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

   1 from __future__ import unicode_literals
   2
   3 from .canvas import CanvasIE
   4 from .common import InfoExtractor
   5
   6
   7 class KetnetIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?ketnet\.be/(?:[^/]+/)*(?P<id>[^/?#&]+)'
   9     _TESTS = [{
  10         'url': 'https://www.ketnet.be/kijken/zomerse-filmpjes',
  11         'md5': '6bdeb65998930251bbd1c510750edba9',
  12         'info_dict': {
  13             'id': 'zomerse-filmpjes',
  14             'ext': 'mp4',
  15             'title': 'Gluur mee op de filmset en op Pennenzakkenrock',
  16             'description': 'Gluur mee met Ghost Rockers op de filmset',
  17             'thumbnail': r're:^https?://.*\.jpg$',
  18         }
  19     }, {
  20         # mzid in playerConfig instead of sources
  21         'url': 'https://www.ketnet.be/kijken/nachtwacht/de-greystook',
  22         'md5': '90139b746a0a9bd7bb631283f6e2a64e',
  23         'info_dict': {
  24             'id': 'md-ast-4ac54990-ce66-4d00-a8ca-9eac86f4c475',
  25             'display_id': 'md-ast-4ac54990-ce66-4d00-a8ca-9eac86f4c475',
  26             'ext': 'flv',
  27             'title': 'Nachtwacht: De Greystook',
  28             'description': 'md5:1db3f5dc4c7109c821261e7512975be7',
  29             'thumbnail': r're:^https?://.*\.jpg$',
  30             'duration': 1468.03,
  31         },
  32         'expected_warnings': ['is not a supported codec', 'Unknown MIME type'],
  33     }, {
  34         'url': 'https://www.ketnet.be/kijken/karrewiet/uitzending-8-september-2016',
  35         'only_matching': True,
  36     }, {
  37         'url': 'https://www.ketnet.be/achter-de-schermen/sien-repeteert-voor-stars-for-life',
  38         'only_matching': True,
  39     }, {
  40         # mzsource, geo restricted to Belgium
  41         'url': 'https://www.ketnet.be/kijken/nachtwacht/de-bermadoe',
  42         'only_matching': True,
  43     }]
  44
  45     def _real_extract(self, url):
  46         video_id = self._match_id(url)
  47
  48         webpage = self._download_webpage(url, video_id)
  49
  50         config = self._parse_json(
  51             self._search_regex(
  52                 r'(?s)playerConfig\s*=\s*({.+?})\s*;', webpage,
  53                 'player config'),
  54             video_id)
  55
  56         mzid = config.get('mzid')
  57         if mzid:
  58             return self.url_result(
  59                 'https://mediazone.vrt.be/api/v1/ketnet/assets/%s' % mzid,
  60                 CanvasIE.ie_key(), video_id=mzid)
  61
  62         title = config['title']
  63
  64         formats = []
  65         for source_key in ('', 'mz'):
  66             source = config.get('%ssource' % source_key)
  67             if not isinstance(source, dict):
  68                 continue
  69             for format_id, format_url in source.items():
  70                 if format_id == 'hls':
  71                     formats.extend(self._extract_m3u8_formats(
  72                         format_url, video_id, 'mp4',
  73                         entry_protocol='m3u8_native', m3u8_id=format_id,
  74                         fatal=False))
  75                 elif format_id == 'hds':
  76                     formats.extend(self._extract_f4m_formats(
  77                         format_url, video_id, f4m_id=format_id, fatal=False))
  78                 else:
  79                     formats.append({
  80                         'url': format_url,
  81                         'format_id': format_id,
  82                     })
  83         self._sort_formats(formats)
  84
  85         return {
  86             'id': video_id,
  87             'title': title,
  88             'description': config.get('description'),
  89             'thumbnail': config.get('image'),
  90             'series': config.get('program'),
  91             'episode': config.get('episode'),
  92             'formats': formats,
  93         }