]> gitweb @ CieloNegro.org - youtube-dl.git/commitdiff
[kaltura] Improve iframe embeds detection (closes #16337)
authorSergey M․ <dstftw@gmail.com>
Mon, 30 Apr 2018 20:09:04 +0000 (03:09 +0700)
committerSergey M․ <dstftw@gmail.com>
Mon, 30 Apr 2018 20:09:04 +0000 (03:09 +0700)
youtube_dl/extractor/generic.py
youtube_dl/extractor/kaltura.py

index 252f97c2617c2322ca454263536e6839b996cefc..73980601c585401c5385759df3afa054f7e6d29e 100644 (file)
@@ -1282,6 +1282,23 @@ class GenericIE(InfoExtractor):
             },
             'add_ie': ['Kaltura'],
         },
+        {
+            # Kaltura iframe embed, more sophisticated
+            'url': 'http://www.cns.nyu.edu/~eero/math-tools/Videos/lecture-05sep2017.html',
+            'info_dict': {
+                'id': '1_9gzouybz',
+                'ext': 'mp4',
+                'title': 'lecture-05sep2017',
+                'description': 'md5:40f347d91fd4ba047e511c5321064b49',
+                'upload_date': '20170913',
+                'uploader_id': 'eps2',
+                'timestamp': 1505340777,
+            },
+            'params': {
+                'skip_download': True,
+            },
+            'add_ie': ['Kaltura'],
+        },
         {
             # meta twitter:player
             'url': 'http://thechive.com/2017/12/08/all-i-want-for-christmas-is-more-twerk/',
index 0ea89e4d66d9fb20a9e9d9cf6635d7dff09f4ba6..04f68fce41fc129f7cf65e0e1cddb0e85636fb9e 100644 (file)
@@ -136,9 +136,10 @@ class KalturaIE(InfoExtractor):
             re.search(
                 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)
         )