/]+/)*(?:p|partner_id)/(\d+)',
112 webpage, 'kaltura partner id', default=None)
116 self._kaltura_video(partner_id, m.group('id'))
117 for m in re.finditer(
118 r'data-id=(["\'])(?P<id>(?:(?!\1).)+)\1', webpage)]
122 self.url_result(m.group('url'), ie=AZMedienIE.ie_key())
123 for m in re.finditer(
124 r'<a[^>]+data-real=(["\'])(?P<url>http.+?)\1', webpage)]
126 title = self._search_regex(
127 r'episodeShareTitle\s*=\s*(["\'])(?P<title>(?:(?!\1).)+)\1',
129 default=strip_or_none(get_element_by_class(
130 'title-block-cell', webpage)), group='title')
132 return self.playlist_result(entries, show_id, title)