Support building WebRTC without audio and video for Android

webrtc/sdk/android/src/jni/null_audio_jni.cc

+/*
+ *  Copyright 2017 The WebRTC project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+#include "webrtc/api/peerconnectioninterface.h"
+#include "webrtc/base/bind.h"
+#include "webrtc/base/scoped_ref_ptr.h"
+#include "webrtc/base/thread.h"
+#include "webrtc/call/callfactoryinterface.h"
+// Adding 'nogncheck' to disable the gn include headers check.
+// We don't want this null implementation target depend on audio and video
+// related targets
+#include "webrtc/media/engine/webrtcmediaengine.h"  // nogncheck
+
+namespace webrtc {
+
+// This implementation is used for building WebRTC without audio and video
+// support.

[Taylor Brandstetter, 2017/06/16 22:19:08]

It's odd that this has an implementation of a CreatePeerConnectionFactory
method, which is normally implemented in a build target at a different level
(webrtc/pc/create_pc_factory). This is also related to the "video/null_audio"
combination not working. So it would be nice to have a comment summarizing the
current state and describing what would need to be done to fix it.

But if it's not too much extra work, I think it would make sense to have:

- video_jni: Real implementation of CreateVideo[En|De]coderFactory
- null_video_jni: "Null" implementation of CreateVideo[En|De]coderFactory
- audio_jni: Real implementation of CreateAudio[En|De]coderFactory, similar to
video_jni
- null_audio_jni: "Null" implementation of CreateAudio[En|De]coderFactory
- media_jni: Implementation of nativeCreatePeerConnectionFactory that creates
media engine and call factory, and calls CreateModularPeerConnectionFactory.
- null_media_jni: Implementation of nativeCreatePeerConnectionFactory that
passes null pointers into CreateModularPeerConnectionFactory.

Then that limitation would be fixed, and we'd have a slightly cleaner version of
this CL. Though maybe there are some issues I haven't considered that would make
this difficult.

[Zhi Huang, 2017/06/17 01:46:43]

Instead of moving entire nativeCreatePeerConnectionFactory to media_jni.cc, I
made a helper function to create native peerconnectionfactory for two reasons:

1.It is simpler and less code duplication.

2. nativeCreatePeerConnectionFactory relies on some method and static global
variables in peerconnection_jni.cc. It requires some extra work to move
everything. 

[Taylor Brandstetter, 2017/06/17 04:41:40]

That makes sense; it's still the same general idea, just an extra level of
indirection.

+rtc::scoped_refptr<PeerConnectionFactoryInterface> CreatePeerConnectionFactory(
+    rtc::Thread* network_thread,
+    rtc::Thread* worker_thread,
+    rtc::Thread* signaling_thread,
+    AudioDeviceModule* default_adm,
+    cricket::WebRtcVideoEncoderFactory* encoder_factory,
+    cricket::WebRtcVideoDecoderFactory* decoder_factory) {
+  return CreateModularPeerConnectionFactory(
+      network_thread, worker_thread, signaling_thread, default_adm,
+      nullptr /*audio_encoder_factory*/, nullptr /*audio_decoder_factory*/,
+      encoder_factory, decoder_factory, nullptr /*audio_mixer*/,
+      std::unique_ptr<cricket::MediaEngineInterface>(),
+      std::unique_ptr<CallFactoryInterface>(),
+      std::unique_ptr<RtcEventLogFactoryInterface>());
+}
+
+}  // namespace webrtc