youtube_dl/extractor/ndr.py

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     ExtractorError,
   7     int_or_none,
   8 )
   9
  10
  11 preferences = {'xl': 4, 'l': 3, 'm': 2, 's': 1, 'xs': 0,}
  12
  13
  14 class NDRBaseIE(InfoExtractor):
  15     def _real_extract(self, url):
  16         video_id = self._match_id(url)
  17
  18         json_data = self._download_json('http://www.ndr.de/%s-ppjson.json' % video_id, video_id, 'Downloading page')
  19
  20         formats = []
  21         objetType = json_data.get('config').get('objectType')
  22         if objetType == 'video':
  23             for key, f in json_data.get('playlist').items():
  24                 if key != 'config':
  25                     src = f['src']
  26                     if '.f4m' in src:
  27                         formats.extend(self._extract_f4m_formats(src, video_id))
  28                     elif '.m3u8' in src:
  29                         formats.extend(self._extract_m3u8_formats(src, video_id))
  30                     else:
  31                         quality = f.get('quality')
  32                         formats.append({
  33                             'url': src,
  34                             'format_id': quality,
  35                             'preference': preferences.get(quality),
  36                         })
  37         elif objetType == 'audio':
  38             for key, f in json_data.get('playlist').items():
  39                 if key != 'config':
  40                     formats.append({
  41                         'url': f['src'],
  42                         'format_id': 'mp3',
  43
  44                     })
  45         else:
  46             raise ExtractorError('No media links available for %s' % video_id)
  47
  48         self._sort_formats(formats)
  49
  50         config = json_data.get('playlist').get('config')
  51
  52         title = config['title']
  53         duration = int_or_none(config.get('duration'))
  54         thumbnails = [{
  55             'id': thumbnail.get('quality'),
  56             'url': thumbnail.get('src'),
  57             'preference': preferences.get(thumbnail.get('quality'))
  58         } for thumbnail in config.get('poster').values()]
  59
  60         return {
  61             'id': video_id,
  62             'title': title,
  63             'thumbnails': thumbnails,
  64             'duration': duration,
  65             'formats': formats,
  66         }
  67
  68
  69 class NDRIE(NDRBaseIE):
  70     IE_NAME = 'ndr'
  71     IE_DESC = 'NDR.de - Mediathek'
  72     _VALID_URL = r'https?://www\.ndr\.de/.+?,(?P<id>\w+)\.html'
  73
  74     _TESTS = [
  75         {
  76             'url': 'http://www.ndr.de/fernsehen/sendungen/nordmagazin/Kartoffeltage-in-der-Lewitz,nordmagazin25866.html',
  77             'md5': '5bc5f5b92c82c0f8b26cddca34f8bb2c',
  78             'note': 'Video file',
  79             'info_dict': {
  80                 'id': 'nordmagazin25866',
  81                 'ext': 'mp4',
  82                 'title': 'Kartoffeltage in der Lewitz',
  83                 'duration': 166,
  84             },
  85             'skip': '404 Not found',
  86         },
  87         {
  88             'url': 'http://www.ndr.de/fernsehen/Party-Poette-und-Parade,hafengeburtstag988.html',
  89             'md5': 'dadc003c55ae12a5d2f6bd436cd73f59',
  90             'info_dict': {
  91                 'id': 'hafengeburtstag988',
  92                 'ext': 'mp4',
  93                 'title': 'Party, Pötte und Parade',
  94                 'duration': 3498,
  95             },
  96         },
  97         {
  98             'url': 'http://www.ndr.de/info/La-Valette-entgeht-der-Hinrichtung,audio51535.html',
  99             'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
 100             'note': 'Audio file',
 101             'info_dict': {
 102                 'id': 'audio51535',
 103                 'ext': 'mp3',
 104                 'title': 'La Valette entgeht der Hinrichtung',
 105                 'duration': 884,
 106             }
 107         }
 108     ]
 109
 110
 111 class NJoyIE(NDRBaseIE):
 112     IE_NAME = 'N-JOY'
 113     _VALID_URL = r'https?://www\.n-joy\.de/.+?,(?P<id>\w+)\.html'
 114
 115     _TEST = {
 116         'url': 'http://www.n-joy.de/entertainment/comedy/comedy_contest/Benaissa-beim-NDR-Comedy-Contest,comedycontest2480.html',
 117         'md5': 'cb63be60cd6f9dd75218803146d8dc67',
 118         'info_dict': {
 119             'id': '2480',
 120             'ext': 'mp4',
 121             'title': 'Benaissa beim NDR Comedy Contest',
 122             'duration': 654,
 123         }
 124     }