media: Add MediaDrmStorage

chrome/browser/media/android/cdm/media_drm_storage_factory.cc

Index: chrome/browser/media/android/cdm/media_drm_storage_factory.cc
diff --git a/chrome/browser/media/android/cdm/media_drm_storage_factory.cc b/chrome/browser/media/android/cdm/media_drm_storage_factory.cc
new file mode 100644
index 0000000000000000000000000000000000000000..fc54ccc614f314b448618045d0111adee326521a
--- /dev/null
+++ b/chrome/browser/media/android/cdm/media_drm_storage_factory.cc
@@ -0,0 +1,48 @@
+// 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.
+
+#include "chrome/browser/media/android/cdm/media_drm_storage_factory.h"
+
+#include "base/logging.h"
+#include "base/memory/ptr_util.h"
+#include "chrome/browser/profiles/profile.h"
+#include "components/cdm/browser/media_drm_storage_impl.h"
+#include "components/prefs/pref_service.h"
+#include "content/public/browser/browser_context.h"
+#include "content/public/browser/browser_thread.h"
+#include "content/public/browser/render_frame_host.h"
+#include "content/public/browser/web_contents.h"
+#include "mojo/public/cpp/bindings/strong_binding.h"
+
+namespace chrome {
+
+void CreateMediaDrmStorage(content::RenderFrameHost* render_frame_host,
+                           media::mojom::MediaDrmStorageRequest request) {
+  DVLOG(2) << __func__;
+  DCHECK_CURRENTLY_ON(content::BrowserThread::UI);
+  DCHECK(render_frame_host);
+
+  content::WebContents* web_contents =
+      content::WebContents::FromRenderFrameHost(render_frame_host);
+  if (!web_contents)

[dcheng, 2017/03/29 01:17:22]

I'm curious how many of these early returns are actually needed. For example, is
it possible for this to return nothing if RenderFrameHost is not null?

[xhwang, 2017/03/29 20:47:25]

Well, I have the same question. But I'd rather code against the API [1] which
doesn't provide any comment :) I could check the implementation but I choose not
to, because I don't want to code against implementation details, which isn't
recommended...

+    return;
+
+  content::BrowserContext* browser_context = web_contents->GetBrowserContext();

[dcheng, 2017/03/29 01:17:22]

Similarly, it doesn't seem like this should /ever/ be null.

[xhwang, 2017/03/29 20:47:25]

Similarly, it's not clear to me whether this could be null from the API :(
https://cs.chromium.org/chromium/src/content/public/browser/web_contents.h?rc...

+  if (!browser_context)
+    return;
+
+  Profile* profile = Profile::FromBrowserContext(browser_context);
+  if (!profile)
+    return;
+
+  PrefService* pref_service = profile->GetPrefs();

[liberato (no reviews please), 2017/03/28 16:42:43]

how does lifetime for |pref_service| work?  for incognito sessions, at least,
the prefs service might be owned by the profile.

[xhwang, 2017/03/29 20:47:25]

I don't really know. But I am following how PrefService are used throughout the
code base. I think it makes sense to assume that if a RenderFrameHost is alive,
then the profile associated with it is also alive.

+  if (!pref_service)
+    return;

[yucliu1, 2017/03/27 19:06:41]

Are we able to launch the page if these early returns happen? If so, should we
just disable all the persistent license request?

[xhwang, 2017/03/27 20:28:46]

All of these should never happen in practice :)

If they do happen for whatever reason, the MojoMediaDrmStorage will get a
ConnectionError. We can make the Initialize() call asyc so that we'll know
whether it's initialized successfully. If it failed, we can reject all calls
related to persistent license.

I'll add a TODO to handle connection error in MojoMediaDrmStorage.

[xhwang, 2017/03/29 20:47:25]

Done.

+
+  mojo::MakeStrongBinding(
+      base::MakeUnique<cdm::MediaDrmStorageImpl>(pref_service),
+      std::move(request));

[dcheng, 2017/03/29 01:17:22]

Nit: #include <utility>

[xhwang, 2017/03/29 20:47:25]

Done.

+}
+
+}  // namespace chrome