Populate canPlayType to renderer

chrome/browser/media/encrypted_media_message_filter_android.cc

Index: chrome/browser/media/encrypted_media_message_filter_android.cc
diff --git a/chrome/browser/media/encrypted_media_message_filter_android.cc b/chrome/browser/media/encrypted_media_message_filter_android.cc
new file mode 100644
index 0000000000000000000000000000000000000000..89ec0b39ebf456bb639deaace4b57a38f33d7f37
--- /dev/null
+++ b/chrome/browser/media/encrypted_media_message_filter_android.cc
@@ -0,0 +1,82 @@
+// Copyright 2013 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/encrypted_media_message_filter_android.h"
+
+#include <string>
+
+#include "chrome/common/encrypted_media_messages_android.h"
+#include "content/public/browser/browser_thread.h"
+#include "ipc/ipc_message_macros.h"
+#include "media/base/android/media_codec_bridge.h"
+#include "media/base/android/media_drm_bridge.h"
+
+using content::BrowserThread;
+using media::MediaCodecBridge;
+using media::MediaDrmBridge;
+
+namespace chrome {
+
+// UUID for widevine.
+static const uint8 kWidevineUuid[16] =

[ddorwin, 2013/09/16 17:58:10]

Should this be in widevine_cdm_constants.h? It appears in other key system code.
However, see below.

[qinmin, 2013/09/17 19:21:06]

passing UUID from IPC instead
On 2013/09/16 17:58:10, ddorwin wrote:

+    { 0xED, 0xEF, 0x8B, 0xA9, 0x79, 0xD6, 0x4A, 0xCE,
+      0xA3, 0xC8, 0x27, 0xDC, 0xD5, 0x1D, 0x21, 0xED };
+
+EncryptedMediaMessageFilterAndroid::EncryptedMediaMessageFilterAndroid() {}
+
+EncryptedMediaMessageFilterAndroid::~EncryptedMediaMessageFilterAndroid() {}
+
+bool EncryptedMediaMessageFilterAndroid::OnMessageReceived(
+    const IPC::Message& message, bool* message_was_ok) {
+  DCHECK(BrowserThread::CurrentlyOn(BrowserThread::IO));

[jam, 2013/09/17 20:26:02]

nit: this check isn't necessary, you're checking here that BrowserMessageFilter
is doing its job.. we shouldn't need to test that in every user of it

[qinmin, 2013/09/17 21:26:23]

Done.

+  bool handled = true;
+  IPC_BEGIN_MESSAGE_MAP_EX(
+      EncryptedMediaMessageFilterAndroid, message, *message_was_ok)
+    IPC_MESSAGE_HANDLER(ChromeViewHostMsg_GetSupportedKeySystems,
+                        OnGetSupportedKeySystems)
+    IPC_MESSAGE_UNHANDLED(handled = false)
+  IPC_END_MESSAGE_MAP_EX()
+  return handled;
+}
+
+void EncryptedMediaMessageFilterAndroid::OnGetSupportedKeySystems(

[ddorwin, 2013/09/16 17:58:10]

We should think about the best way to poll for the information the renderer
wants. For example, list of UUIDs and bitmask of codecs. These might also appear
in the output.

[qinmin, 2013/09/17 19:21:06]

Ok. passing the UUID and SupportedCodecs in the IPC. and moving SupportedCodecs
to widevine_cdm_constants.h

On 2013/09/16 17:58:10, ddorwin wrote:

+    std::vector<SupportedKeySystem>* result) {
+  if (!MediaDrmBridge::IsAvailable() || !MediaCodecBridge::IsAvailable())
+    return;
+
+  std::vector<uint8> widevine_uuid(kWidevineUuid, kWidevineUuid + 16);

[ddorwin, 2013/09/16 17:58:10]

Why does this code need to know about Widevine? Shouldn't the key systems code
ask about specific key systems?

[qinmin, 2013/09/17 21:26:23]

Passing UUIDs in IPC instead, so no more widevine keywords here.

On 2013/09/16 17:58:10, ddorwin wrote:

+  if (!MediaDrmBridge::IsCryptoSchemeSupported(widevine_uuid))
+    return;
+
+  // Treat secure decoders with a separate key system due to video composition.
+  SupportedKeySystem video_composition_enabled_system;
+  SupportedKeySystem video_composition_disabled_system;
+  std::vector<MediaCodecBridge::CodecsInfo> codecs_info;
+  MediaCodecBridge::GetCodecsInfo(&codecs_info);
+  for (unsigned i = 0; i < codecs_info.size(); ++i) {
+    video_composition_enabled_system.codecs.push_back(codecs_info[i].codecs);

[ddorwin, 2013/09/16 17:58:10]

We can't just all all codecs because the platform currently only supports ISO
CENC. Passing VP8, etc. up would lead to unexpected results.

[qinmin, 2013/09/17 21:26:23]

The renderer will send a list of available codecs here, so if vp8 is not listed,
then we don't need to check it.
On 2013/09/16 17:58:10, ddorwin wrote:

+    if (codecs_info[i].secure_decoder_supported)
+      video_composition_disabled_system.codecs.push_back(codecs_info[i].codecs);

[ddorwin, 2013/09/16 17:58:10]

TODO/bug to check whether composition is actually supported.

[qinmin, 2013/09/17 21:26:23]

Done.

+  }
+
+  if (video_composition_enabled_system.codecs.size() > 0) {
+    video_composition_enabled_system.uuid = widevine_uuid;
+    video_composition_enabled_system.key_system = "com.widevine.alpha";

[ddorwin, 2013/09/16 17:58:10]

These two strings should not be known at this level IMO. See comment in struct
definition.

[qinmin, 2013/09/17 19:21:06]

Done.

+    video_composition_enabled_system.parent_key_system = "com.widevine";
+    video_composition_enabled_system.video_composition_enabled = true;
+    result->push_back(video_composition_enabled_system);
+  }
+
+  if (video_composition_disabled_system.codecs.size() > 0) {
+    // TODO(qinmin): we may use a different key_system for non-compositable
+    // video.
+    video_composition_disabled_system.uuid = widevine_uuid;
+    video_composition_disabled_system.key_system = "com.widevine.alpha";
+    video_composition_enabled_system.parent_key_system = "com.widevine";
+    video_composition_disabled_system.video_composition_enabled = false;
+    result->push_back(video_composition_disabled_system);
+  }
+}
+
+}  // namespace chrome