зеркало из https://github.com/mozilla/gecko-dev.git
356 строки
12 KiB
C++
356 строки
12 KiB
C++
/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
|
|
/* vim:set ts=2 sw=2 sts=2 et cindent: */
|
|
/* 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/. */
|
|
|
|
#include "OpusDecoder.h"
|
|
#include "OpusParser.h"
|
|
#include "TimeUnits.h"
|
|
#include "VorbisUtils.h"
|
|
#include "VorbisDecoder.h" // For VorbisLayout
|
|
#include "mozilla/EndianUtils.h"
|
|
#include "mozilla/PodOperations.h"
|
|
#include "mozilla/SyncRunnable.h"
|
|
#include "mozilla/SizePrintfMacros.h"
|
|
|
|
#include <inttypes.h> // For PRId64
|
|
|
|
#include "opus/opus.h"
|
|
extern "C" {
|
|
#include "opus/opus_multistream.h"
|
|
}
|
|
|
|
#define OPUS_DEBUG(arg, ...) MOZ_LOG(sPDMLog, mozilla::LogLevel::Debug, \
|
|
("OpusDataDecoder(%p)::%s: " arg, this, __func__, ##__VA_ARGS__))
|
|
|
|
namespace mozilla {
|
|
|
|
OpusDataDecoder::OpusDataDecoder(const CreateDecoderParams& aParams)
|
|
: mInfo(aParams.AudioConfig())
|
|
, mTaskQueue(aParams.mTaskQueue)
|
|
, mOpusDecoder(nullptr)
|
|
, mSkip(0)
|
|
, mDecodedHeader(false)
|
|
, mPaddingDiscarded(false)
|
|
, mFrames(0)
|
|
{
|
|
}
|
|
|
|
OpusDataDecoder::~OpusDataDecoder()
|
|
{
|
|
if (mOpusDecoder) {
|
|
opus_multistream_decoder_destroy(mOpusDecoder);
|
|
mOpusDecoder = nullptr;
|
|
}
|
|
}
|
|
|
|
RefPtr<ShutdownPromise>
|
|
OpusDataDecoder::Shutdown()
|
|
{
|
|
RefPtr<OpusDataDecoder> self = this;
|
|
return InvokeAsync(mTaskQueue, __func__, [self]() {
|
|
return ShutdownPromise::CreateAndResolve(true, __func__);
|
|
});
|
|
}
|
|
|
|
void
|
|
OpusDataDecoder::AppendCodecDelay(MediaByteBuffer* config, uint64_t codecDelayUS)
|
|
{
|
|
uint8_t buffer[sizeof(uint64_t)];
|
|
BigEndian::writeUint64(buffer, codecDelayUS);
|
|
config->AppendElements(buffer, sizeof(uint64_t));
|
|
}
|
|
|
|
RefPtr<MediaDataDecoder::InitPromise>
|
|
OpusDataDecoder::Init()
|
|
{
|
|
size_t length = mInfo.mCodecSpecificConfig->Length();
|
|
uint8_t *p = mInfo.mCodecSpecificConfig->Elements();
|
|
if (length < sizeof(uint64_t)) {
|
|
OPUS_DEBUG("CodecSpecificConfig too short to read codecDelay!");
|
|
return InitPromise::CreateAndReject(NS_ERROR_DOM_MEDIA_FATAL_ERR, __func__);
|
|
}
|
|
int64_t codecDelay = BigEndian::readUint64(p);
|
|
length -= sizeof(uint64_t);
|
|
p += sizeof(uint64_t);
|
|
if (NS_FAILED(DecodeHeader(p, length))) {
|
|
OPUS_DEBUG("Error decoding header!");
|
|
return InitPromise::CreateAndReject(NS_ERROR_DOM_MEDIA_FATAL_ERR, __func__);
|
|
}
|
|
|
|
int r;
|
|
mOpusDecoder = opus_multistream_decoder_create(mOpusParser->mRate,
|
|
mOpusParser->mChannels,
|
|
mOpusParser->mStreams,
|
|
mOpusParser->mCoupledStreams,
|
|
mMappingTable,
|
|
&r);
|
|
mSkip = mOpusParser->mPreSkip;
|
|
mPaddingDiscarded = false;
|
|
|
|
if (codecDelay != FramesToUsecs(mOpusParser->mPreSkip,
|
|
mOpusParser->mRate).value()) {
|
|
NS_WARNING("Invalid Opus header: CodecDelay and pre-skip do not match!");
|
|
return InitPromise::CreateAndReject(NS_ERROR_DOM_MEDIA_FATAL_ERR, __func__);
|
|
}
|
|
|
|
if (mInfo.mRate != (uint32_t)mOpusParser->mRate) {
|
|
NS_WARNING("Invalid Opus header: container and codec rate do not match!");
|
|
}
|
|
if (mInfo.mChannels != (uint32_t)mOpusParser->mChannels) {
|
|
NS_WARNING("Invalid Opus header: container and codec channels do not match!");
|
|
}
|
|
|
|
return r == OPUS_OK ? InitPromise::CreateAndResolve(TrackInfo::kAudioTrack, __func__)
|
|
: InitPromise::CreateAndReject(NS_ERROR_DOM_MEDIA_FATAL_ERR, __func__);
|
|
}
|
|
|
|
nsresult
|
|
OpusDataDecoder::DecodeHeader(const unsigned char* aData, size_t aLength)
|
|
{
|
|
MOZ_ASSERT(!mOpusParser);
|
|
MOZ_ASSERT(!mOpusDecoder);
|
|
MOZ_ASSERT(!mDecodedHeader);
|
|
mDecodedHeader = true;
|
|
|
|
mOpusParser = new OpusParser;
|
|
if (!mOpusParser->DecodeHeader(const_cast<unsigned char*>(aData), aLength)) {
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
int channels = mOpusParser->mChannels;
|
|
|
|
AudioConfig::ChannelLayout layout(channels);
|
|
if (!layout.IsValid()) {
|
|
OPUS_DEBUG("Invalid channel mapping. Source is %d channels", channels);
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
AudioConfig::ChannelLayout vorbisLayout(
|
|
channels, VorbisDataDecoder::VorbisLayout(channels));
|
|
AudioConfig::ChannelLayout smpteLayout(channels);
|
|
static_assert(sizeof(mOpusParser->mMappingTable) / sizeof(mOpusParser->mMappingTable[0]) >= MAX_AUDIO_CHANNELS,
|
|
"Invalid size set");
|
|
uint8_t map[sizeof(mOpusParser->mMappingTable) / sizeof(mOpusParser->mMappingTable[0])];
|
|
if (vorbisLayout.MappingTable(smpteLayout, map)) {
|
|
for (int i = 0; i < channels; i++) {
|
|
mMappingTable[i] = mOpusParser->mMappingTable[map[i]];
|
|
}
|
|
} else {
|
|
// Should never get here as vorbis layout is always convertible to SMPTE
|
|
// default layout.
|
|
PodCopy(mMappingTable, mOpusParser->mMappingTable, MAX_AUDIO_CHANNELS);
|
|
}
|
|
|
|
return NS_OK;
|
|
}
|
|
|
|
RefPtr<MediaDataDecoder::DecodePromise>
|
|
OpusDataDecoder::Decode(MediaRawData* aSample)
|
|
{
|
|
return InvokeAsync<MediaRawData*>(mTaskQueue, this, __func__,
|
|
&OpusDataDecoder::ProcessDecode, aSample);
|
|
}
|
|
|
|
RefPtr<MediaDataDecoder::DecodePromise>
|
|
OpusDataDecoder::ProcessDecode(MediaRawData* aSample)
|
|
{
|
|
uint32_t channels = mOpusParser->mChannels;
|
|
|
|
if (mPaddingDiscarded) {
|
|
// Discard padding should be used only on the final packet, so
|
|
// decoding after a padding discard is invalid.
|
|
OPUS_DEBUG("Opus error, discard padding on interstitial packet");
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_FATAL_ERR,
|
|
RESULT_DETAIL("Discard padding on interstitial packet")),
|
|
__func__);
|
|
}
|
|
|
|
if (!mLastFrameTime ||
|
|
mLastFrameTime.ref() != aSample->mTime.ToMicroseconds()) {
|
|
// We are starting a new block.
|
|
mFrames = 0;
|
|
mLastFrameTime = Some(aSample->mTime.ToMicroseconds());
|
|
}
|
|
|
|
// Maximum value is 63*2880, so there's no chance of overflow.
|
|
int frames_number =
|
|
opus_packet_get_nb_frames(aSample->Data(), aSample->Size());
|
|
if (frames_number <= 0) {
|
|
OPUS_DEBUG("Invalid packet header: r=%d length=%" PRIuSIZE, frames_number,
|
|
aSample->Size());
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
|
|
RESULT_DETAIL("Invalid packet header: r=%d length=%u",
|
|
frames_number, uint32_t(aSample->Size()))),
|
|
__func__);
|
|
}
|
|
|
|
int samples = opus_packet_get_samples_per_frame(
|
|
aSample->Data(), opus_int32(mOpusParser->mRate));
|
|
|
|
// A valid Opus packet must be between 2.5 and 120 ms long (48kHz).
|
|
CheckedInt32 totalFrames =
|
|
CheckedInt32(frames_number) * CheckedInt32(samples);
|
|
if (!totalFrames.isValid()) {
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
|
|
RESULT_DETAIL("Frames count overflow")),
|
|
__func__);
|
|
}
|
|
|
|
int frames = totalFrames.value();
|
|
if (frames < 120 || frames > 5760) {
|
|
OPUS_DEBUG("Invalid packet frames: %d", frames);
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
|
|
RESULT_DETAIL("Invalid packet frames:%d", frames)),
|
|
__func__);
|
|
}
|
|
|
|
AlignedAudioBuffer buffer(frames * channels);
|
|
if (!buffer) {
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_OUT_OF_MEMORY, __func__), __func__);
|
|
}
|
|
|
|
// Decode to the appropriate sample type.
|
|
#ifdef MOZ_SAMPLE_TYPE_FLOAT32
|
|
int ret = opus_multistream_decode_float(mOpusDecoder,
|
|
aSample->Data(), aSample->Size(),
|
|
buffer.get(), frames, false);
|
|
#else
|
|
int ret = opus_multistream_decode(mOpusDecoder,
|
|
aSample->Data(), aSample->Size(),
|
|
buffer.get(), frames, false);
|
|
#endif
|
|
if (ret < 0) {
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
|
|
RESULT_DETAIL("Opus decoding error:%d", ret)),
|
|
__func__);
|
|
}
|
|
NS_ASSERTION(ret == frames, "Opus decoded too few audio samples");
|
|
auto startTime = aSample->mTime;
|
|
|
|
// Trim the initial frames while the decoder is settling.
|
|
if (mSkip > 0) {
|
|
int32_t skipFrames = std::min<int32_t>(mSkip, frames);
|
|
int32_t keepFrames = frames - skipFrames;
|
|
OPUS_DEBUG(
|
|
"Opus decoder skipping %d of %d frames", skipFrames, frames);
|
|
PodMove(buffer.get(),
|
|
buffer.get() + skipFrames * channels,
|
|
keepFrames * channels);
|
|
startTime = startTime + FramesToTimeUnit(skipFrames, mOpusParser->mRate);
|
|
frames = keepFrames;
|
|
mSkip -= skipFrames;
|
|
}
|
|
|
|
if (aSample->mDiscardPadding > 0) {
|
|
OPUS_DEBUG("Opus decoder discarding %u of %d frames",
|
|
aSample->mDiscardPadding, frames);
|
|
// Padding discard is only supposed to happen on the final packet.
|
|
// Record the discard so we can return an error if another packet is
|
|
// decoded.
|
|
if (aSample->mDiscardPadding > uint32_t(frames)) {
|
|
// Discarding more than the entire packet is invalid.
|
|
OPUS_DEBUG("Opus error, discard padding larger than packet");
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_FATAL_ERR,
|
|
RESULT_DETAIL("Discard padding larger than packet")),
|
|
__func__);
|
|
}
|
|
|
|
mPaddingDiscarded = true;
|
|
frames = frames - aSample->mDiscardPadding;
|
|
}
|
|
|
|
// Apply the header gain if one was specified.
|
|
#ifdef MOZ_SAMPLE_TYPE_FLOAT32
|
|
if (mOpusParser->mGain != 1.0f) {
|
|
float gain = mOpusParser->mGain;
|
|
uint32_t samples = frames * channels;
|
|
for (uint32_t i = 0; i < samples; i++) {
|
|
buffer[i] *= gain;
|
|
}
|
|
}
|
|
#else
|
|
if (mOpusParser->mGain_Q16 != 65536) {
|
|
int64_t gain_Q16 = mOpusParser->mGain_Q16;
|
|
uint32_t samples = frames * channels;
|
|
for (uint32_t i = 0; i < samples; i++) {
|
|
int32_t val = static_cast<int32_t>((gain_Q16*buffer[i] + 32768)>>16);
|
|
buffer[i] = static_cast<AudioDataValue>(MOZ_CLIP_TO_15(val));
|
|
}
|
|
}
|
|
#endif
|
|
|
|
auto duration = FramesToTimeUnit(frames, mOpusParser->mRate);
|
|
if (!duration.IsValid()) {
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_OVERFLOW_ERR,
|
|
RESULT_DETAIL("Overflow converting WebM audio duration")),
|
|
__func__);
|
|
}
|
|
auto time = startTime -
|
|
FramesToTimeUnit(mOpusParser->mPreSkip, mOpusParser->mRate) +
|
|
FramesToTimeUnit(mFrames, mOpusParser->mRate);
|
|
if (!time.IsValid()) {
|
|
return DecodePromise::CreateAndReject(
|
|
MediaResult(NS_ERROR_DOM_MEDIA_OVERFLOW_ERR,
|
|
RESULT_DETAIL("Overflow shifting tstamp by codec delay")),
|
|
__func__);
|
|
};
|
|
|
|
|
|
mFrames += frames;
|
|
|
|
return DecodePromise::CreateAndResolve(
|
|
DecodedData{ new AudioData(aSample->mOffset, time, duration,
|
|
frames, Move(buffer), mOpusParser->mChannels,
|
|
mOpusParser->mRate) },
|
|
__func__);
|
|
}
|
|
|
|
RefPtr<MediaDataDecoder::DecodePromise>
|
|
OpusDataDecoder::Drain()
|
|
{
|
|
RefPtr<OpusDataDecoder> self = this;
|
|
// InvokeAsync dispatches a task that will be run after any pending decode
|
|
// completes. As such, once the drain task run, there's nothing more to do.
|
|
return InvokeAsync(mTaskQueue, __func__, [] {
|
|
return DecodePromise::CreateAndResolve(DecodedData(), __func__);
|
|
});
|
|
}
|
|
|
|
RefPtr<MediaDataDecoder::FlushPromise>
|
|
OpusDataDecoder::Flush()
|
|
{
|
|
if (!mOpusDecoder) {
|
|
return FlushPromise::CreateAndResolve(true, __func__);
|
|
}
|
|
|
|
RefPtr<OpusDataDecoder> self = this;
|
|
return InvokeAsync(mTaskQueue, __func__, [self, this]() {
|
|
MOZ_ASSERT(mOpusDecoder);
|
|
// Reset the decoder.
|
|
opus_multistream_decoder_ctl(mOpusDecoder, OPUS_RESET_STATE);
|
|
mSkip = mOpusParser->mPreSkip;
|
|
mPaddingDiscarded = false;
|
|
mLastFrameTime.reset();
|
|
return FlushPromise::CreateAndResolve(true, __func__);
|
|
});
|
|
}
|
|
|
|
/* static */
|
|
bool
|
|
OpusDataDecoder::IsOpus(const nsACString& aMimeType)
|
|
{
|
|
return aMimeType.EqualsLiteral("audio/opus");
|
|
}
|
|
|
|
} // namespace mozilla
|
|
#undef OPUS_DEBUG
|