Add the ability to read/write to WAV files in FakeAudioDevice

webrtc/test/fake_audio_device.cc

Index: webrtc/test/fake_audio_device.cc
diff --git a/webrtc/test/fake_audio_device.cc b/webrtc/test/fake_audio_device.cc
index 623ff518d7b0e9f3aba9e0e98e4d0505f77dbb3f..ad7e49bf0471735b8cdef7be146b169fdbdf9113 100644
--- a/webrtc/test/fake_audio_device.cc
+++ b/webrtc/test/fake_audio_device.cc
@@ -11,10 +11,11 @@
 #include "webrtc/test/fake_audio_device.h"
 
 #include <algorithm>
+#include <utility>
 
-#include "webrtc/base/array_view.h"
 #include "webrtc/base/checks.h"
 #include "webrtc/base/random.h"
+#include "webrtc/common_audio/wav_file.h"
 #include "webrtc/system_wrappers/include/event_wrapper.h"
 
 namespace webrtc {
@@ -24,58 +25,171 @@ namespace {
 constexpr int kFrameLengthMs = 10;
 constexpr int kFramesPerSecond = 1000 / kFrameLengthMs;
 
-}  // namespace
-namespace test {
-
 // Assuming 10ms audio packets..
-class FakeAudioDevice::PulsedNoiseCapturer {
+class PulsedNoiseCapturer final : public test::FakeAudioDevice::Capturer {
  public:
-  PulsedNoiseCapturer(size_t num_samples_per_frame, int16_t max_amplitude)
-      : fill_with_zero_(false),
+  PulsedNoiseCapturer(int16_t max_amplitude, int sampling_frequency_in_hz)
+      : sampling_frequency_in_hz_(sampling_frequency_in_hz),
+        fill_with_zero_(false),
         random_generator_(1),
-        max_amplitude_(max_amplitude),
-        random_audio_(num_samples_per_frame),
-        silent_audio_(num_samples_per_frame, 0) {
+        max_amplitude_(max_amplitude) {
     RTC_DCHECK_GT(max_amplitude, 0);
   }
 
-  rtc::ArrayView<const int16_t> Capture() {
+  int SamplingFrequency() const override {
+    return sampling_frequency_in_hz_;
+  }
+
+  bool Capture(rtc::BufferT<int16_t>* buffer) override {
     fill_with_zero_ = !fill_with_zero_;
-    if (!fill_with_zero_) {
-      std::generate(random_audio_.begin(), random_audio_.end(), [&]() {
-        return random_generator_.Rand(-max_amplitude_, max_amplitude_);
-      });
-    }
-    return fill_with_zero_ ? silent_audio_ : random_audio_;
+    buffer->SetData(buffer->size(), [&](rtc::ArrayView<int16_t> data) {
+      if (fill_with_zero_) {
+        std::fill(data.begin(), data.end(), 0);
+      } else {
+        std::generate(data.begin(), data.end(), [&]() {
+          return random_generator_.Rand(-max_amplitude_, max_amplitude_);
+        });
+      }
+      return data.size();
+    });
+    return true;
   }
 
  private:
+  int sampling_frequency_in_hz_;
   bool fill_with_zero_;
   Random random_generator_;
   const int16_t max_amplitude_;
-  std::vector<int16_t> random_audio_;
-  std::vector<int16_t> silent_audio_;
 };
 
-FakeAudioDevice::FakeAudioDevice(float speed,
-                                 int sampling_frequency_in_hz,
-                                 int16_t max_amplitude)
-    : sampling_frequency_in_hz_(sampling_frequency_in_hz),
-      num_samples_per_frame_(
-          rtc::CheckedDivExact(sampling_frequency_in_hz_, kFramesPerSecond)),
+class WavFileReader final : public test::FakeAudioDevice::Capturer {
+ public:
+  WavFileReader(std::string filename, int sampling_frequency_in_hz)
+      : sampling_frequency_in_hz_(sampling_frequency_in_hz),
+        wav_reader_(filename) {
+    RTC_CHECK_EQ(wav_reader_.sample_rate(), sampling_frequency_in_hz);
+    RTC_CHECK_EQ(wav_reader_.num_channels(), 1);
+  }
+
+  int SamplingFrequency() const override {
+    return sampling_frequency_in_hz_;
+  }
+
+  bool Capture(rtc::BufferT<int16_t>* buffer) override {
+    buffer->SetData(buffer->size(), [&](rtc::ArrayView<int16_t> data) {
+      return wav_reader_.ReadSamples(data.size(), data.data());
+    });
+    return buffer->size() > 0;
+  }
+
+ private:
+  int sampling_frequency_in_hz_;
+  WavReader wav_reader_;
+};
+
+class WavFileWriter final : public test::FakeAudioDevice::Renderer {
+ public:
+  WavFileWriter(std::string filename, int sampling_frequency_in_hz)
+      : sampling_frequency_in_hz_(sampling_frequency_in_hz),
+        wav_writer_(filename, sampling_frequency_in_hz, 1) {}
+
+  int SamplingFrequency() const override {
+    return sampling_frequency_in_hz_;
+  }
+
+  bool Render(rtc::ArrayView<const int16_t> data) override {
+    wav_writer_.WriteSamples(data.data(), data.size());
+    return true;
+  }
+
+ private:
+  int sampling_frequency_in_hz_;
+  WavWriter wav_writer_;
+};
+
+class DiscardRenderer final : public test::FakeAudioDevice::Renderer {
+ public:
+  DiscardRenderer(int sampling_frequency_in_hz)
+      : sampling_frequency_in_hz_(sampling_frequency_in_hz) {}
+
+  int SamplingFrequency() const override {
+    return sampling_frequency_in_hz_;
+  }
+
+  bool Render(rtc::ArrayView<const int16_t> data) override {
+    return true;
+  }
+
+ private:
+  int sampling_frequency_in_hz_;
+};
+
+}  // namespace
+namespace test {
+
+size_t FakeAudioDevice::SamplesPerFrame(int sampling_frequency_in_hz) {
+  return rtc::CheckedDivExact(sampling_frequency_in_hz, kFramesPerSecond);
+}
+
+std::unique_ptr<FakeAudioDevice::Capturer>
+    FakeAudioDevice::CreatePulsedNoiseCapturer(
+        int16_t max_amplitude, int sampling_frequency_in_hz) {
+  return std::unique_ptr<FakeAudioDevice::Capturer>(
+      new PulsedNoiseCapturer(max_amplitude, sampling_frequency_in_hz));
+}
+
+std::unique_ptr<FakeAudioDevice::Capturer> FakeAudioDevice::CreateWavFileReader(
+    std::string filename, int sampling_frequency_in_hz) {
+  return std::unique_ptr<FakeAudioDevice::Capturer>(
+      new WavFileReader(filename, sampling_frequency_in_hz));
+}
+
+std::unique_ptr<FakeAudioDevice::Capturer> FakeAudioDevice::CreateWavFileReader(
+    std::string filename) {
+  int sampling_frequency_in_hz = WavReader(filename).sample_rate();
+  return std::unique_ptr<FakeAudioDevice::Capturer>(
+      new WavFileReader(filename, sampling_frequency_in_hz));
+}
+
+std::unique_ptr<FakeAudioDevice::Renderer> FakeAudioDevice::CreateWavFileWriter(
+    std::string filename, int sampling_frequency_in_hz) {
+  return std::unique_ptr<FakeAudioDevice::Renderer>(
+      new WavFileWriter(filename, sampling_frequency_in_hz));
+}
+
+std::unique_ptr<FakeAudioDevice::Renderer>
+    FakeAudioDevice::CreateDiscardRenderer(int sampling_frequency_in_hz) {
+  return std::unique_ptr<FakeAudioDevice::Renderer>(
+      new DiscardRenderer(sampling_frequency_in_hz));
+}
+
+
+FakeAudioDevice::FakeAudioDevice(std::unique_ptr<Capturer> capturer,
+                                 std::unique_ptr<Renderer> renderer,
+                                 float speed)
+    : capturer_(std::move(capturer)),
+      renderer_(std::move(renderer)),
       speed_(speed),
       audio_callback_(nullptr),
       rendering_(false),
       capturing_(false),
-      capturer_(new FakeAudioDevice::PulsedNoiseCapturer(num_samples_per_frame_,
-                                                         max_amplitude)),
-      playout_buffer_(num_samples_per_frame_, 0),
+      done_rendering_(true, true),
+      done_capturing_(true, true),
       tick_(EventTimerWrapper::Create()),
       thread_(FakeAudioDevice::Run, this, "FakeAudioDevice") {
-  RTC_DCHECK(
-      sampling_frequency_in_hz == 8000 || sampling_frequency_in_hz == 16000 ||
-      sampling_frequency_in_hz == 32000 || sampling_frequency_in_hz == 44100 ||
-      sampling_frequency_in_hz == 48000);
+  if (renderer_) {
+    const int sample_rate = renderer_->SamplingFrequency();
+    playout_buffer_.resize(SamplesPerFrame(sample_rate), 0);
+    RTC_CHECK(
+        sample_rate == 8000 || sample_rate == 16000 || sample_rate == 32000 ||
+        sample_rate == 44100 || sample_rate == 48000);
+  }
+  if (capturer_) {
+    const int sample_rate = capturer_->SamplingFrequency();
+    RTC_CHECK(
+        sample_rate == 8000 || sample_rate == 16000 || sample_rate == 32000 ||
+        sample_rate == 44100 || sample_rate == 48000);

[kwiberg-webrtc, 2017/03/13 14:22:28]

Do the CHECK first in this function, so you don't have to write that mouthful
twice? You don't have to bend over backwards to not trigger a check for illegal
sample_rate if both renderer_ and capturer_ are null...

[oprypin_webrtc, 2017/03/13 15:11:36]

I'm checking two different sample rates.

[kwiberg-webrtc, 2017/03/14 10:02:03]

Right, I would have seen that if I hadn't been so sloppy when reading the
code... How about a subroutine then?

  auto good_sample_rate = [](int sr){
    return sr == 8000 || sr == 16000 || sr == 32000
        || sr == 44100 || sr == 48000;
  }
  [...]
  RTC_CHECK(good_sample_rate(renderer_->SamplingFrequency()));
  [...]

[oprypin_webrtc, 2017/03/14 11:58:13]

Done.

+  }
 }
 
 FakeAudioDevice::~FakeAudioDevice() {
@@ -86,25 +200,31 @@ FakeAudioDevice::~FakeAudioDevice() {
 
 int32_t FakeAudioDevice::StartPlayout() {
   rtc::CritScope cs(&lock_);
+  RTC_CHECK(renderer_);
   rendering_ = true;
+  done_rendering_.Reset();
   return 0;
 }
 
 int32_t FakeAudioDevice::StopPlayout() {
   rtc::CritScope cs(&lock_);
   rendering_ = false;
+  done_rendering_.Set();
   return 0;
 }
 
 int32_t FakeAudioDevice::StartRecording() {
   rtc::CritScope cs(&lock_);
+  RTC_CHECK(capturer_);
   capturing_ = true;
+  done_capturing_.Reset();
   return 0;
 }
 
 int32_t FakeAudioDevice::StopRecording() {
   rtc::CritScope cs(&lock_);
   capturing_ = false;
+  done_capturing_.Set();
   return 0;
 }
 
@@ -117,7 +237,7 @@ int32_t FakeAudioDevice::Init() {
 
 int32_t FakeAudioDevice::RegisterAudioCallback(AudioTransport* callback) {
   rtc::CritScope cs(&lock_);
-  RTC_DCHECK(callback || audio_callback_ != nullptr);
+  RTC_DCHECK(callback || audio_callback_);
   audio_callback_ = callback;
   return 0;
 }
@@ -132,6 +252,14 @@ bool FakeAudioDevice::Recording() const {
   return capturing_;
 }
 
+bool FakeAudioDevice::WaitForPlayoutEnd(int timeout_ms) {
+  return done_rendering_.Wait(timeout_ms);
+}
+
+bool FakeAudioDevice::WaitForRecordingEnd(int timeout_ms) {
+  return done_capturing_.Wait(timeout_ms);
+}
+
 bool FakeAudioDevice::Run(void* obj) {
   static_cast<FakeAudioDevice*>(obj)->ProcessAudio();
   return true;
@@ -142,19 +270,34 @@ void FakeAudioDevice::ProcessAudio() {
     rtc::CritScope cs(&lock_);
     if (capturing_) {
       // Capture 10ms of audio. 2 bytes per sample.
-      rtc::ArrayView<const int16_t> audio_data = capturer_->Capture();
-      uint32_t new_mic_level = 0;
-      audio_callback_->RecordedDataIsAvailable(
-          audio_data.data(), audio_data.size(), 2, 1, sampling_frequency_in_hz_,
-          0, 0, 0, false, new_mic_level);
+      recording_buffer_.SetSize(
+          SamplesPerFrame(capturer_->SamplingFrequency()));
+      const bool keep_capturing = capturer_->Capture(&recording_buffer_);
+      uint32_t new_mic_level;
+      if (recording_buffer_.size() > 0) {

[kwiberg-webrtc, 2017/03/13 14:22:27]

This condition is always true, right?

(You could use SetData with a lambda here, if you wanted.)

[oprypin_webrtc, 2017/03/13 15:11:36]

It is not true at the end of the recording. Don't want to "bother" the receiver
when there is actually no data.

[kwiberg-webrtc, 2017/03/14 10:02:03]

On line 273, you SetSize() the buffer to
SamplesPerFrame(capturer_->SamplingFrequency()). Are you saying this number can
be zero?

[oprypin_webrtc, 2017/03/14 11:58:13]

No. The Capture method is free to change the buffer's size, possibly even zero
size at the end of the recording.

[kwiberg-webrtc, 2017/03/14 13:46:47]

Ah, right! Sorry for being dense.

However, since Capture will discard the contents of recording_buffer_ and fill
it with however many elements it wants, the SetSize call on line 273 is entirely
redundant, isn't it?

[oprypin_webrtc, 2017/03/14 14:01:40]

Yes, I removed that line in a later patchset.

In this patchset it implemented that weird behavior that we agreed to remove:
"The final size of the buffer should not exceed its initial size, which is equal
to SamplesPerFrame."

[kwiberg-webrtc, 2017/03/14 14:08:56]

Acknowledged.

+        audio_callback_->RecordedDataIsAvailable(
+            recording_buffer_.data(), recording_buffer_.size(), 2, 1,
+            capturer_->SamplingFrequency(), 0, 0, 0, false, new_mic_level);
+      }
+      if (!keep_capturing) {
+        capturing_ = false;
+        done_capturing_.Set();
+      }
     }
     if (rendering_) {
-      size_t samples_out = 0;
-      int64_t elapsed_time_ms = -1;
-      int64_t ntp_time_ms = -1;
+      size_t samples_out;
+      int64_t elapsed_time_ms;
+      int64_t ntp_time_ms;
+      const int sampling_frequency = renderer_->SamplingFrequency();
       audio_callback_->NeedMorePlayData(
-          num_samples_per_frame_, 2, 1, sampling_frequency_in_hz_,
+          SamplesPerFrame(sampling_frequency), 2, 1, sampling_frequency,
           playout_buffer_.data(), samples_out, &elapsed_time_ms, &ntp_time_ms);
+      const bool keep_rendering = renderer_->Render(
+          rtc::ArrayView<const int16_t>(playout_buffer_.data(), samples_out));
+      if (!keep_rendering) {
+        rendering_ = false;
+        done_rendering_.Set();
+      }
     }
   }
   tick_->Wait(WEBRTC_EVENT_INFINITE);