Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/servus.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7
   8
   9 class ServusIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?servus\.com/(?:(?:at|de)/p/[^/]+|tv/videos)/(?P<id>[aA]{2}-\w+|\d+-\d+)'
  11     _TESTS = [{
  12         'url': 'https://www.servus.com/de/p/Die-Gr%C3%BCnen-aus-Sicht-des-Volkes/AA-1T6VBU5PW1W12/',
  13         'md5': '3e1dd16775aa8d5cbef23628cfffc1f4',
  14         'info_dict': {
  15             'id': 'AA-1T6VBU5PW1W12',
  16             'ext': 'mp4',
  17             'title': 'Die Grünen aus Sicht des Volkes',
  18             'description': 'md5:1247204d85783afe3682644398ff2ec4',
  19             'thumbnail': r're:^https?://.*\.jpg',
  20         }
  21     }, {
  22         'url': 'https://www.servus.com/at/p/Wie-das-Leben-beginnt/1309984137314-381415152/',
  23         'only_matching': True,
  24     }, {
  25         'url': 'https://www.servus.com/tv/videos/aa-1t6vbu5pw1w12/',
  26         'only_matching': True,
  27     }, {
  28         'url': 'https://www.servus.com/tv/videos/1380889096408-1235196658/',
  29         'only_matching': True,
  30     }]
  31
  32     def _real_extract(self, url):
  33         video_id = self._match_id(url).upper()
  34         webpage = self._download_webpage(url, video_id)
  35
  36         title = self._search_regex(
  37             (r'videoLabel\s*=\s*(["\'])(?P<title>(?:(?!\1).)+)\1',
  38              r'<h\d+[^>]+\bclass=["\']heading--(?:one|two)["\'][^>]*>(?P<title>[^<]+)'),
  39             webpage, 'title', default=None,
  40             group='title') or self._og_search_title(webpage)
  41         title = re.sub(r'\s*-\s*Servus TV\s*$', '', title)
  42         description = self._og_search_description(webpage)
  43         thumbnail = self._og_search_thumbnail(webpage)
  44
  45         formats = self._extract_m3u8_formats(
  46             'https://stv.rbmbtnx.net/api/v1/manifests/%s.m3u8' % video_id,
  47             video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id='hls')
  48         self._sort_formats(formats)
  49
  50         return {
  51             'id': video_id,
  52             'title': title,
  53             'description': description,
  54             'thumbnail': thumbnail,
  55             'formats': formats,
  56         }