RenderFrameProxyHost::OnOpenURL needs to validate resource request body.

content/browser/resource_request_body_browser_utils.h

Index: content/browser/resource_request_body_browser_utils.h
diff --git a/content/browser/resource_request_body_browser_utils.h b/content/browser/resource_request_body_browser_utils.h
new file mode 100644
index 0000000000000000000000000000000000000000..7b4a0e3bda5184059781e81e9bf49605a877fa72
--- /dev/null
+++ b/content/browser/resource_request_body_browser_utils.h
@@ -0,0 +1,24 @@
+// Copyright 2017 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef CONTENT_BROWSER_RESOURCE_REQUEST_BODY_BROWSER_UTILS_H_
+#define CONTENT_BROWSER_RESOURCE_REQUEST_BODY_BROWSER_UTILS_H_
+
+#include "base/memory/ref_counted.h"
+
+namespace content {
+
+class SiteInstance;
+class ResourceRequestBodyImpl;
+
+// Checks if |site_instance| can read all elements of |body|.

[alexmos, 2017/05/25 23:44:06]

nit: Maybe elaborate a little bit?  I.e., validate that the renderer process for
|site_instance| is allowed to access files in the POST upload data specified by
|body|.  

Also, probably more for your other CL, but maybe also mention what can be called
on what thread (since you'll need the deduplicated helper on both UI and IO
threads, though this one takes a SiteInstance so should be UI thread).

[Łukasz Anforowicz, 2017/05/26 00:05:14]

Done.
Done.

+// TODO(lukasza): Remove code duplication - the function below should be reused
+// by RenderFrameHostImpl::OnBeginNavigation and
+// ResourceDispatcherHostImpl::ShouldServiceRequest.
+bool CanReadRequestBody(SiteInstance* site_instance,
+                        const scoped_refptr<ResourceRequestBodyImpl>& body);

[Łukasz Anforowicz, 2017/05/25 19:56:01]

Does this look okay?

I had some comments about the file/dir for this new function in
https://codereview.chromium.org/2905763002/diff/20001/content/browser/resourc...

mmenke@ had some (optional) suggestions about possibly putting the new function
into a static-only class or namespace.

I'll rebase https://codereview.chromium.org/2905763002 in a minute.

[alexmos, 2017/05/25 23:44:06]

Hmm, this might be ok.  A static-only class might make it easier to discover,
but I think it'd make more sense if you had multiple of these helpers that it
could help group together.  But maybe also run this by nick@?

[Łukasz Anforowicz, 2017/05/26 00:05:14]

Ok - see
https://codereview.chromium.org/2908433003/diff/80001/content/browser/resourc...

+
+}  // namespace content
+
+#endif  // CONTENT_BROWSER_RESOURCE_REQUEST_BODY_BROWSER_UTILS_H_