Add ProxyDecryptor which wraps concrete Decryptor implementations.

webkit/media/crypto/key_systems.h

Index: webkit/media/crypto/key_systems.h
diff --git a/webkit/media/key_systems.h b/webkit/media/crypto/key_systems.h
similarity index 80%
rename from webkit/media/key_systems.h
rename to webkit/media/crypto/key_systems.h
index dedd279e0184accdee5f5a08ff9c29e75be6c680..4271244724d7ce3e1c1cd7d46a59489cdc8da1db 100644
--- a/webkit/media/key_systems.h
+++ b/webkit/media/crypto/key_systems.h
@@ -2,8 +2,8 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
-#ifndef WEBKIT_MEDIA_KEY_SYSTEMS_H_
-#define WEBKIT_MEDIA_KEY_SYSTEMS_H_
+#ifndef WEBKIT_MEDIA_CRYPTO_KEY_SYSTEMS_H_
+#define WEBKIT_MEDIA_CRYPTO_KEY_SYSTEMS_H_
 
 #include <string>
 #include <vector>
@@ -14,6 +14,8 @@ class WebString;
 
 namespace webkit_media {
 
+const char kClearKeyKeySystem[] = "webkit-org.w3.clearkey";

[scherkus (not reviewing), 2012/06/27 01:10:18]

this can be declared here but initialized in the .cc

[ddorwin, 2012/06/27 04:07:15]

This is only used to determine the CDM. We eventually want a static registry
that tells us what to create. Rather than move this string and expose it
elsewhere, I suggest adding UseAesDecryptor(string) or something to this file.
(I'm not exactly sure how we will specify AES vs. external CDMs, but this seems
close enough for now.)

[scherkus (not reviewing), 2012/06/27 04:22:23]

Could you explain the UseAesDecryptor() approach a bit more? Not sure I
understand.

Also is this something that needs to be done now? Seems like something that
could wait until we have a proper registry.

[ddorwin, 2012/06/27 05:24:41]

The string that is being exposed is only used in the .cc file and
proxy_decryptor.cc line 33, which will use this registry in the future. However,
AES decryptor/clear key will be different than most/all items in the registry,
so the registry might need something like UseAesDecryptor() anyway.

If we expose the necessary function, we don't need to expose the string.
It could wait, but I'd rather not expose the string now (and allow other uses to
get added). Exposing a potentially temporary function seems better than exposing
a string that should not be exposed.

[scherkus (not reviewing), 2012/06/27 05:34:11]

I don't think it's really that big of an issue but if you feel strongly about it
I'd be OK implementing a crude registry now

Considering ProxyDecryptor only references a Decryptor (and not an AesDecryptor)
we can:
  1) define a function here that returns a Decryptor or NULL given a key system
  2) update ProxyDecryptor to know nothing about clear key

WDYT?

[xhwang, 2012/06/27 21:41:26]

Done.

+
 // Returns whether |key_sytem| is supported at all.
 // Call IsSupportedKeySystemWithMediaMimeType() to determine whether a
 // |key_system| supports a specific type of media.
@@ -27,4 +29,4 @@ bool IsSupportedKeySystemWithMediaMimeType(
 
 }  // namespace webkit_media
 
-#endif  // WEBKIT_MEDIA_KEY_SYSTEMS_H_
+#endif  // WEBKIT_MEDIA_CRYPTO_KEY_SYSTEMS_H_