2 from __future__ import unicode_literals
7 from .common import InfoExtractor
21 class KalturaIE(InfoExtractor):
24 kaltura:(?P<partner_id>\d+):(?P<id>[0-9a-z_]+)|
26 (:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/
32 html5/html5lib/[^/]+/mwEmbedFrame\.php
34 )(?:/(?P<path>[^?]+))?(?:\?(?P<query>.*))?
37 _API_BASE = 'http://cdnapi.kaltura.com/api_v3/index.php?'
40 'url': 'kaltura:269692:1_1jc2y3e4',
41 'md5': '3adcbdb3dcc02d647539e53f284ba171',
45 'title': 'Straight from the Heart',
46 'upload_date': '20131219',
47 'uploader_id': 'mlundberg@wolfgangsvault.com',
48 'description': 'The Allman Brothers Band, 12/16/1981',
49 'thumbnail': 're:^https?://.*/thumbnail/.*',
54 'url': 'http://www.kaltura.com/index.php/kwidget/cache_st/1300318621/wid/_269692/uiconf_id/3873291/entry_id/1_1jc2y3e4',
55 'only_matching': True,
58 'url': 'https://cdnapisec.kaltura.com/index.php/kwidget/wid/_557781/uiconf_id/22845202/entry_id/1_plr1syf3',
59 'only_matching': True,
62 'url': 'https://cdnapisec.kaltura.com/html5/html5lib/v2.30.2/mwEmbedFrame.php/p/1337/uiconf_id/20540612/entry_id/1_sf5ovm7u?wid=_243342',
63 'only_matching': True,
67 def _kaltura_api_call(self, video_id, actions, *args, **kwargs):
70 for i, a in enumerate(actions[1:], start=1):
71 for k, v in a.items():
72 params['%d:%s' % (i, k)] = v
74 query = compat_urllib_parse.urlencode(params)
75 url = self._API_BASE + query
76 data = self._download_json(url, video_id, *args, **kwargs)
78 status = data if len(actions) == 1 else data[0]
79 if status.get('objectType') == 'KalturaAPIException':
81 '%s said: %s' % (self.IE_NAME, status['message']))
85 def _get_kaltura_signature(self, video_id, partner_id):
91 'action': 'startWidgetSession',
92 'widgetId': '_%s' % partner_id,
94 return self._kaltura_api_call(
95 video_id, actions, note='Downloading Kaltura signature')['ks']
97 def _get_video_info(self, video_id, partner_id):
98 signature = self._get_kaltura_signature(video_id, partner_id)
102 'apiVersion': '3.1.5',
103 'clientTag': 'kdp:v3.8.5',
104 'format': 1, # JSON, 2 = XML, 3 = PHP
105 'service': 'multirequest',
111 'service': 'baseentry',
115 'action': 'getbyentryid',
117 'service': 'flavorAsset',
120 return self._kaltura_api_call(
121 video_id, actions, note='Downloading video info JSON')
123 def _real_extract(self, url):
124 url, smuggled_data = unsmuggle_url(url, {})
126 mobj = re.match(self._VALID_URL, url)
127 partner_id, entry_id = mobj.group('partner_id', 'id')
128 info, flavor_assets = None, None
129 if partner_id and entry_id:
130 info, flavor_assets = self._get_video_info(entry_id, partner_id)
132 path, query = mobj.group('path', 'query')
133 if not path and not query:
134 raise ExtractorError('Invalid URL', expected=True)
137 params = compat_parse_qs(query)
139 splitted_path = path.split('/')
140 if not splitted_path[-1]:
141 splitted_path = splitted_path[:-1]
142 for i in range(0, len(splitted_path), 2):
143 params[splitted_path[i]] = [splitted_path[i + 1]]
145 partner_id = params['wid'][0][1:]
147 partner_id = params['p'][0]
149 raise ExtractorError('Invalid URL', expected=True)
150 if 'entry_id' in params:
151 entry_id = params['entry_id'][0]
152 info, flavor_assets = self._get_video_info(entry_id, partner_id)
153 elif 'uiconf_id' in params and 'flashvars[referenceId]' in params:
154 reference_id = params['flashvars[referenceId]'][0]
155 webpage = self._download_webpage(url, reference_id)
156 entry_data = self._parse_json(self._search_regex(
157 r'window\.kalturaIframePackageData\s*=\s*({.*});',
158 webpage, 'kalturaIframePackageData'),
159 reference_id)['entryResult']
160 info, flavor_assets = entry_data['meta'], entry_data['contextData']['flavorAssets']
161 entry_id = info['id']
163 raise ExtractorError('Invalid URL', expected=True)
165 source_url = smuggled_data.get('source_url')
167 referrer = base64.b64encode(
168 '://'.join(compat_urlparse.urlparse(source_url)[:2])
169 .encode('utf-8')).decode('utf-8')
174 for f in flavor_assets:
175 # Continue if asset is not ready
178 video_url = '%s/flavorId/%s' % (info['dataUrl'], f['id'])
180 video_url += '?referrer=%s' % referrer
182 'format_id': '%(fileExt)s-%(bitrate)s' % f,
183 'ext': f.get('fileExt'),
184 'tbr': int_or_none(f['bitrate']),
185 'fps': int_or_none(f.get('frameRate')),
186 'filesize_approx': int_or_none(f.get('size'), invscale=1024),
187 'container': f.get('containerFormat'),
188 'vcodec': f.get('videoCodecId'),
189 'height': int_or_none(f.get('height')),
190 'width': int_or_none(f.get('width')),
193 m3u8_url = info['dataUrl'].replace('format/url', 'format/applehttp')
195 m3u8_url += '?referrer=%s' % referrer
196 formats.extend(self._extract_m3u8_formats(
197 m3u8_url, entry_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
199 self._check_formats(formats, entry_id)
200 self._sort_formats(formats)
204 'title': info['name'],
206 'description': clean_html(info.get('description')),
207 'thumbnail': info.get('thumbnailUrl'),
208 'duration': info.get('duration'),
209 'timestamp': info.get('createdAt'),
210 'uploader_id': info.get('userId'),
211 'view_count': info.get('plays'),