From fe05191b8c59538a48b6cbc95f4fe54fc7e6a0ac Mon Sep 17 00:00:00 2001
From: Ben Rog-Wilhelm <zorba-github@pavlovian.net>
Date: Tue, 4 May 2021 14:14:35 -0500
Subject: [kaltura] Improve iframe extraction (#28969)

Co-authored-by: Sergey M. <dstftw@gmail.com>
---
 youtube_dl/extractor/gdcvault.py | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

(limited to 'youtube_dl/extractor/gdcvault.py')

diff --git a/youtube_dl/extractor/gdcvault.py b/youtube_dl/extractor/gdcvault.py
index 2f555c1d4..5ad40ee23 100644
--- a/youtube_dl/extractor/gdcvault.py
+++ b/youtube_dl/extractor/gdcvault.py
@@ -102,6 +102,21 @@ class GDCVaultIE(InfoExtractor):
                 'format': 'mp4-408',
             },
         },
+        {
+            # Kaltura embed, whitespace between quote and embedded URL in iframe's src
+            'url': 'https://www.gdcvault.com/play/1025699',
+            'info_dict': {
+                'id': '0_zagynv0a',
+                'ext': 'mp4',
+                'title': 'Tech Toolbox',
+                'upload_date': '20190408',
+                'uploader_id': 'joe@blazestreaming.com',
+                'timestamp': 1554764629,
+            },
+            'params': {
+                'skip_download': True,
+            },
+        },
     ]
 
     def _login(self, webpage_url, display_id):
-- 
cgit v1.2.3


From 504e4d804df0ee666d80ba6796017cf97e026c0b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sergey=20M=E2=80=A4?= <dstftw@gmail.com>
Date: Wed, 5 May 2021 02:44:29 +0700
Subject: [gdcvault] Add support for HTML5 videos

---
 youtube_dl/extractor/gdcvault.py | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

(limited to 'youtube_dl/extractor/gdcvault.py')

diff --git a/youtube_dl/extractor/gdcvault.py b/youtube_dl/extractor/gdcvault.py
index 5ad40ee23..acc6478b8 100644
--- a/youtube_dl/extractor/gdcvault.py
+++ b/youtube_dl/extractor/gdcvault.py
@@ -6,6 +6,7 @@ from .common import InfoExtractor
 from .kaltura import KalturaIE
 from ..utils import (
     HEADRequest,
+    remove_start,
     sanitized_Request,
     smuggle_url,
     urlencode_postdata,
@@ -117,6 +118,11 @@ class GDCVaultIE(InfoExtractor):
                 'skip_download': True,
             },
         },
+        {
+            # HTML5 video
+            'url': 'http://www.gdcvault.com/play/1014846/Conference-Keynote-Shigeru',
+            'only_matching': True,
+        },
     ]
 
     def _login(self, webpage_url, display_id):
@@ -190,7 +196,18 @@ class GDCVaultIE(InfoExtractor):
 
             xml_name = self._html_search_regex(
                 r'<iframe src=".*?\?xml(?:=|URL=xml/)(.+?\.xml).*?".*?</iframe>',
-                start_page, 'xml filename')
+                start_page, 'xml filename', default=None)
+            if not xml_name:
+                info = self._parse_html5_media_entries(url, start_page, video_id)[0]
+                info.update({
+                    'title': remove_start(self._search_regex(
+                        r'>Session Name:\s*<.*?>\s*<td>(.+?)</td>', start_page,
+                        'title', default=None) or self._og_search_title(
+                        start_page, default=None), 'GDC Vault - '),
+                    'id': video_id,
+                    'display_id': display_id,
+                })
+                return info
             embed_url = '%s/xml/%s' % (xml_root, xml_name)
             ie_key = 'DigitallySpeaking'
 
-- 
cgit v1.2.3