DXR is a code search and navigation tool aimed at making sense of large projects. It supports full-text and regex searches as well as structural queries.

Implementation

Mercurial (8aa8bbbf0bee)

VCS Links

Line Code
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this file,
 * You can obtain one at http://mozilla.org/MPL/2.0/. */

#ifndef MOZILLA_AUDIOSEGMENT_H_
#define MOZILLA_AUDIOSEGMENT_H_

#include "MediaSegment.h"
#include "AudioSampleFormat.h"
#include "AudioChannelFormat.h"
#include "SharedBuffer.h"
#include "WebAudioUtils.h"
#ifdef MOZILLA_INTERNAL_API
#  include "mozilla/TimeStamp.h"
#endif
#include <float.h>

namespace mozilla {
struct AudioChunk;
class AudioSegment;
}  // namespace mozilla
DECLARE_USE_COPY_CONSTRUCTORS(mozilla::AudioChunk)

/**
 * This allows compilation of nsTArray<AudioSegment> and
 * AutoTArray<AudioSegment> since without it, static analysis fails on the
 * mChunks member being a non-memmovable AutoTArray.
 *
 * Note that AudioSegment(const AudioSegment&) is deleted, so this should
 * never come into effect.
 */
DECLARE_USE_COPY_CONSTRUCTORS(mozilla::AudioSegment)

namespace mozilla {

template <typename T>
class SharedChannelArrayBuffer : public ThreadSharedObject {
 public:
  explicit SharedChannelArrayBuffer(nsTArray<nsTArray<T> >* aBuffers) {
    mBuffers.SwapElements(*aBuffers);
  }

  size_t SizeOfExcludingThis(MallocSizeOf aMallocSizeOf) const override {
    size_t amount = 0;
    amount += mBuffers.ShallowSizeOfExcludingThis(aMallocSizeOf);
    for (size_t i = 0; i < mBuffers.Length(); i++) {
      amount += mBuffers[i].ShallowSizeOfExcludingThis(aMallocSizeOf);
    }

    return amount;
  }

  size_t SizeOfIncludingThis(MallocSizeOf aMallocSizeOf) const override {
    return aMallocSizeOf(this) + SizeOfExcludingThis(aMallocSizeOf);
  }

  nsTArray<nsTArray<T> > mBuffers;
};

class AudioMixer;

/**
 * For auto-arrays etc, guess this as the common number of channels.
 */
const int GUESS_AUDIO_CHANNELS = 2;

// We ensure that the graph advances in steps that are multiples of the Web
// Audio block size
const uint32_t WEBAUDIO_BLOCK_SIZE_BITS = 7;
const uint32_t WEBAUDIO_BLOCK_SIZE = 1 << WEBAUDIO_BLOCK_SIZE_BITS;

template <typename SrcT, typename DestT>
static void InterleaveAndConvertBuffer(const SrcT* const* aSourceChannels,
                                       uint32_t aLength, float aVolume,
                                       uint32_t aChannels, DestT* aOutput) {
  DestT* output = aOutput;
  for (size_t i = 0; i < aLength; ++i) {
    for (size_t channel = 0; channel < aChannels; ++channel) {
      float v = AudioSampleToFloat(aSourceChannels[channel][i]) * aVolume;
      *output = FloatToAudioSample<DestT>(v);
      ++output;
    }
  }
}

template <typename SrcT, typename DestT>
static void DeinterleaveAndConvertBuffer(const SrcT* aSourceBuffer,
                                         uint32_t aFrames, uint32_t aChannels,
                                         DestT** aOutput) {
  for (size_t i = 0; i < aChannels; i++) {
    size_t interleavedIndex = i;
    for (size_t j = 0; j < aFrames; j++) {
      ConvertAudioSample(aSourceBuffer[interleavedIndex], aOutput[i][j]);
      interleavedIndex += aChannels;
    }
  }
}

class SilentChannel {
 public:
  static const int AUDIO_PROCESSING_FRAMES = 640; /* > 10ms of 48KHz audio */
  static const uint8_t
      gZeroChannel[MAX_AUDIO_SAMPLE_SIZE * AUDIO_PROCESSING_FRAMES];
  // We take advantage of the fact that zero in float and zero in int have the
  // same all-zeros bit layout.
  template <typename T>
  static const T* ZeroChannel();
};

/**
 * Given an array of input channels (aChannelData), downmix to aOutputChannels,
 * interleave the channel data. A total of aOutputChannels*aDuration
 * interleaved samples will be copied to a channel buffer in aOutput.
 */
template <typename SrcT, typename DestT>
void DownmixAndInterleave(const nsTArray<const SrcT*>& aChannelData,
                          int32_t aDuration, float aVolume,
                          uint32_t aOutputChannels, DestT* aOutput) {
  if (aChannelData.Length() == aOutputChannels) {
    InterleaveAndConvertBuffer(aChannelData.Elements(), aDuration, aVolume,
                               aOutputChannels, aOutput);
  } else {
    AutoTArray<SrcT*, GUESS_AUDIO_CHANNELS> outputChannelData;
    AutoTArray<SrcT,
               SilentChannel::AUDIO_PROCESSING_FRAMES * GUESS_AUDIO_CHANNELS>
        outputBuffers;
    outputChannelData.SetLength(aOutputChannels);
    outputBuffers.SetLength(aDuration * aOutputChannels);
    for (uint32_t i = 0; i < aOutputChannels; i++) {
      outputChannelData[i] = outputBuffers.Elements() + aDuration * i;
    }
    AudioChannelsDownMix(aChannelData, outputChannelData.Elements(),
                         aOutputChannels, aDuration);
    InterleaveAndConvertBuffer(outputChannelData.Elements(), aDuration, aVolume,
                               aOutputChannels, aOutput);
  }
}

/**
 * An AudioChunk represents a multi-channel buffer of audio samples.
 * It references an underlying ThreadSharedObject which manages the lifetime
 * of the buffer. An AudioChunk maintains its own duration and channel data
 * pointers so it can represent a subinterval of a buffer without copying.
 * An AudioChunk can store its individual channels anywhere; it maintains
 * separate pointers to each channel's buffer.
 */
struct AudioChunk {
  typedef mozilla::AudioSampleFormat SampleFormat;

  // Generic methods
  void SliceTo(TrackTime aStart, TrackTime aEnd) {
    MOZ_ASSERT(aStart >= 0 && aStart < aEnd && aEnd <= mDuration,
               "Slice out of bounds");
    if (mBuffer) {
      MOZ_ASSERT(aStart < INT32_MAX,
                 "Can't slice beyond 32-bit sample lengths");
      for (uint32_t channel = 0; channel < mChannelData.Length(); ++channel) {
        mChannelData[channel] = AddAudioSampleOffset(
            mChannelData[channel], mBufferFormat, int32_t(aStart));
      }
    }
    mDuration = aEnd - aStart;
  }
  TrackTime GetDuration() const { return mDuration; }
  bool CanCombineWithFollowing(const AudioChunk& aOther) const {
    if (aOther.mBuffer != mBuffer) {
      return false;
    }
    if (!mBuffer) {
      return true;
    }
    if (aOther.mVolume != mVolume) {
      return false;
    }
    if (aOther.mPrincipalHandle != mPrincipalHandle) {
      return false;
    }
    NS_ASSERTION(aOther.mBufferFormat == mBufferFormat,
                 "Wrong metadata about buffer");
    NS_ASSERTION(aOther.mChannelData.Length() == mChannelData.Length(),
                 "Mismatched channel count");
    if (mDuration > INT32_MAX) {
      return false;
    }
    for (uint32_t channel = 0; channel < mChannelData.Length(); ++channel) {
      if (aOther.mChannelData[channel] !=
          AddAudioSampleOffset(mChannelData[channel], mBufferFormat,
                               int32_t(mDuration))) {
        return false;
      }
    }
    return true;
  }
  bool IsNull() const { return mBuffer == nullptr; }
  void SetNull(TrackTime aDuration) {
    mBuffer = nullptr;
    mChannelData.Clear();
    mDuration = aDuration;
    mVolume = 1.0f;
    mBufferFormat = AUDIO_FORMAT_SILENCE;
    mPrincipalHandle = PRINCIPAL_HANDLE_NONE;
  }

  size_t ChannelCount() const { return mChannelData.Length(); }

  bool IsMuted() const { return mVolume == 0.0f; }

  bool IsAudible() const {
    for (auto&& channel : mChannelData) {
      // Transform sound into dB RMS and assume that the value smaller than -100
      // is inaudible.
      float dbrms = 0.0;
      for (uint32_t idx = 0; idx < mDuration; idx++) {
        dbrms += std::pow(static_cast<const AudioDataValue*>(channel)[idx], 2);
      }
      dbrms /= mDuration;
      dbrms = std::sqrt(dbrms) != 0.0 ? 20 * log10(dbrms) : -1000.0;
      if (dbrms > -100.0) {
        return true;
      }
    }
    return false;
  }

  size_t SizeOfExcludingThisIfUnshared(MallocSizeOf aMallocSizeOf) const {
    return SizeOfExcludingThis(aMallocSizeOf, true);
  }

  size_t SizeOfExcludingThis(MallocSizeOf aMallocSizeOf, bool aUnshared) const {
    size_t amount = 0;

    // Possibly owned:
    // - mBuffer - Can hold data that is also in the decoded audio queue. If it
    //             is not shared, or unshared == false it gets counted.
    if (mBuffer && (!aUnshared || !mBuffer->IsShared())) {
      amount += mBuffer->SizeOfIncludingThis(aMallocSizeOf);
    }

    // Memory in the array is owned by mBuffer.
    amount += mChannelData.ShallowSizeOfExcludingThis(aMallocSizeOf);
    return amount;
  }

  template <typename T>
  const nsTArray<const T*>& ChannelData() const {
    MOZ_ASSERT(AudioSampleTypeToFormat<T>::Format == mBufferFormat);
    return *reinterpret_cast<const AutoTArray<const T*, GUESS_AUDIO_CHANNELS>*>(
        &mChannelData);
  }

  /**
   * ChannelFloatsForWrite() should be used only when mBuffer is owned solely
   * by the calling thread.
   */
  template <typename T>
  T* ChannelDataForWrite(size_t aChannel) {
    MOZ_ASSERT(AudioSampleTypeToFormat<T>::Format == mBufferFormat);
    MOZ_ASSERT(!mBuffer->IsShared());
    return static_cast<T*>(const_cast<void*>(mChannelData[aChannel]));
  }

  const PrincipalHandle& GetPrincipalHandle() const { return mPrincipalHandle; }

  TrackTime mDuration = 0;             // in frames within the buffer
  RefPtr<ThreadSharedObject> mBuffer;  // the buffer object whose lifetime is
                                       // managed; null means data is all zeroes
  // one pointer per channel; empty if and only if mBuffer is null
  AutoTArray<const void*, GUESS_AUDIO_CHANNELS> mChannelData;
  float mVolume = 1.0f;  // volume multiplier to apply
  // format of frames in mBuffer (or silence if mBuffer is null)
  SampleFormat mBufferFormat = AUDIO_FORMAT_SILENCE;
  // principalHandle for the data in this chunk.
  // This can be compared to an nsIPrincipal* when back on main thread.
  PrincipalHandle mPrincipalHandle = PRINCIPAL_HANDLE_NONE;
};

/**
 * A list of audio samples consisting of a sequence of slices of SharedBuffers.
 * The audio rate is determined by the track, not stored in this class.
 */
class AudioSegment : public MediaSegmentBase<AudioSegment, AudioChunk> {
 public:
  typedef mozilla::AudioSampleFormat SampleFormat;

  AudioSegment() : MediaSegmentBase<AudioSegment, AudioChunk>(AUDIO) {}

  AudioSegment(AudioSegment&& aSegment)
      : MediaSegmentBase<AudioSegment, AudioChunk>(std::move(aSegment)) {}

  AudioSegment(const AudioSegment&) = delete;
  AudioSegment& operator=(const AudioSegment&) = delete;

  ~AudioSegment() {}

  // Resample the whole segment in place.
  template <typename T>
  void Resample(SpeexResamplerState* aResampler, uint32_t aInRate,
                uint32_t aOutRate) {
    mDuration = 0;
#ifdef DEBUG
    uint32_t segmentChannelCount = ChannelCount();
#endif

    for (ChunkIterator ci(*this); !ci.IsEnded(); ci.Next()) {
      AutoTArray<nsTArray<T>, GUESS_AUDIO_CHANNELS> output;
      AutoTArray<const T*, GUESS_AUDIO_CHANNELS> bufferPtrs;
      AudioChunk& c = *ci;
      // If this chunk is null, don't bother resampling, just alter its duration
      if (c.IsNull()) {
        c.mDuration = (c.mDuration * aOutRate) / aInRate;
        mDuration += c.mDuration;
        continue;
      }
      uint32_t channels = c.mChannelData.Length();
      MOZ_ASSERT(channels == segmentChannelCount);
      output.SetLength(channels);
      bufferPtrs.SetLength(channels);
      uint32_t inFrames = c.mDuration;
      // Round up to allocate; the last frame may not be used.
      NS_ASSERTION((UINT32_MAX - aInRate + 1) / c.mDuration >= aOutRate,
                   "Dropping samples");
      uint32_t outSize = (c.mDuration * aOutRate + aInRate - 1) / aInRate;
      for (uint32_t i = 0; i < channels; i++) {
        T* out = output[i].AppendElements(outSize);
        uint32_t outFrames = outSize;

        const T* in = static_cast<const T*>(c.mChannelData[i]);
        dom::WebAudioUtils::SpeexResamplerProcess(aResampler, i, in, &inFrames,
                                                  out, &outFrames);
        MOZ_ASSERT(inFrames == c.mDuration);

        bufferPtrs[i] = out;
        output[i].SetLength(outFrames);
      }
      MOZ_ASSERT(channels > 0);
      c.mDuration = output[0].Length();
      c.mBuffer = new mozilla::SharedChannelArrayBuffer<T>(&output);
      for (uint32_t i = 0; i < channels; i++) {
        c.mChannelData[i] = bufferPtrs[i];
      }
      mDuration += c.mDuration;
    }
  }

  void ResampleChunks(SpeexResamplerState* aResampler, uint32_t aInRate,
                      uint32_t aOutRate);
  void AppendFrames(already_AddRefed<ThreadSharedObject> aBuffer,
                    const nsTArray<const float*>& aChannelData,
                    int32_t aDuration,
                    const PrincipalHandle& aPrincipalHandle) {
    AudioChunk* chunk = AppendChunk(aDuration);
    chunk->mBuffer = aBuffer;

    MOZ_ASSERT(chunk->mBuffer || aChannelData.IsEmpty(),
               "Appending invalid data ?");

    for (uint32_t channel = 0; channel < aChannelData.Length(); ++channel) {
      chunk->mChannelData.AppendElement(aChannelData[channel]);
    }
    chunk->mBufferFormat = AUDIO_FORMAT_FLOAT32;
    chunk->mPrincipalHandle = aPrincipalHandle;
  }
  void AppendFrames(already_AddRefed<ThreadSharedObject> aBuffer,
                    const nsTArray<const int16_t*>& aChannelData,
                    int32_t aDuration,
                    const PrincipalHandle& aPrincipalHandle) {
    AudioChunk* chunk = AppendChunk(aDuration);
    chunk->mBuffer = aBuffer;

    MOZ_ASSERT(chunk->mBuffer || aChannelData.IsEmpty(),
               "Appending invalid data ?");

    for (uint32_t channel = 0; channel < aChannelData.Length(); ++channel) {
      chunk->mChannelData.AppendElement(aChannelData[channel]);
    }
    chunk->mBufferFormat = AUDIO_FORMAT_S16;
    chunk->mPrincipalHandle = aPrincipalHandle;
  }
  // Consumes aChunk, and returns a pointer to the persistent copy of aChunk
  // in the segment.
  AudioChunk* AppendAndConsumeChunk(AudioChunk* aChunk) {
    AudioChunk* chunk = AppendChunk(aChunk->mDuration);
    chunk->mBuffer = aChunk->mBuffer.forget();
    chunk->mChannelData.SwapElements(aChunk->mChannelData);

    MOZ_ASSERT(chunk->mBuffer || aChunk->mChannelData.IsEmpty(),
               "Appending invalid data ?");

    chunk->mVolume = aChunk->mVolume;
    chunk->mBufferFormat = aChunk->mBufferFormat;
    chunk->mPrincipalHandle = aChunk->mPrincipalHandle;
    return chunk;
  }
  void ApplyVolume(float aVolume);
  // Mix the segment into a mixer, interleaved. This is useful to output a
  // segment to a system audio callback. It up or down mixes to aChannelCount
  // channels.
  void WriteTo(AudioMixer& aMixer, uint32_t aChannelCount,
               uint32_t aSampleRate);
  // Mix the segment into a mixer, keeping it planar, up or down mixing to
  // aChannelCount channels.
  void Mix(AudioMixer& aMixer, uint32_t aChannelCount, uint32_t aSampleRate);

  int ChannelCount() {
    NS_WARNING_ASSERTION(
        !mChunks.IsEmpty(),
        "Cannot query channel count on a AudioSegment with no chunks.");
    // Find the first chunk that has non-zero channels. A chunk that hs zero
    // channels is just silence and we can simply discard it.
    for (ChunkIterator ci(*this); !ci.IsEnded(); ci.Next()) {
      if (ci->ChannelCount()) {
        return ci->ChannelCount();
      }
    }
    return 0;
  }

  static Type StaticType() { return AUDIO; }

  size_t SizeOfIncludingThis(MallocSizeOf aMallocSizeOf) const override {
    return aMallocSizeOf(this) + SizeOfExcludingThis(aMallocSizeOf);
  }
};

template <typename SrcT>
void WriteChunk(AudioChunk& aChunk, uint32_t aOutputChannels,
                AudioDataValue* aOutputBuffer) {
  AutoTArray<const SrcT*, GUESS_AUDIO_CHANNELS> channelData;

  channelData = aChunk.ChannelData<SrcT>();

  if (channelData.Length() < aOutputChannels) {
    // Up-mix. Note that this might actually make channelData have more
    // than aOutputChannels temporarily.
    AudioChannelsUpMix(&channelData, aOutputChannels,
                       SilentChannel::ZeroChannel<SrcT>());
  }
  if (channelData.Length() > aOutputChannels) {
    // Down-mix.
    DownmixAndInterleave(channelData, aChunk.mDuration, aChunk.mVolume,
                         aOutputChannels, aOutputBuffer);
  } else {
    InterleaveAndConvertBuffer(channelData.Elements(), aChunk.mDuration,
                               aChunk.mVolume, aOutputChannels, aOutputBuffer);
  }
}

}  // namespace mozilla

#endif /* MOZILLA_AUDIOSEGMENT_H_ */