_ Git - youtube-dl/blob - youtube_dl/extractor/azmedientv.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from .kaltura import KalturaIE
   7 from ..utils import get_element_by_class
   8
   9
  10 class AZMedienTVIE(InfoExtractor):
  11     IE_DESC = 'telezueri.ch, telebaern.tv and telem1.ch videos'
  12     _VALID_URL = r'http://(?:www\.)?(?:telezueri\.ch|telebaern\.tv|telem1\.ch)/[0-9]+-show-[^/#]+(?:/[0-9]+-episode-[^/#]+(?:/[0-9]+-segment-(?:[^/#]+#)?|#)|#)(?P<id>[^#]+)'
  13
  14     _TESTS = [{
  15         # URL with 'segment'
  16         'url': 'http://www.telezueri.ch/62-show-zuerinews/13772-episode-sonntag-18-dezember-2016/32419-segment-massenabweisungen-beim-hiltl-club-wegen-pelzboom',
  17         'md5': 'fda85ada1299cee517a622bfbc5f6b66',
  18         'info_dict': {
  19             'id': '1_2444peh4',
  20             'ext': 'mov',
  21             'title': 'Massenabweisungen beim Hiltl Club wegen Pelzboom',
  22             'description': 'md5:9ea9dd1b159ad65b36ddcf7f0d7c76a8',
  23             'uploader_id': 'TeleZ?ri',
  24             'upload_date': '20161218',
  25             'timestamp': 1482084490,
  26         }
  27     }, {
  28         # URL with 'segment' and fragment:
  29         'url': 'http://www.telebaern.tv/118-show-news/14240-episode-dienstag-17-januar-2017/33666-segment-achtung-gefahr#zu-wenig-pflegerinnen-und-pfleger',
  30         'only_matching': True
  31     }, {
  32         # URL with 'episode' and fragment:
  33         'url': 'http://www.telem1.ch/47-show-sonntalk/13986-episode-soldaten-fuer-grenzschutz-energiestrategie-obama-bilanz#soldaten-fuer-grenzschutz-energiestrategie-obama-bilanz',
  34         'only_matching': True
  35     }, {
  36         # URL with 'show' and fragment:
  37         'url': 'http://www.telezueri.ch/66-show-sonntalk#burka-plakate-trump-putin-china-besuch',
  38         'only_matching': True
  39     }]
  40
  41     def _real_extract(self, url):
  42         video_id = self._match_id(url)
  43         webpage = self._download_webpage(url, video_id)
  44
  45         kaltura_partner_id = self._html_search_regex(
  46             r'<script[^>]+src=["\']https?://www\.kaltura\.com/.*/partner_id/([0-9]+)',
  47             webpage, 'Kaltura partner ID')
  48         kaltura_entry_id = self._html_search_regex(
  49             r'<a[^>]+data-id=["\'](.*?)["\'][^>]+data-slug=["\']%s' % video_id,
  50             webpage, 'Kaltura entry ID')
  51
  52         return self.url_result(
  53             'kaltura:%s:%s' % (kaltura_partner_id, kaltura_entry_id),
  54             ie=KalturaIE.ie_key())
  55
  56
  57 class AZMedienTVShowIE(InfoExtractor):
  58     IE_DESC = 'telezueri.ch, telebaern.tv and telem1.ch shows'
  59     _VALID_URL = r'http://(?:www\.)?(?:telezueri\.ch|telebaern\.tv|telem1\.ch)/(?P<id>[0-9]+-show-[^/#]+(?:/[0-9]+-episode-[^/#]+)?)$'
  60
  61     _TESTS = [{
  62         # URL with 'episode':
  63         'url': 'http://www.telebaern.tv/118-show-news/13735-episode-donnerstag-15-dezember-2016',
  64         'info_dict': {
  65             'id': '118-show-news/13735-episode-donnerstag-15-dezember-2016',
  66             'title': 'News',
  67         },
  68         'playlist_count': 9,
  69     }, {
  70         # URL with 'show' only:
  71         'url': 'http://www.telezueri.ch/86-show-talktaeglich',
  72         'only_matching': True
  73     }]
  74
  75     def _real_extract(self, url):
  76         show_id = self._match_id(url)
  77         webpage = self._download_webpage(url, show_id)
  78
  79         title = get_element_by_class('title-block-cell', webpage)
  80         if title:
  81             title = title.strip()
  82
  83         entries = [self.url_result(m.group('url'), ie=AZMedienTVIE.ie_key()) for m in re.finditer(
  84             r'<a href=["\']#["\'][^>]+data-real=["\'](?P<url>.+?)["\']', webpage)]
  85
  86         return self.playlist_result(
  87             entries, show_id, title)