MediaStream audio object graph untangling and clean-ups.

content/renderer/media/webrtc_audio_capturer.h

 // Copyright (c) 2012 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_RENDERER_MEDIA_WEBRTC_AUDIO_CAPTURER_H_
 #define CONTENT_RENDERER_MEDIA_WEBRTC_AUDIO_CAPTURER_H_
 
 #include <list>
 #include <string>
 
 #include "base/callback.h"
 #include "base/files/file.h"
 #include "base/macros.h"
 #include "base/memory/ref_counted.h"
+#include "base/memory/scoped_ptr.h"
+#include "base/memory/weak_ptr.h"
 #include "base/synchronization/lock.h"
 #include "base/threading/thread_checker.h"
 #include "base/time/time.h"
 #include "content/common/media/media_stream_options.h"
+#include "content/renderer/media/media_stream_audio_level_calculator.h"
 #include "content/renderer/media/tagged_list.h"
 #include "media/audio/audio_input_device.h"
 #include "media/base/audio_capturer_source.h"
 #include "third_party/WebKit/public/platform/WebMediaConstraints.h"
 
 namespace media {
 class AudioBus;
 }
 
 namespace content {
 
 class MediaStreamAudioProcessor;
 class MediaStreamAudioSource;
 class WebRtcAudioDeviceImpl;
 class WebRtcLocalAudioRenderer;
 class WebRtcLocalAudioTrack;
 
 // This class manages the capture data flow by getting data from its
 // |source_|, and passing it to its |tracks_|.
 // The threading model for this class is rather complex since it will be
 // created on the main render thread, captured data is provided on a dedicated
 // AudioInputDevice thread, and methods can be called either on the Libjingle
 // thread or on the main render thread but also other client threads
 // if an alternative AudioCapturerSource has been set.
 class CONTENT_EXPORT WebRtcAudioCapturer
-    : public base::RefCountedThreadSafe<WebRtcAudioCapturer>,
-      NON_EXPORTED_BASE(public media::AudioCapturerSource::CaptureCallback) {
+    : NON_EXPORTED_BASE(public media::AudioCapturerSource::CaptureCallback) {
  public:
   // Used to construct the audio capturer. |render_frame_id| specifies the
   // RenderFrame consuming audio for capture; -1 is used for tests.
   // |device_info| contains all the device information that the capturer is
   // created for. |constraints| contains the settings for audio processing.
   // TODO(xians): Implement the interface for the audio source and move the
   // |constraints| to ApplyConstraints(). Called on the main render thread.
-  static scoped_refptr<WebRtcAudioCapturer> CreateCapturer(
+  static scoped_ptr<WebRtcAudioCapturer> CreateCapturer(
       int render_frame_id,
       const StreamDeviceInfo& device_info,
       const blink::WebMediaConstraints& constraints,
       WebRtcAudioDeviceImpl* audio_device,
       MediaStreamAudioSource* audio_source);
 
+  ~WebRtcAudioCapturer() override;
+
   // Add a audio track to the sinks of the capturer.
   // WebRtcAudioDeviceImpl calls this method on the main render thread but
   // other clients may call it from other threads. The current implementation
   // does not support multi-thread calling.
   // The first AddTrack will implicitly trigger the Start() of this object.
   void AddTrack(WebRtcLocalAudioTrack* track);
 
   // Remove a audio track from the sinks of the capturer.
   // If the track has been added to the capturer, it  must call RemoveTrack()
   // before it goes away.
   // Called on the main render thread or libjingle working thread.
   void RemoveTrack(WebRtcLocalAudioTrack* track);
 
   // Called when a stream is connecting to a peer connection. This will set
   // up the native buffer size for the stream in order to optimize the
   // performance for peer connection.
   void EnablePeerConnectionMode();
 
   // Volume APIs used by WebRtcAudioDeviceImpl.
   // Called on the AudioInputDevice audio thread.
   void SetVolume(int volume);
   int Volume() const;
   int MaxVolume() const;
 
   // Audio parameters utilized by the source of the audio capturer.
   // TODO(phoglund): Think over the implications of this accessor and if we can
   // remove it.
-  media::AudioParameters source_audio_parameters() const;
+  media::AudioParameters GetInputFormat() const;

[mcasas, 2016/02/26 01:28:19]

One thing that surprised me of audio processing
vs video is that the parameters of the audio stream
are passed partially in a preamble (as AudioParameters)
and partially with the AudioBus (inside AudioBus), and
there's still an overarching AudioBuffer that transports
all necessary data with each blob of audio, but a copy
is needed from AudioBus to AudioBuffer. Is AudioBuffer
the future? 

[miu, 2016/02/27 03:46:37]

I've never used AudioBuffer.  It's sole purpose seems to be to copy audio data
once, hold it for some time, and then copy it again.  That seems like an
anti-pattern to me, but perhaps there's a rationale I'm missing.

<rant>IMO, the purpose of copying data should be to share data, and only when it
would otherwise be impractical to guarantee the original data remains unchanged
while being shared.  Fast machines have allowed developers to become lazy w/ all
the copying and re-copying...</rant>

I'd rather see AudioBus become a little more like VideoFrame, where the audio
format is available as metadata alongside the audio samples.  Then, perhaps we
could all agree to use AudioBus everywhere and kill AudioBuffer and other
"solutions."

 
-  // Gets information about the paired output device. Returns true if such a
-  // device exists.
-  bool GetPairedOutputParameters(int* session_id,
-                                 int* output_sample_rate,
-                                 int* output_frames_per_buffer) const;
-
-  const std::string& device_id() const { return device_info_.device.id; }
-  int session_id() const { return device_info_.session_id; }
+  const StreamDeviceInfo& device_info() const { return device_info_; }
 
   // Stops recording audio. This method will empty its track lists since
   // stopping the capturer will implicitly invalidate all its tracks.
   // This method is exposed to the public because the MediaStreamAudioSource can
   // call Stop()
   void Stop();
 
   // Returns the output format.
   // Called on the main render thread.
   media::AudioParameters GetOutputFormat() const;
 
   // Used by clients to inject their own source to the capturer.
   void SetCapturerSource(
       const scoped_refptr<media::AudioCapturerSource>& source,
       media::AudioParameters params);
 
- protected:
-  friend class base::RefCountedThreadSafe<WebRtcAudioCapturer>;
-  ~WebRtcAudioCapturer() override;
-
  private:
   class TrackOwner;
   typedef TaggedList<TrackOwner> TrackList;
 
   WebRtcAudioCapturer(int render_frame_id,
                       const StreamDeviceInfo& device_info,
                       const blink::WebMediaConstraints& constraints,
                       WebRtcAudioDeviceImpl* audio_device,
                       MediaStreamAudioSource* audio_source);
 
@@ skipping 10 matching lines @@
   bool Initialize();
 
   // SetCapturerSourceInternal() is called if the client on the source side
   // desires to provide their own captured audio data. Client is responsible
   // for calling Start() on its own source to get the ball rolling.
   // Called on the main render thread.
   // buffer_size is optional. Set to 0 to let it be chosen automatically.
   void SetCapturerSourceInternal(
       const scoped_refptr<media::AudioCapturerSource>& source,
       media::ChannelLayout channel_layout,
-      int sample_rate,
-      int buffer_size);
+      int sample_rate);
 
   // Starts recording audio.
   // Triggered by AddSink() on the main render thread or a Libjingle working
   // thread. It should NOT be called under |lock_|.
   void Start();
 
   // Helper function to get the buffer size based on |peer_connection_mode_|
   // and sample rate;
   int GetBufferSize(int sample_rate) const;
 
@@ skipping 10 matching lines @@
   TrackList tracks_;
 
   // The audio data source from the browser process.
   scoped_refptr<media::AudioCapturerSource> source_;
 
   // Cached audio constraints for the capturer.
   blink::WebMediaConstraints constraints_;
 
   // Audio processor doing processing like FIFO, AGC, AEC and NS. Its output
   // data is in a unit of 10 ms data chunk.
-  scoped_refptr<MediaStreamAudioProcessor> audio_processor_;
+  const scoped_refptr<MediaStreamAudioProcessor> audio_processor_;
 
   bool running_;
 
   int render_frame_id_;
 
   // Cached information of the device used by the capturer.
   const StreamDeviceInfo device_info_;
 
   // Stores latest microphone volume received in a CaptureData() callback.
   // Range is [0, 255].
   int volume_;
 
   // Flag which affects the buffer size used by the capturer.
   bool peer_connection_mode_;
 
   // Raw pointer to the WebRtcAudioDeviceImpl, which is valid for the lifetime
   // of RenderThread.
   WebRtcAudioDeviceImpl* audio_device_;
 
   // Raw pointer to the MediaStreamAudioSource object that holds a reference
   // to this WebRtcAudioCapturer.
   // Since |audio_source_| is owned by a blink::WebMediaStreamSource object and
   // blink guarantees that the blink::WebMediaStreamSource outlives any
   // blink::WebMediaStreamTrack connected to the source, |audio_source_| is
   // guaranteed to exist as long as a WebRtcLocalAudioTrack is connected to this
   // WebRtcAudioCapturer.
   MediaStreamAudioSource* const audio_source_;
 
+  // Used to calculate the signal level that shows in the UI.
+  MediaStreamAudioLevelCalculator level_calculator_;
+
+  // Provides weak pointers so that the stop callbacks given to
+  // WebRtcLocalAudioTracks can be canceled once they are no longer necessary.
+  base::WeakPtrFactory<WebRtcAudioCapturer> weak_factory_;
+
   DISALLOW_COPY_AND_ASSIGN(WebRtcAudioCapturer);
 };
 
 }  // namespace content
 
 #endif  // CONTENT_RENDERER_MEDIA_WEBRTC_AUDIO_CAPTURER_H_