nsresult
MediaDataBox::Write()
{
nsresult rv;
BoxSizeChecker checker(mControl, size);
Box::Write();
nsTArray<uint32_t> types;
types.AppendElement(Audio_Track);
types.AppendElement(Video_Track);
for (uint32_t l = 0; l < types.Length(); l++) {
if (mTrackType & types[l]) {
FragmentBuffer* frag = mControl->GetFragment(types[l]);
nsTArray<nsRefPtr<EncodedFrame>> frames;
// Here is the last time we get fragment frames, flush it!
rv = frag->GetFirstFragment(frames, true);
NS_ENSURE_SUCCESS(rv, rv);
uint32_t len = frames.Length();
for (uint32_t i = 0; i < len; i++) {
mControl->Write((uint8_t*)frames.ElementAt(i)->GetFrameData().Elements(),
frames.ElementAt(i)->GetFrameData().Length());
}
}
}
return NS_OK;
}
uint32_t
TrackRunBox::fillSampleTable()
{
uint32_t table_size = 0;
nsresult rv;
nsTArray<nsRefPtr<EncodedFrame>> frames;
FragmentBuffer* frag = mControl->GetFragment(mTrackType);
rv = frag->GetFirstFragment(frames);
if (NS_FAILED(rv)) {
return 0;
}
uint32_t len = frames.Length();
sample_info_table = new tbl[len];
for (uint32_t i = 0; i < len; i++) {
sample_info_table[i].sample_duration = 0;
sample_info_table[i].sample_size = frames.ElementAt(i)->GetFrameData().Length();
mAllSampleSize += sample_info_table[i].sample_size;
table_size += sizeof(uint32_t);
if (flags.to_ulong() & flags_sample_flags_present) {
sample_info_table[i].sample_flags =
set_sample_flags(
(frames.ElementAt(i)->GetFrameType() == EncodedFrame::I_FRAME));
table_size += sizeof(uint32_t);
} else {
sample_info_table[i].sample_flags = 0;
}
sample_info_table[i].sample_composition_time_offset = 0;
}
return table_size;
}
nsresult
AVCConfigurationBox::Generate(uint32_t* aBoxSize)
{
nsresult rv;
FragmentBuffer* frag = mControl->GetFragment(Video_Track);
rv = frag->GetCSD(avcConfig);
NS_ENSURE_SUCCESS(rv, rv);
size += avcConfig.Length();
*aBoxSize = size;
return NS_OK;
}
nsresult
EVRCSpecificBox::Generate(uint32_t* aBoxSize)
{
nsresult rv;
FragmentBuffer* frag = mControl->GetFragment(Audio_Track);
rv = frag->GetCSD(evrcDecSpecInfo);
NS_ENSURE_SUCCESS(rv, rv);
size += evrcDecSpecInfo.Length();
*aBoxSize = size;
return NS_OK;
}
nsresult
ES_Descriptor::Generate(uint32_t* aBoxSize)
{
nsresult rv;
// 14496-1 '8.3.4 DecoderConfigDescriptor'
// 14496-1 '10.2.3 SL Packet Header Configuration'
Box::MetaHelper meta;
meta.Init(mControl);
FragmentBuffer* frag = mControl->GetFragment(Audio_Track);
rv = frag->GetCSD(DecodeSpecificInfo);
NS_ENSURE_SUCCESS(rv, rv);
length = sizeof(ES_ID) + 1;
length += DecodeSpecificInfo.Length();
*aBoxSize = sizeof(tag) + sizeof(length) + length;
return NS_OK;
}
nsresult
MediaDataBox::Generate(uint32_t* aBoxSize)
{
mFirstSampleOffset = size;
mAllSampleSize = 0;
if (mTrackType & Audio_Track) {
FragmentBuffer* frag = mControl->GetFragment(Audio_Track);
mAllSampleSize += frag->GetFirstFragmentSampleSize();
}
if (mTrackType & Video_Track) {
FragmentBuffer* frag = mControl->GetFragment(Video_Track);
mAllSampleSize += frag->GetFirstFragmentSampleSize();
}
size += mAllSampleSize;
*aBoxSize = size;
return NS_OK;
}
nsresult
TrackRunBox::Generate(uint32_t* aBoxSize)
{
FragmentBuffer* frag = mControl->GetFragment(mTrackType);
sample_count = frag->GetFirstFragmentSampleNumber();
size += sizeof(sample_count);
// data_offset needs to be updated if there is other
// TrackRunBox before this one.
if (flags.to_ulong() & flags_data_offset_present) {
data_offset = 0;
size += sizeof(data_offset);
}
size += fillSampleTable();
*aBoxSize = size;
return NS_OK;
}
void test_smoke()
{
FragmentBuffer fb;
NUT_TA(fb.readable_size() == 0);
const void* bufs[2];
size_t lens[2];
NUT_TA(fb.readable_pointers(bufs, lens, 2) == 0);
FragmentBuffer::Fragment *f = FragmentBuffer::new_fragment(10);
f->size = 2;
f->buffer[1] = 0x12;
f = fb.write_fragment(f);
NUT_TA(nullptr == f);
NUT_TA(fb.readable_size() == 2);
NUT_TA(fb.readable_pointers(bufs, lens, 2) == 1);
NUT_TA(lens[0] == 2);
f = FragmentBuffer::new_fragment(10);
f->size = 2;
f->buffer[0] = 0x34;
f->buffer[1] = 0x56;
f = fb.write_fragment(f);
NUT_TA(nullptr != f);
NUT_TA(fb.readable_size() == 4);
NUT_TA(fb.readable_pointers(bufs, lens, 2) == 1);
NUT_TA(lens[0] == 4);
f->size = 7;
f->buffer[0] = 0x78;
f = fb.write_fragment(f);
NUT_TA(nullptr == f);
NUT_TA(fb.readable_size() == 11);
NUT_TA(fb.readable_pointers(bufs, lens, 2) == 2);
NUT_TA(lens[0] == 4 && lens[1] == 7);
fb.skip_read(1);
NUT_TA(fb.readable_pointers(bufs, lens, 2) == 2);
NUT_TA(lens[0] == 3 && lens[1] == 7);
uint16_t v = 0;
fb.read(&v, 2);
NUT_TA(v == 0x3412);
fb.read(&v, 2);
NUT_TA(v == 0x7856);
NUT_TA(fb.readable_size() == 6);
}
nsresult
ISOMediaWriter::WriteEncodedTrack(const EncodedFrameContainer& aData,
uint32_t aFlags)
{
// Muxing complete, it doesn't allowed to reentry again.
if (mState == MUXING_DONE) {
MOZ_ASSERT(false);
return NS_ERROR_FAILURE;
}
FragmentBuffer* frag = nullptr;
uint32_t len = aData.GetEncodedFrames().Length();
if (!len) {
// no frame? why bother to WriteEncodedTrack
return NS_OK;
}
for (uint32_t i = 0; i < len; i++) {
nsRefPtr<EncodedFrame> frame(aData.GetEncodedFrames()[i]);
EncodedFrame::FrameType type = frame->GetFrameType();
if (type == EncodedFrame::AAC_AUDIO_FRAME ||
type == EncodedFrame::AAC_CSD ||
type == EncodedFrame::AMR_AUDIO_FRAME ||
type == EncodedFrame::AMR_AUDIO_CSD) {
frag = mAudioFragmentBuffer;
} else if (type == EncodedFrame::AVC_I_FRAME ||
type == EncodedFrame::AVC_P_FRAME ||
type == EncodedFrame::AVC_B_FRAME ||
type == EncodedFrame::AVC_CSD) {
frag = mVideoFragmentBuffer;
} else {
MOZ_ASSERT(0);
return NS_ERROR_FAILURE;
}
frag->AddFrame(frame);
}
// Encoder should send CSD (codec specific data) frame before sending the
// audio/video frames. When CSD data is ready, it is sufficient to generate a
// moov data. If encoder doesn't send CSD yet, muxer needs to wait before
// generating anything.
if (mType & Audio_Track && (!mAudioFragmentBuffer ||
!mAudioFragmentBuffer->HasCSD())) {
return NS_OK;
}
if (mType & Video_Track && (!mVideoFragmentBuffer ||
!mVideoFragmentBuffer->HasCSD())) {
return NS_OK;
}
// Only one FrameType in EncodedFrameContainer so it doesn't need to be
// inside the for-loop.
if (frag && (aFlags & END_OF_STREAM)) {
frag->SetEndOfStream();
}
nsresult rv;
bool EOS;
if (ReadyToRunState(EOS)) {
// TODO:
// The MediaEncoder doesn't use nsRunnable, so thread will be
// stocked on that part and the new added nsRunnable won't get to run
// before MediaEncoder completing. Before MediaEncoder change, it needs
// to call RunState directly.
// https://bugzilla.mozilla.org/show_bug.cgi?id=950429
rv = RunState();
NS_ENSURE_SUCCESS(rv, rv);
}
return NS_OK;
}
nsresult
ISOMediaWriter::WriteEncodedTrack(const EncodedFrameContainer& aData,
uint32_t aFlags)
{
PROFILER_LABEL("ISOMediaWriter", "WriteEncodedTrack",
js::ProfileEntry::Category::OTHER);
// Muxing complete, it doesn't allowed to reentry again.
if (mState == MUXING_DONE) {
MOZ_ASSERT(false);
return NS_ERROR_FAILURE;
}
FragmentBuffer* frag = nullptr;
uint32_t len = aData.GetEncodedFrames().Length();
if (!len) {
// no frame? why bother to WriteEncodedTrack
return NS_OK;
}
for (uint32_t i = 0; i < len; i++) {
nsRefPtr<EncodedFrame> frame(aData.GetEncodedFrames()[i]);
EncodedFrame::FrameType type = frame->GetFrameType();
if (type == EncodedFrame::AAC_AUDIO_FRAME ||
type == EncodedFrame::AAC_CSD ||
type == EncodedFrame::AMR_AUDIO_FRAME ||
type == EncodedFrame::AMR_AUDIO_CSD) {
frag = mAudioFragmentBuffer;
} else if (type == EncodedFrame::AVC_I_FRAME ||
type == EncodedFrame::AVC_P_FRAME ||
type == EncodedFrame::AVC_B_FRAME ||
type == EncodedFrame::AVC_CSD) {
frag = mVideoFragmentBuffer;
} else {
MOZ_ASSERT(0);
return NS_ERROR_FAILURE;
}
frag->AddFrame(frame);
}
// Encoder should send CSD (codec specific data) frame before sending the
// audio/video frames. When CSD data is ready, it is sufficient to generate a
// moov data. If encoder doesn't send CSD yet, muxer needs to wait before
// generating anything.
if (mType & Audio_Track && (!mAudioFragmentBuffer ||
!mAudioFragmentBuffer->HasCSD())) {
return NS_OK;
}
if (mType & Video_Track && (!mVideoFragmentBuffer ||
!mVideoFragmentBuffer->HasCSD())) {
return NS_OK;
}
// Only one FrameType in EncodedFrameContainer so it doesn't need to be
// inside the for-loop.
if (frag && (aFlags & END_OF_STREAM)) {
frag->SetEndOfStream();
}
nsresult rv;
bool EOS;
if (ReadyToRunState(EOS)) {
// Because track encoder won't generate new data after EOS, it needs to make
// sure the state reaches MUXING_DONE when EOS is signaled.
do {
rv = RunState();
} while (EOS && mState != MUXING_DONE);
NS_ENSURE_SUCCESS(rv, rv);
}
return NS_OK;
}
uint32_t
TrackRunBox::fillSampleTable()
{
uint32_t table_size = 0;
nsresult rv;
nsTArray<RefPtr<EncodedFrame>> frames;
FragmentBuffer* frag = mControl->GetFragment(mTrackType);
rv = frag->GetFirstFragment(frames);
if (NS_FAILED(rv)) {
return 0;
}
uint32_t len = frames.Length();
sample_info_table = MakeUnique<tbl[]>(len);
// Create sample table according to 14496-12 8.8.8.2.
for (uint32_t i = 0; i < len; i++) {
// Sample size.
sample_info_table[i].sample_size = 0;
if (flags.to_ulong() & flags_sample_size_present) {
sample_info_table[i].sample_size = frames.ElementAt(i)->GetFrameData().Length();
mAllSampleSize += sample_info_table[i].sample_size;
table_size += sizeof(uint32_t);
}
// Sample flags.
sample_info_table[i].sample_flags = 0;
if (flags.to_ulong() & flags_sample_flags_present) {
sample_info_table[i].sample_flags =
set_sample_flags(
(frames.ElementAt(i)->GetFrameType() == EncodedFrame::AVC_I_FRAME));
table_size += sizeof(uint32_t);
}
// Sample duration.
sample_info_table[i].sample_duration = 0;
if (flags.to_ulong() & flags_sample_duration_present) {
// Calculate each frame's duration, it is decided by "current frame
// timestamp - last frame timestamp".
uint64_t frame_time = 0;
if (i == 0) {
frame_time = frames.ElementAt(i)->GetTimeStamp() -
frag->GetLastFragmentLastFrameTime();
} else {
frame_time = frames.ElementAt(i)->GetTimeStamp() -
frames.ElementAt(i - 1)->GetTimeStamp();
// Keep the last frame time of current fagment, it will be used to calculate
// the first frame duration of next fragment.
if ((len - 1) == i) {
frag->SetLastFragmentLastFrameTime(frames.ElementAt(i)->GetTimeStamp());
}
}
// In TrackRunBox, there should be exactly one type, either audio or video.
MOZ_ASSERT((mTrackType & Video_Track) ^ (mTrackType & Audio_Track));
sample_info_table[i].sample_duration = (mTrackType & Video_Track ?
frame_time * mVideoMeta->GetVideoClockRate() / USECS_PER_S :
frame_time * mAudioMeta->GetAudioSampleRate() / USECS_PER_S);
table_size += sizeof(uint32_t);
}
sample_info_table[i].sample_composition_time_offset = 0;
}
return table_size;
}
