Annotate some highlights of the release.

[youtubedl] / youtube_dl / extractor / kaltura.py

diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py
index 0ea89e4d66d9fb20a9e9d9cf6635d7dff09f4ba6..49d13460df7f0edd4d2a08f97deaf831ba9d6a46 100644 (file)
--- a/youtube_dl/extractor/kaltura.py
+++ b/youtube_dl/extractor/kaltura.py
@@ -103,14 +103,24 @@ class KalturaIE(InfoExtractor):
         {
             'url': 'https://www.kaltura.com:443/index.php/extwidget/preview/partner_id/1770401/uiconf_id/37307382/entry_id/0_58u8kme7/embed/iframe?&flashvars[streamerType]=auto',
             'only_matching': True,
+        },
+        {
+            # unavailable source format
+            'url': 'kaltura:513551:1_66x4rg7o',
+            'only_matching': True,
         }
     ]
 
     @staticmethod
     def _extract_url(webpage):
+        urls = KalturaIE._extract_urls(webpage)
+        return urls[0] if urls else None
+
+    @staticmethod
+    def _extract_urls(webpage):
         # Embed codes: https://knowledge.kaltura.com/embedding-kaltura-media-players-your-site
-        mobj = (
-            re.search(
+        finditer = (
+            re.finditer(
                 r"""(?xs)
                     kWidget\.(?:thumb)?[Ee]mbed\(
                     \{.*?
@@ -118,8 +128,8 @@ class KalturaIE(InfoExtractor):
                         (?P<q2>['"])_?(?P<partner_id>(?:(?!(?P=q2)).)+)(?P=q2),.*?
                         (?P<q3>['"])entry_?[Ii]d(?P=q3)\s*:\s*
                         (?P<q4>['"])(?P<id>(?:(?!(?P=q4)).)+)(?P=q4)(?:,|\s*\})
-                """, webpage) or
-            re.search(
+                """, webpage)
+            or re.finditer(
                 r'''(?xs)
                     (?P<q1>["'])
                         (?:https?:)?//cdnapi(?:sec)?\.kaltura\.com(?::\d+)?/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+)(?:(?!(?P=q1)).)*
@@ -132,26 +142,32 @@ class KalturaIE(InfoExtractor):
                         \[\s*(?P<q2_1>["'])entry_?[Ii]d(?P=q2_1)\s*\]\s*=\s*
                     )
                     (?P<q3>["'])(?P<id>(?:(?!(?P=q3)).)+)(?P=q3)
-                ''', webpage) or
-            re.search(
+                ''', webpage)
+            or re.finditer(
                 r'''(?xs)
                     <(?:iframe[^>]+src|meta[^>]+\bcontent)=(?P<q1>["'])
-                      (?:https?:)?//(?:(?:www|cdnapi)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+)
+                      (?:https?:)?//(?:(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+)
                       (?:(?!(?P=q1)).)*
                       [?&;]entry_id=(?P<id>(?:(?!(?P=q1))[^&])+)
+                      (?:(?!(?P=q1)).)*
                     (?P=q1)
                 ''', webpage)
         )
-        if mobj:
+        urls = []
+        for mobj in finditer:
             embed_info = mobj.groupdict()
+            for k, v in embed_info.items():
+                if v:
+                    embed_info[k] = v.strip()
             url = 'kaltura:%(partner_id)s:%(id)s' % embed_info
             escaped_pid = re.escape(embed_info['partner_id'])
-            service_url = re.search(
-                r'<script[^>]+src=["\']((?:https?:)?//.+?)/p/%s/sp/%s00/embedIframeJs' % (escaped_pid, escaped_pid),
+            service_mobj = re.search(
+                r'<script[^>]+src=(["\'])(?P<id>(?:https?:)?//(?:(?!\1).)+)/p/%s/sp/%s00/embedIframeJs' % (escaped_pid, escaped_pid),
                 webpage)
-            if service_url:
-                url = smuggle_url(url, {'service_url': service_url.group(1)})
-            return url
+            if service_mobj:
+                url = smuggle_url(url, {'service_url': service_mobj.group('id')})
+            urls.append(url)
+        return urls
 
     def _kaltura_api_call(self, video_id, actions, service_url=None, *args, **kwargs):
         params = actions[0]
@@ -191,6 +207,8 @@ class KalturaIE(InfoExtractor):
                 'entryId': video_id,
                 'service': 'baseentry',
                 'ks': '{1:result:ks}',
+                'responseProfile:fields': 'createdAt,dataUrl,duration,name,plays,thumbnailUrl,userId',
+                'responseProfile:type': 1,
             },
             {
                 'action': 'getbyentryid',
@@ -301,12 +319,17 @@ class KalturaIE(InfoExtractor):
                     f['fileExt'] = 'mp4'
             video_url = sign_url(
                 '%s/flavorId/%s' % (data_url, f['id']))
+            format_id = '%(fileExt)s-%(bitrate)s' % f
+            # Source format may not be available (e.g. kaltura:513551:1_66x4rg7o)
+            if f.get('isOriginal') is True and not self._is_valid_url(
+                    video_url, entry_id, format_id):
+                continue
             # audio-only has no videoCodecId (e.g. kaltura:1926081:0_c03e1b5g
             # -f mp4-56)
             vcodec = 'none' if 'videoCodecId' not in f and f.get(
                 'frameRate') == 0 else f.get('videoCodecId')
             formats.append({
-                'format_id': '%(fileExt)s-%(bitrate)s' % f,
+                'format_id': format_id,
                 'ext': f.get('fileExt'),
                 'tbr': int_or_none(f['bitrate']),
                 'fps': int_or_none(f.get('frameRate')),