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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..compat import compat_urllib_parse
   8 from ..utils import (
   9     ExtractorError,
  10     int_or_none,
  11 )
  12
  13
  14 class KalturaIE(InfoExtractor):
  15     _VALID_URL = r'''(?x)
  16                 (?:
  17                     kaltura:(?P<partner_id_s>\d+):(?P<id_s>[0-9a-z_]+)|
  18                     https?://
  19                     (:?
  20                         (?:www|cdnapisec)\.)?kaltura\.com/
  21                         (?:
  22                             (?:
  23                                 # flash player
  24                                 index\.php/kwidget/
  25                                 (?:[^/]+/)*?wid/_(?P<partner_id>\d+)/
  26                                 (?:[^/]+/)*?entry_id/(?P<id>[0-9a-z_]+)|
  27                                 # html player
  28                                 html5/html5lib/
  29                                 (?:[^/]+/)*?entry_id/(?P<id_html5>[0-9a-z_]+)
  30                                 .*\?.*\bwid=_(?P<partner_id_html5>\d+)
  31                         )
  32                     )
  33                 )
  34                 '''
  35     _API_BASE = 'http://cdnapi.kaltura.com/api_v3/index.php?'
  36     _TESTS = [
  37         {
  38             'url': 'kaltura:269692:1_1jc2y3e4',
  39             'md5': '3adcbdb3dcc02d647539e53f284ba171',
  40             'info_dict': {
  41                 'id': '1_1jc2y3e4',
  42                 'ext': 'mp4',
  43                 'title': 'Track 4',
  44                 'upload_date': '20131219',
  45                 'uploader_id': 'mlundberg@wolfgangsvault.com',
  46                 'description': 'The Allman Brothers Band, 12/16/1981',
  47                 'thumbnail': 're:^https?://.*/thumbnail/.*',
  48                 'timestamp': int,
  49             },
  50         },
  51         {
  52             'url': 'http://www.kaltura.com/index.php/kwidget/cache_st/1300318621/wid/_269692/uiconf_id/3873291/entry_id/1_1jc2y3e4',
  53             'only_matching': True,
  54         },
  55         {
  56             'url': 'https://cdnapisec.kaltura.com/index.php/kwidget/wid/_557781/uiconf_id/22845202/entry_id/1_plr1syf3',
  57             'only_matching': True,
  58         },
  59         {
  60             'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.30.2/mwEmbedFrame.php/p/1337/uiconf_id/20540612/entry_id/1_sf5ovm7u?wid=_243342',
  61             'only_matching': True,
  62         }
  63     ]
  64
  65     def _kaltura_api_call(self, video_id, actions, *args, **kwargs):
  66         params = actions[0]
  67         if len(actions) > 1:
  68             for i, a in enumerate(actions[1:], start=1):
  69                 for k, v in a.items():
  70                     params['%d:%s' % (i, k)] = v
  71
  72         query = compat_urllib_parse.urlencode(params)
  73         url = self._API_BASE + query
  74         data = self._download_json(url, video_id, *args, **kwargs)
  75
  76         status = data if len(actions) == 1 else data[0]
  77         if status.get('objectType') == 'KalturaAPIException':
  78             raise ExtractorError(
  79                 '%s said: %s' % (self.IE_NAME, status['message']))
  80
  81         return data
  82
  83     def _get_kaltura_signature(self, video_id, partner_id):
  84         actions = [{
  85             'apiVersion': '3.1',
  86             'expiry': 86400,
  87             'format': 1,
  88             'service': 'session',
  89             'action': 'startWidgetSession',
  90             'widgetId': '_%s' % partner_id,
  91         }]
  92         return self._kaltura_api_call(
  93             video_id, actions, note='Downloading Kaltura signature')['ks']
  94
  95     def _get_video_info(self, video_id, partner_id):
  96         signature = self._get_kaltura_signature(video_id, partner_id)
  97         actions = [
  98             {
  99                 'action': 'null',
 100                 'apiVersion': '3.1.5',
 101                 'clientTag': 'kdp:v3.8.5',
 102                 'format': 1,  # JSON, 2 = XML, 3 = PHP
 103                 'service': 'multirequest',
 104                 'ks': signature,
 105             },
 106             {
 107                 'action': 'get',
 108                 'entryId': video_id,
 109                 'service': 'baseentry',
 110                 'version': '-1',
 111             },
 112             {
 113                 'action': 'getContextData',
 114                 'contextDataParams:objectType': 'KalturaEntryContextDataParams',
 115                 'contextDataParams:referrer': 'http://www.kaltura.com/',
 116                 'contextDataParams:streamerType': 'http',
 117                 'entryId': video_id,
 118                 'service': 'baseentry',
 119             },
 120         ]
 121         return self._kaltura_api_call(
 122             video_id, actions, note='Downloading video info JSON')
 123
 124     def _real_extract(self, url):
 125         mobj = re.match(self._VALID_URL, url)
 126         partner_id = mobj.group('partner_id_s') or mobj.group('partner_id') or mobj.group('partner_id_html5')
 127         entry_id = mobj.group('id_s') or mobj.group('id') or mobj.group('id_html5')
 128
 129         info, source_data = self._get_video_info(entry_id, partner_id)
 130
 131         formats = [{
 132             'format_id': '%(fileExt)s-%(bitrate)s' % f,
 133             'ext': f['fileExt'],
 134             'tbr': f['bitrate'],
 135             'fps': f.get('frameRate'),
 136             'filesize_approx': int_or_none(f.get('size'), invscale=1024),
 137             'container': f.get('containerFormat'),
 138             'vcodec': f.get('videoCodecId'),
 139             'height': f.get('height'),
 140             'width': f.get('width'),
 141             'url': '%s/flavorId/%s' % (info['dataUrl'], f['id']),
 142         } for f in source_data['flavorAssets']]
 143         self._sort_formats(formats)
 144
 145         return {
 146             'id': entry_id,
 147             'title': info['name'],
 148             'formats': formats,
 149             'description': info.get('description'),
 150             'thumbnail': info.get('thumbnailUrl'),
 151             'duration': info.get('duration'),
 152             'timestamp': info.get('createdAt'),
 153             'uploader_id': info.get('userId'),
 154             'view_count': info.get('plays'),
 155         }