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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5 import base64
   6
   7 from .common import InfoExtractor
   8 from ..compat import (
   9     compat_urllib_parse,
  10     compat_urlparse,
  11 )
  12 from ..utils import (
  13     ExtractorError,
  14     int_or_none,
  15     unsmuggle_url,
  16 )
  17
  18
  19 class KalturaIE(InfoExtractor):
  20     _VALID_URL = r'''(?x)
  21                 (?:
  22                     kaltura:(?P<partner_id_s>\d+):(?P<id_s>[0-9a-z_]+)|
  23                     https?://
  24                         (:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/
  25                         (?:
  26                             (?:
  27                                 # flash player
  28                                 index\.php/kwidget/
  29                                 (?:[^/]+/)*?wid/_(?P<partner_id>\d+)/
  30                                 (?:[^/]+/)*?entry_id/(?P<id>[0-9a-z_]+)|
  31                                 # html5 player
  32                                 html5/html5lib/
  33                                 (?:[^/]+/)*?entry_id/(?P<id_html5>[0-9a-z_]+)
  34                                 .*\?.*\bwid=_(?P<partner_id_html5>\d+)
  35                             )
  36                         )
  37                 )
  38                 '''
  39     _API_BASE = 'http://cdnapi.kaltura.com/api_v3/index.php?'
  40     _TESTS = [
  41         {
  42             'url': 'kaltura:269692:1_1jc2y3e4',
  43             'md5': '3adcbdb3dcc02d647539e53f284ba171',
  44             'info_dict': {
  45                 'id': '1_1jc2y3e4',
  46                 'ext': 'mp4',
  47                 'title': 'Track 4',
  48                 'upload_date': '20131219',
  49                 'uploader_id': 'mlundberg@wolfgangsvault.com',
  50                 'description': 'The Allman Brothers Band, 12/16/1981',
  51                 'thumbnail': 're:^https?://.*/thumbnail/.*',
  52                 'timestamp': int,
  53             },
  54         },
  55         {
  56             'url': 'http://www.kaltura.com/index.php/kwidget/cache_st/1300318621/wid/_269692/uiconf_id/3873291/entry_id/1_1jc2y3e4',
  57             'only_matching': True,
  58         },
  59         {
  60             'url': 'https://cdnapisec.kaltura.com/index.php/kwidget/wid/_557781/uiconf_id/22845202/entry_id/1_plr1syf3',
  61             'only_matching': True,
  62         },
  63         {
  64             'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.30.2/mwEmbedFrame.php/p/1337/uiconf_id/20540612/entry_id/1_sf5ovm7u?wid=_243342',
  65             'only_matching': True,
  66         }
  67     ]
  68
  69     def _kaltura_api_call(self, video_id, actions, *args, **kwargs):
  70         params = actions[0]
  71         if len(actions) > 1:
  72             for i, a in enumerate(actions[1:], start=1):
  73                 for k, v in a.items():
  74                     params['%d:%s' % (i, k)] = v
  75
  76         query = compat_urllib_parse.urlencode(params)
  77         url = self._API_BASE + query
  78         data = self._download_json(url, video_id, *args, **kwargs)
  79
  80         status = data if len(actions) == 1 else data[0]
  81         if status.get('objectType') == 'KalturaAPIException':
  82             raise ExtractorError(
  83                 '%s said: %s' % (self.IE_NAME, status['message']))
  84
  85         return data
  86
  87     def _get_kaltura_signature(self, video_id, partner_id):
  88         actions = [{
  89             'apiVersion': '3.1',
  90             'expiry': 86400,
  91             'format': 1,
  92             'service': 'session',
  93             'action': 'startWidgetSession',
  94             'widgetId': '_%s' % partner_id,
  95         }]
  96         return self._kaltura_api_call(
  97             video_id, actions, note='Downloading Kaltura signature')['ks']
  98
  99     def _get_video_info(self, video_id, partner_id):
 100         signature = self._get_kaltura_signature(video_id, partner_id)
 101         actions = [
 102             {
 103                 'action': 'null',
 104                 'apiVersion': '3.1.5',
 105                 'clientTag': 'kdp:v3.8.5',
 106                 'format': 1,  # JSON, 2 = XML, 3 = PHP
 107                 'service': 'multirequest',
 108                 'ks': signature,
 109             },
 110             {
 111                 'action': 'get',
 112                 'entryId': video_id,
 113                 'service': 'baseentry',
 114                 'version': '-1',
 115             },
 116             {
 117                 'action': 'getContextData',
 118                 'contextDataParams:objectType': 'KalturaEntryContextDataParams',
 119                 'contextDataParams:referrer': 'http://www.kaltura.com/',
 120                 'contextDataParams:streamerType': 'http',
 121                 'entryId': video_id,
 122                 'service': 'baseentry',
 123             },
 124         ]
 125         return self._kaltura_api_call(
 126             video_id, actions, note='Downloading video info JSON')
 127
 128     def _real_extract(self, url):
 129         url, smuggled_data = unsmuggle_url(url, {})
 130
 131         mobj = re.match(self._VALID_URL, url)
 132         partner_id = mobj.group('partner_id_s') or mobj.group('partner_id') or mobj.group('partner_id_html5')
 133         entry_id = mobj.group('id_s') or mobj.group('id') or mobj.group('id_html5')
 134
 135         info, source_data = self._get_video_info(entry_id, partner_id)
 136
 137         source_url = smuggled_data.get('source_url')
 138         if source_url:
 139             referrer = base64.b64encode(
 140                 '://'.join(compat_urlparse.urlparse(source_url)[:2])
 141                 .encode('utf-8')).decode('utf-8')
 142         else:
 143             referrer = None
 144
 145         formats = []
 146         for f in source_data['flavorAssets']:
 147             video_url = '%s/flavorId/%s' % (info['dataUrl'], f['id'])
 148             if referrer:
 149                 video_url += '?referrer=%s' % referrer
 150             formats.append({
 151                 'format_id': '%(fileExt)s-%(bitrate)s' % f,
 152                 'ext': f.get('fileExt'),
 153                 'tbr': int_or_none(f['bitrate']),
 154                 'fps': int_or_none(f.get('frameRate')),
 155                 'filesize_approx': int_or_none(f.get('size'), invscale=1024),
 156                 'container': f.get('containerFormat'),
 157                 'vcodec': f.get('videoCodecId'),
 158                 'height': int_or_none(f.get('height')),
 159                 'width': int_or_none(f.get('width')),
 160                 'url': video_url,
 161             })
 162         self._check_formats(formats, entry_id)
 163         self._sort_formats(formats)
 164
 165         return {
 166             'id': entry_id,
 167             'title': info['name'],
 168             'formats': formats,
 169             'description': info.get('description'),
 170             'thumbnail': info.get('thumbnailUrl'),
 171             'duration': info.get('duration'),
 172             'timestamp': info.get('createdAt'),
 173             'uploader_id': info.get('userId'),
 174             'view_count': info.get('plays'),
 175         }