From 30d894749655ab81087c51c96092fd717926d26d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=3D=3FUTF-8=3Fq=3FSergey=3D20M=3DE2=3D80=3DA4=3F=3D?=
 <dstftw@gmail.com>
Date: Mon, 31 May 2021 23:20:33 +0200
Subject: [PATCH] [gdcvault] Add support for HTML5 videos

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

diff --git a/haruhi_dl/extractor/gdcvault.py b/haruhi_dl/extractor/gdcvault.py
index 5ad40ee23..acc6478b8 100644
--- a/haruhi_dl/extractor/gdcvault.py
+++ b/haruhi_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'