RefPtr<MediaDataDecoder::DecodePromise>
VorbisDataDecoder::ProcessDecode(MediaRawData* aSample)
{
MOZ_ASSERT(mTaskQueue->IsCurrentThreadIn());
const unsigned char* aData = aSample->Data();
size_t aLength = aSample->Size();
int64_t aOffset = aSample->mOffset;
MOZ_ASSERT(mPacketCount >= 3);
if (!mLastFrameTime ||
mLastFrameTime.ref() != aSample->mTime.ToMicroseconds()) {
// We are starting a new block.
mFrames = 0;
mLastFrameTime = Some(aSample->mTime.ToMicroseconds());
}
ogg_packet pkt = InitVorbisPacket(
aData, aLength, false, aSample->mEOS,
aSample->mTimecode.ToMicroseconds(), mPacketCount++);
int err = vorbis_synthesis(&mVorbisBlock, &pkt);
if (err) {
return DecodePromise::CreateAndReject(
MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
RESULT_DETAIL("vorbis_synthesis:%d", err)),
__func__);
}
err = vorbis_synthesis_blockin(&mVorbisDsp, &mVorbisBlock);
if (err) {
return DecodePromise::CreateAndReject(
MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
RESULT_DETAIL("vorbis_synthesis_blockin:%d", err)),
__func__);
}
VorbisPCMValue** pcm = 0;
int32_t frames = vorbis_synthesis_pcmout(&mVorbisDsp, &pcm);
if (frames == 0) {
return DecodePromise::CreateAndResolve(DecodedData(), __func__);
}
DecodedData results;
while (frames > 0) {
uint32_t channels = mVorbisDsp.vi->channels;
uint32_t rate = mVorbisDsp.vi->rate;
AlignedAudioBuffer buffer(frames*channels);
if (!buffer) {
return DecodePromise::CreateAndReject(
MediaResult(NS_ERROR_OUT_OF_MEMORY, __func__), __func__);
}
for (uint32_t j = 0; j < channels; ++j) {
VorbisPCMValue* channel = pcm[j];
for (uint32_t i = 0; i < uint32_t(frames); ++i) {
buffer[i*channels + j] = MOZ_CONVERT_VORBIS_SAMPLE(channel[i]);
}
}
auto duration = FramesToTimeUnit(frames, rate);
if (!duration.IsValid()) {
return DecodePromise::CreateAndReject(
MediaResult(NS_ERROR_DOM_MEDIA_OVERFLOW_ERR,
RESULT_DETAIL("Overflow converting audio duration")),
__func__);
}
auto total_duration = FramesToTimeUnit(mFrames, rate);
if (!total_duration.IsValid()) {
return DecodePromise::CreateAndReject(
MediaResult(NS_ERROR_DOM_MEDIA_OVERFLOW_ERR,
RESULT_DETAIL("Overflow converting audio total_duration")),
__func__);
}
auto time = total_duration + aSample->mTime;
if (!time.IsValid()) {
return DecodePromise::CreateAndReject(
MediaResult(
NS_ERROR_DOM_MEDIA_OVERFLOW_ERR,
RESULT_DETAIL("Overflow adding total_duration and aSample->mTime")),
__func__);
};
if (!mAudioConverter) {
AudioConfig in(
AudioConfig::ChannelLayout(channels, VorbisLayout(channels)), rate);
AudioConfig out(channels, rate);
if (!in.IsValid() || !out.IsValid()) {
return DecodePromise::CreateAndReject(
MediaResult(NS_ERROR_DOM_MEDIA_FATAL_ERR,
RESULT_DETAIL("Invalid channel layout:%u", channels)),
__func__);
}
mAudioConverter = MakeUnique<AudioConverter>(in, out);
}
MOZ_ASSERT(mAudioConverter->CanWorkInPlace());
AudioSampleBuffer data(Move(buffer));
data = mAudioConverter->Process(Move(data));
results.AppendElement(new AudioData(aOffset, time, duration,
frames, data.Forget(), channels, rate));
mFrames += frames;
err = vorbis_synthesis_read(&mVorbisDsp, frames);
if (err) {
return DecodePromise::CreateAndReject(
MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
RESULT_DETAIL("vorbis_synthesis_read:%d", err)),
__func__);
}
frames = vorbis_synthesis_pcmout(&mVorbisDsp, &pcm);
}
return DecodePromise::CreateAndResolve(Move(results), __func__);
}
MediaResult
FFmpegAudioDecoder<LIBAV_VER>::DoDecode(MediaRawData* aSample,
uint8_t* aData,
int aSize,
bool* aGotFrame,
DecodedData& aResults)
{
AVPacket packet;
mLib->av_init_packet(&packet);
packet.data = const_cast<uint8_t*>(aData);
packet.size = aSize;
if (aGotFrame) {
*aGotFrame = false;
}
if (!PrepareFrame()) {
return MediaResult(
NS_ERROR_OUT_OF_MEMORY,
RESULT_DETAIL("FFmpeg audio decoder failed to allocate frame"));
}
int64_t samplePosition = aSample->mOffset;
media::TimeUnit pts = aSample->mTime;
while (packet.size > 0) {
int decoded;
int bytesConsumed =
mLib->avcodec_decode_audio4(mCodecContext, mFrame, &decoded, &packet);
if (bytesConsumed < 0) {
NS_WARNING("FFmpeg audio decoder error.");
return MediaResult(NS_ERROR_DOM_MEDIA_DECODE_ERR,
RESULT_DETAIL("FFmpeg audio error:%d", bytesConsumed));
}
if (decoded) {
if (mFrame->format != AV_SAMPLE_FMT_FLT &&
mFrame->format != AV_SAMPLE_FMT_FLTP &&
mFrame->format != AV_SAMPLE_FMT_S16 &&
mFrame->format != AV_SAMPLE_FMT_S16P &&
mFrame->format != AV_SAMPLE_FMT_S32 &&
mFrame->format != AV_SAMPLE_FMT_S32P) {
return MediaResult(
NS_ERROR_DOM_MEDIA_DECODE_ERR,
RESULT_DETAIL(
"FFmpeg audio decoder outputs unsupported audio format"));
}
uint32_t numChannels = mCodecContext->channels;
uint32_t samplingRate = mCodecContext->sample_rate;
AlignedAudioBuffer audio =
CopyAndPackAudio(mFrame, numChannels, mFrame->nb_samples);
if (!audio) {
return MediaResult(NS_ERROR_OUT_OF_MEMORY, __func__);
}
media::TimeUnit duration =
FramesToTimeUnit(mFrame->nb_samples, samplingRate);
if (!duration.IsValid()) {
return MediaResult(NS_ERROR_DOM_MEDIA_OVERFLOW_ERR,
RESULT_DETAIL("Invalid sample duration"));
}
media::TimeUnit newpts = pts + duration;
if (!newpts.IsValid()) {
return MediaResult(
NS_ERROR_DOM_MEDIA_OVERFLOW_ERR,
RESULT_DETAIL("Invalid count of accumulated audio samples"));
}
aResults.AppendElement(new AudioData(samplePosition,
pts,
duration,
mFrame->nb_samples,
Move(audio),
numChannels,
samplingRate,
mCodecContext->channel_layout));
pts = newpts;
if (aGotFrame) {
*aGotFrame = true;
}
}
packet.data += bytesConsumed;
packet.size -= bytesConsumed;
samplePosition += bytesConsumed;
}
return NS_OK;
}
