X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fkaltura.py;h=44d7c84a13f9bef9aa1d68dc5d38fe81b0af4a5f;hb=2e7e561c1d9dedf1a8e5a206e1ef86cfa4599956;hp=1eaa09fea0e6528656c1596cc65a0df8df14e640;hpb=d80a39cec89e8ddfc3090c3d1b10d3fe105791e1;p=youtube-dl

diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py
index 1eaa09fea..44d7c84a1 100644
--- a/youtube_dl/extractor/kaltura.py
+++ b/youtube_dl/extractor/kaltura.py
@@ -8,8 +8,10 @@ from .common import InfoExtractor
 from ..compat import (
     compat_urllib_parse,
     compat_urlparse,
+    compat_parse_qs,
 )
 from ..utils import (
+    clean_html,
     ExtractorError,
     int_or_none,
     unsmuggle_url,
@@ -19,21 +21,17 @@ from ..utils import (
 class KalturaIE(InfoExtractor):
     _VALID_URL = r'''(?x)
                 (?:
-                    kaltura:(?P<partner_id_s>\d+):(?P<id_s>[0-9a-z_]+)|
+                    kaltura:(?P<partner_id>\d+):(?P<id>[0-9a-z_]+)|
                     https?://
                         (:?(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/
                         (?:
                             (?:
                                 # flash player
-                                index\.php/kwidget/
-                                (?:[^/]+/)*?wid/_(?P<partner_id>\d+)/
-                                (?:[^/]+/)*?entry_id/(?P<id>[0-9a-z_]+)|
+                                index\.php/kwidget|
                                 # html5 player
-                                html5/html5lib/
-                                (?:[^/]+/)*?entry_id/(?P<id_html5>[0-9a-z_]+)
-                                .*\?.*\bwid=_(?P<partner_id_html5>\d+)
+                                html5/html5lib/[^/]+/mwEmbedFrame\.php
                             )
-                        )
+                        )(?:/(?P<path>[^?]+))?(?:\?(?P<query>.*))?
                 )
                 '''
     _API_BASE = 'http://cdnapi.kaltura.com/api_v3/index.php?'
@@ -44,7 +42,7 @@ class KalturaIE(InfoExtractor):
             'info_dict': {
                 'id': '1_1jc2y3e4',
                 'ext': 'mp4',
-                'title': 'Track 4',
+                'title': 'Straight from the Heart',
                 'upload_date': '20131219',
                 'uploader_id': 'mlundberg@wolfgangsvault.com',
                 'description': 'The Allman Brothers Band, 12/16/1981',
@@ -114,12 +112,9 @@ class KalturaIE(InfoExtractor):
                 'version': '-1',
             },
             {
-                'action': 'getContextData',
-                'contextDataParams:objectType': 'KalturaEntryContextDataParams',
-                'contextDataParams:referrer': 'http://www.kaltura.com/',
-                'contextDataParams:streamerType': 'http',
+                'action': 'getbyentryid',
                 'entryId': video_id,
-                'service': 'baseentry',
+                'service': 'flavorAsset',
             },
         ]
         return self._kaltura_api_call(
@@ -129,10 +124,41 @@ class KalturaIE(InfoExtractor):
         url, smuggled_data = unsmuggle_url(url, {})
 
         mobj = re.match(self._VALID_URL, url)
-        partner_id = mobj.group('partner_id_s') or mobj.group('partner_id') or mobj.group('partner_id_html5')
-        entry_id = mobj.group('id_s') or mobj.group('id') or mobj.group('id_html5')
-
-        info, source_data = self._get_video_info(entry_id, partner_id)
+        partner_id, entry_id = mobj.group('partner_id', 'id')
+        ks = None
+        if partner_id and entry_id:
+            info, flavor_assets = self._get_video_info(entry_id, partner_id)
+        else:
+            path, query = mobj.group('path', 'query')
+            if not path and not query:
+                raise ExtractorError('Invalid URL', expected=True)
+            params = {}
+            if query:
+                params = compat_parse_qs(query)
+            if path:
+                splitted_path = path.split('/')
+                params.update(dict((zip(splitted_path[::2], [[v] for v in splitted_path[1::2]]))))
+            if 'wid' in params:
+                partner_id = params['wid'][0][1:]
+            elif 'p' in params:
+                partner_id = params['p'][0]
+            else:
+                raise ExtractorError('Invalid URL', expected=True)
+            if 'entry_id' in params:
+                entry_id = params['entry_id'][0]
+                info, flavor_assets = self._get_video_info(entry_id, partner_id)
+            elif 'uiconf_id' in params and 'flashvars[referenceId]' in params:
+                reference_id = params['flashvars[referenceId]'][0]
+                webpage = self._download_webpage(url, reference_id)
+                entry_data = self._parse_json(self._search_regex(
+                    r'window\.kalturaIframePackageData\s*=\s*({.*});',
+                    webpage, 'kalturaIframePackageData'),
+                    reference_id)['entryResult']
+                info, flavor_assets = entry_data['meta'], entry_data['contextData']['flavorAssets']
+                entry_id = info['id']
+            else:
+                raise ExtractorError('Invalid URL', expected=True)
+            ks = params.get('flashvars[ks]', [None])[0]
 
         source_url = smuggled_data.get('source_url')
         if source_url:
@@ -142,11 +168,19 @@ class KalturaIE(InfoExtractor):
         else:
             referrer = None
 
-        formats = []
-        for f in source_data['flavorAssets']:
-            video_url = '%s/flavorId/%s' % (info['dataUrl'], f['id'])
+        def sign_url(unsigned_url):
+            if ks:
+                unsigned_url += '/ks/%s' % ks
             if referrer:
-                video_url += '?referrer=%s' % referrer
+                unsigned_url += '?referrer=%s' % referrer
+            return unsigned_url
+
+        formats = []
+        for f in flavor_assets:
+            # Continue if asset is not ready
+            if f['status'] != 2:
+                continue
+            video_url = sign_url('%s/flavorId/%s' % (info['dataUrl'], f['id']))
             formats.append({
                 'format_id': '%(fileExt)s-%(bitrate)s' % f,
                 'ext': f.get('fileExt'),
@@ -159,6 +193,10 @@ class KalturaIE(InfoExtractor):
                 'width': int_or_none(f.get('width')),
                 'url': video_url,
             })
+        m3u8_url = sign_url(info['dataUrl'].replace('format/url', 'format/applehttp'))
+        formats.extend(self._extract_m3u8_formats(
+            m3u8_url, entry_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
+
         self._check_formats(formats, entry_id)
         self._sort_formats(formats)
 
@@ -166,7 +204,7 @@ class KalturaIE(InfoExtractor):
             'id': entry_id,
             'title': info['name'],
             'formats': formats,
-            'description': info.get('description'),
+            'description': clean_html(info.get('description')),
             'thumbnail': info.get('thumbnailUrl'),
             'duration': info.get('duration'),
             'timestamp': info.get('createdAt'),