void
MediaAudioTest::testCopy() {
const int32_t numSamples = 155; // I choose an odd number because HV will align up to 32
const int32_t sampleRate = 22050;
const int32_t channels = 15; // choose a large # of channels to make sure we expand the Frame
const AudioChannel::Layout layout = AudioChannel::CH_LAYOUT_UNKNOWN;
const AudioFormat::Type format = AudioFormat::SAMPLE_FMT_DBLP;
int32_t bufSize = AudioFormat::getBufferSizeNeeded(numSamples, channels,
format);
// test that there is rounding up
int32_t minSize = AudioFormat::getBytesPerSample(format) * numSamples
* channels;
TS_ASSERT_LESS_THAN(minSize, bufSize);
RefPointer<Buffer> src = Buffer::make(0, bufSize);
double* srcData = (double*) src->getBytes(0, bufSize);
// now, let's go nuts!
for (size_t i = 0; i < bufSize / sizeof(double); i++) {
srcData[i] = i;
}
RefPointer<MediaAudio> audio;
audio = MediaAudio::make(src.value(), numSamples, sampleRate, channels,
layout, format);
TS_ASSERT(audio);
TS_ASSERT_EQUALS(channels, audio->getNumDataPlanes());
bool tests[] =
{ true, false };
for (size_t i = 0; i < sizeof(tests) / sizeof(*tests); i++) {
// now let's make a copy
RefPointer<MediaAudio> copy = MediaAudio::make(audio.value(), tests[i]);
TS_ASSERT_EQUALS(copy->getMaxNumSamples(), audio->getMaxNumSamples());
TS_ASSERT_EQUALS(copy->getNumSamples(), audio->getNumSamples());
TS_ASSERT_EQUALS(copy->getChannels(), audio->getChannels());
TS_ASSERT_EQUALS(copy->getNumDataPlanes(), audio->getNumDataPlanes());
TS_ASSERT_EQUALS(copy->getChannelLayout(), audio->getChannelLayout());
TS_ASSERT_EQUALS(copy->getSampleRate(), audio->getSampleRate());
TS_ASSERT_EQUALS(copy->getFormat(), audio->getFormat());
for (int32_t j = 0; j < audio->getNumDataPlanes(); j++) {
RefPointer<Buffer> srcBuf = audio->getData(j);
RefPointer<Buffer> dstBuf = copy->getData(j);
int32_t planeSize = srcBuf->getBufferSize();
TS_ASSERT_EQUALS(planeSize, dstBuf->getBufferSize());
uint8_t* srcBytes = (uint8_t*) srcBuf->getBytes(0, planeSize);
uint8_t* dstBytes = (uint8_t*) dstBuf->getBytes(0, planeSize);
if (tests[i]) {
for (int32_t k = 0; k < planeSize; k++) {
// should be byte-by-byte the same
TS_ASSERT_EQUALS(srcBytes[k], dstBytes[k]);
}
} else {
TS_ASSERT_EQUALS(srcBytes, dstBytes);
}
}
}
}
void
MediaAudioTest::testCreation() {
const int32_t numSamples = 1024;
const int32_t sampleRate = 22050;
const int32_t channels = 8;
const AudioChannel::Layout layout = AudioChannel::CH_LAYOUT_7POINT1;
const AudioFormat::Type format = AudioFormat::SAMPLE_FMT_S16P;
// now let's test invalid methods.
RefPointer<MediaAudio> audio;
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(-1, sampleRate, channels, layout, format),
HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, -1, channels, layout, format),
HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, sampleRate, -1, layout, format),
HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, sampleRate, channels + 1, layout,
format), HumbleInvalidArgument);
}
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(MediaAudio::make(numSamples, sampleRate, channels, layout,
AudioFormat::SAMPLE_FMT_NONE), HumbleInvalidArgument);
}
// And this should be valid
audio = MediaAudio::make(numSamples, sampleRate, channels + 1,
AudioChannel::CH_LAYOUT_UNKNOWN, format);
TS_ASSERT(audio);
audio = MediaAudio::make(numSamples, sampleRate, channels, layout, format);
TS_ASSERT(audio);
// now let's try getting the data
RefPointer<Buffer> buf;
TS_ASSERT_EQUALS(numSamples, audio->getMaxNumSamples());
TS_ASSERT_EQUALS(numSamples, audio->getNumSamples());
TS_ASSERT(!audio->isComplete());
TS_ASSERT_EQUALS(channels, audio->getChannels());
TS_ASSERT_EQUALS(channels, audio->getNumDataPlanes());
TS_ASSERT_EQUALS(layout, audio->getChannelLayout());
TS_ASSERT_EQUALS(sampleRate, audio->getSampleRate());
TS_ASSERT_EQUALS(format, audio->getFormat());
{
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(
audio->setNumSamples(audio->getMaxNumSamples()+1),
HumbleInvalidArgument);
TS_ASSERT_THROWS(audio->setNumSamples(-1), HumbleInvalidArgument);
TS_ASSERT_THROWS(audio->setNumSamples(0), HumbleInvalidArgument);
}
audio->setNumSamples(1);
TS_ASSERT_EQUALS(1, audio->getNumSamples());
audio->setComplete(true);
TS_ASSERT(audio->isComplete());
for (int i = 0; i < channels; i++) {
buf = audio->getData(i);
TS_ASSERT(buf);
TS_ASSERT_EQUALS(audio->getDataPlaneSize(i),
audio->getNumSamples()*AudioFormat::getBytesPerSample(audio->getFormat())*(audio->isPlanar()?1:audio->getChannels()));
}
// now let's try packed audio
audio = MediaAudio::make(numSamples, sampleRate, channels, layout,
AudioFormat::getPackedSampleFormat(format));
TS_ASSERT(audio);
TS_ASSERT_EQUALS(AudioFormat::getPackedSampleFormat(format),
audio->getFormat());
TS_ASSERT_EQUALS(channels, audio->getChannels());
TS_ASSERT_EQUALS(1, audio->getNumDataPlanes());
buf = audio->getData(0);
TS_ASSERT(buf);
for (int i = 1; i < channels; i++) {
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
TS_ASSERT_THROWS(audio->getData(i), HumbleInvalidArgument);
}
}
MediaPictureImpl*
MediaPictureImpl::make(Buffer* buffer, int32_t width, int32_t height,
PixelFormat::Type format) {
if (width <= 0) {
VS_THROW(HumbleInvalidArgument("width must be > 0"));
}
if (height <= 0) {
VS_THROW(HumbleInvalidArgument("height must be > 0"));
}
if (format == PixelFormat::PIX_FMT_NONE) {
VS_THROW(HumbleInvalidArgument("pixel format must be specifie"));
}
if (!buffer) {
VS_THROW(HumbleInvalidArgument("must pass non null buffer"));
}
// let's figure out how big of a buffer we need
int32_t bufSize = PixelFormat::getBufferSizeNeeded(width, height, format);
if (bufSize < buffer->getBufferSize()) {
VS_THROW(
HumbleInvalidArgument(
"passed in buffer too small to fit requested image parameters"));
}
RefPointer<MediaPictureImpl> retval = make();
AVFrame* frame = retval->mFrame;
frame->width = width;
frame->height = height;
frame->format = format;
// buffer is large enough; let's fill the data pointers
uint8_t* data = (uint8_t*) buffer->getBytes(0, bufSize);
int32_t imgSize = av_image_fill_arrays(frame->data, frame->linesize, data,
(enum AVPixelFormat) frame->format, frame->width, frame->height, 1);
if (imgSize != bufSize) {
VS_ASSERT(imgSize == bufSize, "these should always be equal");
VS_THROW(HumbleRuntimeError("could not fill image with data"));
}
// now, set up the reference buffers
frame->extended_data = frame->data;
for (int32_t i = 0; i < AV_NUM_DATA_POINTERS; i++) {
if (frame->data[i])
frame->buf[i] = AVBufferSupport::wrapBuffer(buffer, frame->data[i], frame->linesize[0]*frame->height+16);
}
// now fill in the AVBufferRefs where we pass of to FFmpeg care
// of our buffer. Be kind FFmpeg. Be kind.
RefPointer<PixelFormatDescriptor> desc = PixelFormat::getDescriptor((PixelFormat::Type)frame->format);
if (!desc) {
VS_THROW(HumbleRuntimeError("could not get format descriptor"));
}
if (desc->getFlag(PixelFormatDescriptor::PIX_FMT_FLAG_PAL) ||
desc->getFlag(PixelFormatDescriptor::PIX_FMT_FLAG_PSEUDOPAL)) {
av_buffer_unref(&frame->buf[1]);
frame->buf[1] = AVBufferSupport::wrapBuffer(Buffer::make(retval.value(), 1024));
if (!frame->buf[1]) {
VS_THROW(HumbleRuntimeError("memory failure"));
}
frame->data[1] = frame->buf[1]->data;
}
int32_t n = retval->getNumDataPlanes();
(void) n;
VS_LOG_TRACE("Created MediaPicture: %d x %d (%d). [%d, %d, %d, %d]",
retval->getWidth(),
retval->getHeight(),
retval->getFormat(),
n < 1 ? 0 : retval->getDataPlaneSize(0),
n < 2 ? 0 : retval->getDataPlaneSize(1),
n < 3 ? 0 : retval->getDataPlaneSize(2),
n < 4 ? 0 : retval->getDataPlaneSize(3)
);
// and we're done.
return retval.get();
}
void
MediaPictureResamplerTest::testRescale() {
TestData::Fixture* fixture=mFixtures.getFixture("testfile_h264_mp4a_tmcd.mov");
TS_ASSERT(fixture);
char filepath[2048];
mFixtures.fillPath(fixture, filepath, sizeof(filepath));
RefPointer<Demuxer> source = Demuxer::make();
source->open(filepath, 0, false, true, 0, 0);
int32_t numStreams = source->getNumStreams();
TS_ASSERT_EQUALS(fixture->num_streams, numStreams);
int32_t streamToDecode = 0;
RefPointer<DemuxerStream> stream = source->getStream(streamToDecode);
TS_ASSERT(stream);
RefPointer<Decoder> decoder = stream->getDecoder();
TS_ASSERT(decoder);
RefPointer<Codec> codec = decoder->getCodec();
TS_ASSERT(codec);
TS_ASSERT_EQUALS(Codec::CODEC_ID_H264, codec->getID());
decoder->open(0, 0);
TS_ASSERT_EQUALS(PixelFormat::PIX_FMT_YUV420P, decoder->getPixelFormat());
// now, let's start a decoding loop.
RefPointer<MediaPacket> packet = MediaPacket::make();
RefPointer<MediaPicture> picture = MediaPicture::make(
decoder->getWidth(),
decoder->getHeight(),
decoder->getPixelFormat());
int32_t rescaleW = decoder->getWidth()/4;
int32_t rescaleH = (int32_t)(decoder->getHeight()*5.22);
PixelFormat::Type rescaleFmt = PixelFormat::PIX_FMT_RGBA;
RefPointer<MediaPicture> rescaled = MediaPicture::make(
rescaleW, rescaleH, rescaleFmt);
RefPointer<MediaPictureResampler> resampler =
MediaPictureResampler::make(rescaled->getWidth(),
rescaled->getHeight(),
rescaled->getFormat(),
picture->getWidth(),
picture->getHeight(),
picture->getFormat(),
0);
resampler->open();
int32_t frameNo = 0;
while(source->read(packet.value()) >= 0) {
// got a packet; now we try to decode it.
if (packet->getStreamIndex() == streamToDecode &&
packet->isComplete()) {
int32_t bytesRead = 0;
int32_t byteOffset=0;
do {
bytesRead = decoder->decodeVideo(picture.value(), packet.value(), byteOffset);
if (picture->isComplete()) {
writePicture("MediaPictureResamplerTest_testRescaleVideo",
&frameNo, picture.value(), resampler.value(), rescaled.value());
}
byteOffset += bytesRead;
} while(byteOffset < packet->getSize());
// now, handle the case where bytesRead is 0; we need to flush any
// cached packets
do {
decoder->decodeVideo(picture.value(), 0, 0);
if (picture->isComplete()) {
writePicture("MediaPictureResamplerTest_testRescaleVideo", &frameNo, picture.value(),
resampler.value(), rescaled.value());
}
} while (picture->isComplete());
}
if ((int32_t)(frameNo/30) > 10)
// 20 pictures should be enough to see if it's working.
break;
}
source->close();
}
void
EncoderTest::testRegression36Internal (const Codec::ID codecId,
const int32_t numSamples,
const int32_t sampleRate,
const int32_t channels,
const AudioChannel::Layout channelLayout,
const AudioFormat::Type audioFormat,
const int64_t bitRate,
const char* testOutputName)
{
VS_LOG_DEBUG("Output filename: %s", testOutputName);
RefPointer<Codec> codec = Codec::findEncodingCodec (codecId);
RefPointer<Encoder> encoder = Encoder::make (codec.value ());
RefPointer<FilterGraph> graph = FilterGraph::make ();
RefPointer<MediaAudio> audio = MediaAudio::make (numSamples, sampleRate,
channels, channelLayout,
audioFormat);
// set the encoder properties we need
encoder->setSampleRate (audio->getSampleRate ());
encoder->setSampleFormat (audio->getFormat ());
encoder->setChannelLayout (audio->getChannelLayout ());
encoder->setChannels (audio->getChannels ());
encoder->setProperty ("b", (int64_t) (bitRate));
RefPointer<Rational> tb = Rational::make (1, sampleRate);
encoder->setTimeBase (tb.value ());
// create an output muxer
RefPointer<Muxer> muxer = Muxer::make (testOutputName, 0, 0);
RefPointer<MuxerFormat> format = muxer->getFormat ();
if (format->getFlag (MuxerFormat::GLOBAL_HEADER))
encoder->setFlag (Encoder::FLAG_GLOBAL_HEADER, true);
// open the encoder
encoder->open (0, 0);
RefPointer<FilterAudioSink> fsink = graph->addAudioSink (
"out", audio->getSampleRate (), audio->getChannelLayout (),
audio->getFormat ());
// Generate a 220 Hz sine wave with a 880 Hz beep each second, for 10 seconds.
graph->open ("sine=frequency=220:beep_factor=4:duration=11[out]");
fsink->setFrameSize (numSamples);
// add a stream for the encoded packets
{
RefPointer<MuxerStream> stream = muxer->addNewStream (encoder.value ());
} // and open the muxer
muxer->open (0, 0);
// now we're (in theory) ready to start writing data.
int32_t numCompletePackets = 0;
RefPointer<MediaPacket> packet;
// Get one audio packet that is larger than the frame-size.
fsink->getAudio (audio.value ());
TS_ASSERT(audio->isComplete ());
TS_ASSERT_EQUALS(audio->getNumSamples (), sampleRate);
audio->setTimeStamp (0);
// let's encode
packet = MediaPacket::make ();
encoder->encodeAudio (packet.value (), audio.value ());
if (packet->isComplete ())
{
muxer->write (packet.value (), false);
}
// now flush the encoder
do
{
packet = MediaPacket::make ();
encoder->encodeAudio (packet.value (), 0);
if (packet->isComplete ())
{
muxer->write (packet.value (), false);
++numCompletePackets;
}
}
while (packet->isComplete ());
muxer->close ();
if (!(codec->getCapabilities() & Codec::CAP_VARIABLE_FRAME_SIZE)) {
const int32_t numExpectedPackets = audio->getNumSamples()
/ encoder->getFrameSize();
VS_LOG_DEBUG("%ld vs %ld; framesize: %ld", numCompletePackets, numExpectedPackets, encoder->getFrameSize());
TS_ASSERT(numCompletePackets > 10);
}
}
void
EncoderTest::testEncodeVideo() {
Logger::setGlobalIsLogging(Logger::LEVEL_TRACE, false);
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_INFO, false);
const bool isMemCheck = getenv("VS_TEST_MEMCHECK") ? true : false;
const int32_t maxPics = isMemCheck ? 10 : 500;
int32_t width=176;
int32_t height=144;
RefPointer<Codec> codec = Codec::findEncodingCodec(Codec::CODEC_ID_H264);
RefPointer<Encoder> encoder = Encoder::make(codec.value());
RefPointer<MediaPicture> picture = MediaPicture::make(width*2,height*2,
PixelFormat::PIX_FMT_YUV420P);
// set the encoder properties we need
encoder->setWidth(picture->getWidth());
encoder->setHeight(picture->getHeight());
encoder->setPixelFormat(picture->getFormat());
encoder->setProperty("b", (int64_t)400000); // bitrate
encoder->setProperty("g", (int64_t) 10); // gop
encoder->setProperty("bf", (int64_t)1); // max b frames
RefPointer<Rational> tb = Rational::make(1,25);
encoder->setTimeBase(tb.value());
// mandlebrot, that is then negated, horizontally flipped, and edge detected, before
// final outputting to a new picture with each version in one of 4 quadrants.
char graphCommand[1024];
snprintf(graphCommand,sizeof(graphCommand),"mandelbrot=s=%dx%d[mb];"
"[mb]split=4[0][1][2][3];"
"[0]pad=iw*2:ih*2[a];"
"[1]negate[b];"
"[2]hflip[c];"
"[3]edgedetect[d];"
"[a][b]overlay=w[x];"
"[x][c]overlay=0:h[y];"
"[y][d]overlay=w:h[out]", width, height);
RefPointer<FilterGraph> graph = FilterGraph::make();
RefPointer<FilterPictureSink> fsink = graph->addPictureSink("out", picture->getFormat());
graph->open(graphCommand);
// let's set a frame time base of 1/30
RefPointer<Rational> pictureTb = Rational::make(1,30);
// create an output muxer
RefPointer<Muxer> muxer = Muxer::make("EncoderTest_encodeVideo.mov", 0, 0);
RefPointer<MuxerFormat> format = muxer->getFormat();
// if the container will require a global header, then the encoder needs to set this.
if (format->getFlag(MuxerFormat::GLOBAL_HEADER))
encoder->setFlag(Encoder::FLAG_GLOBAL_HEADER, true);
// open the encoder
encoder->open(0, 0);
// add a stream for the encoded packets
{
RefPointer<MuxerStream> stream = muxer->addNewStream(encoder.value());
}
// and open the muxer
muxer->open(0, 0);
// now we're (in theory) ready to start writing data.
int32_t numPics = 0;
RefPointer<MediaPacket> packet;
while(fsink->getPicture(picture.value()) >= 0 && numPics < maxPics) {
picture->setTimeBase(pictureTb.value());
picture->setTimeStamp(numPics);
// let's encode
packet = MediaPacket::make();
encoder->encodeVideo(packet.value(), picture.value());
if (packet->isComplete()) {
muxer->write(packet.value(), false);
}
++numPics;
}
// now flush the encoder
do {
packet = MediaPacket::make();
encoder->encodeVideo(packet.value(), 0);
if (packet->isComplete()) {
muxer->write(packet.value(), false);
}
} while (packet->isComplete());
muxer->close();
}
/**
* This test will read ironman (FLV, H263 video and mp3 audio) and transcode
* to MP4 (H264 Video and aac audio).
*/
void
EncoderTest::testTranscode()
{
// enable trace logging
Logger::setGlobalIsLogging(Logger::LEVEL_TRACE, false);
const bool isMemCheck = getenv("VS_TEST_MEMCHECK") ? true : false;
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_INFO, false);
TestData::Fixture* fixture;
fixture=mFixtures.getFixture("testfile.flv");
// fixture=mFixtures.getFixture("testfile_h264_mp4a_tmcd.mov");
// fixture=mFixtures.getFixture("bigbuckbunny_h264_aac_5.1.mp4");
TS_ASSERT(fixture);
char filepath[2048];
mFixtures.fillPath(fixture, filepath, sizeof(filepath));
RefPointer<Demuxer> source = Demuxer::make();
source->open(filepath, 0, false, true, 0, 0);
int32_t numStreams = source->getNumStreams();
TS_ASSERT_EQUALS(fixture->num_streams, numStreams);
// Let's create a helper object to help us with decoding
typedef struct {
MediaDescriptor::Type type;
RefPointer<DemuxerStream> stream;
RefPointer<Decoder> decoder;
RefPointer<MediaSampled> media;
} DemuxerStreamHelper;
// I know there are only 2 in the input file.
DemuxerStreamHelper inputHelpers[10];
for(int32_t i = 0; i < numStreams; i++) {
DemuxerStreamHelper* input = &inputHelpers[i];
input->stream = source->getStream(i);
input->decoder = input->stream->getDecoder();
if (!input->decoder)
// skip
break;
input->decoder->open(0, 0);
input->type = input->decoder->getCodecType();
if (input->type == MediaDescriptor::MEDIA_AUDIO)
input->media = MediaAudio::make(
input->decoder->getFrameSize(),
input->decoder->getSampleRate(),
input->decoder->getChannels(),
input->decoder->getChannelLayout(),
input->decoder->getSampleFormat());
else if (input->type == MediaDescriptor::MEDIA_VIDEO)
input->media = MediaPicture::make(
input->decoder->getWidth(),
input->decoder->getHeight(),
input->decoder->getPixelFormat()
);
}
// now, let's set up our output file.
RefPointer<Muxer> muxer = Muxer::make("EncoderTest_testTranscode.mp4", 0, 0);
RefPointer<MuxerFormat> format = muxer->getFormat();
// Let's create a helper object to help us with decoding
typedef struct {
MediaDescriptor::Type type;
RefPointer<MuxerStream> stream;
RefPointer<MediaResampler> resampler;
RefPointer<MediaSampled> media;
RefPointer<Encoder> encoder;
} MuxerStreamHelper;
MuxerStreamHelper outputHelpers[10];
for(int32_t i = 0; i < numStreams; i++) {
DemuxerStreamHelper *input = &inputHelpers[i];
MuxerStreamHelper *output = &outputHelpers[i];
if (!input->decoder)
// skip
break;
output->type = input->type;
RefPointer<Encoder> encoder;
if (output->type == MediaDescriptor::MEDIA_VIDEO) {
RefPointer<Codec> codec = Codec::findEncodingCodec(Codec::CODEC_ID_H264);
encoder = Encoder::make(codec.value());
// set the encoder properties we need
encoder->setWidth(input->decoder->getWidth());
encoder->setHeight(input->decoder->getHeight());
encoder->setPixelFormat(input->decoder->getPixelFormat());
encoder->setProperty("b", (int64_t)400000); // bitrate
encoder->setProperty("g", (int64_t) 10); // gop
encoder->setProperty("bf", (int64_t)0); // max b frames
RefPointer<Rational> tb = Rational::make(1,2997);
encoder->setTimeBase(tb.value());
} else if (output->type == MediaDescriptor::MEDIA_AUDIO) {
RefPointer<Codec> codec = Codec::findEncodingCodec(Codec::CODEC_ID_AAC);
encoder = Encoder::make(codec.value());
// set the encoder properties we need
encoder->setSampleRate(input->decoder->getSampleRate());
encoder->setSampleFormat(input->decoder->getSampleFormat());
encoder->setSampleFormat(AudioFormat::SAMPLE_FMT_S16);
encoder->setChannelLayout(input->decoder->getChannelLayout());
encoder->setChannels(input->decoder->getChannels());
encoder->setProperty("b", (int64_t)64000); // bitrate
RefPointer<Rational> tb = Rational::make(1,encoder->getSampleRate());
encoder->setTimeBase(tb.value());
// //input->decoder->getTimeBase();
// output->encoder->setTimeBase(tb.value());
}
output->encoder.reset(encoder.value(), true);
if (output->encoder) {
if (format->getFlag(MuxerFormat::GLOBAL_HEADER))
output->encoder->setFlag(Encoder::FLAG_GLOBAL_HEADER, true);
output->encoder->open(0,0);
// sometimes encoders need to change parameters to fit; let's see
// if that happened.
output->stream = muxer->addNewStream(output->encoder.value());
}
output->media = input->media;
output->resampler = 0;
if (output->type == MediaDescriptor::MEDIA_AUDIO) {
// sometimes encoders only accept certain media types and discard
// our suggestions. Let's check.
if (
output->encoder->getSampleRate() != input->decoder->getSampleRate() ||
output->encoder->getSampleFormat() != input->decoder->getSampleFormat() ||
output->encoder->getChannelLayout() != input->decoder->getChannelLayout() ||
output->encoder->getChannels() != input->decoder->getChannels()
)
{
// we need a resampler.
VS_LOG_DEBUG("Resampling: [%"PRId32", %"PRId32", %"PRId32"] [%"PRId32", %"PRId32", %"PRId32"]",
(int32_t)output->encoder->getChannelLayout(),
(int32_t)output->encoder->getSampleRate(),
(int32_t)output->encoder->getSampleFormat(),
(int32_t)input->decoder->getChannelLayout(),
(int32_t)input->decoder->getSampleRate(),
(int32_t)input->decoder->getSampleFormat()
);
RefPointer<MediaAudioResampler> resampler = MediaAudioResampler::make(
output->encoder->getChannelLayout(),
output->encoder->getSampleRate(),
output->encoder->getSampleFormat(),
input->decoder->getChannelLayout(),
input->decoder->getSampleRate(),
input->decoder->getSampleFormat()
);
resampler->open();
output->resampler.reset(resampler.value(), true);
output->media = MediaAudio::make(
output->encoder->getFrameSize(),
output->encoder->getSampleRate(),
output->encoder->getChannels(),
output->encoder->getChannelLayout(),
output->encoder->getSampleFormat());
}
}
}
// now we should be ready to open the muxer
muxer->open(0, 0);
// now, let's start a decoding loop.
RefPointer<MediaPacket> packet = MediaPacket::make();
int numPackets = 0;
while(source->read(packet.value()) >= 0) {
// got a packet; now we try to decode it.
if (packet->isComplete()) {
int32_t streamNo = packet->getStreamIndex();
DemuxerStreamHelper *input = &inputHelpers[streamNo];
MuxerStreamHelper* output = &outputHelpers[streamNo];
if (input->decoder) decodeAndEncode(
packet.value(),
input->decoder.value(),
input->media.value(),
output->resampler.value(),
output->media.value(),
muxer.value(),
output->encoder.value());
++numPackets;
if (isMemCheck && numPackets > 100) {
VS_LOG_WARN("Exiting early under valgrind");
break;
}
}
}
// now, flush any cached packets
for(int i = 0; i < numStreams; i++) {
DemuxerStreamHelper *input = &inputHelpers[i];
MuxerStreamHelper* output = &outputHelpers[i];
if (input->decoder) decodeAndEncode(
0,
input->decoder.value(),
input->media.value(),
output->resampler.value(),
output->media.value(),
muxer.value(),
output->encoder.value());
}
source->close();
muxer->close();
}
/**
* Test that memory is not leaked/corrupted during error paths.
*/
void
EncoderTest::testEncodeInvalidParameters()
{
// Sub-test 1
{
int32_t width=176;
int32_t height=360; // invalid dimensions for H263 codec
RefPointer<Codec> codec = Codec::findEncodingCodec(Codec::CODEC_ID_H263);
RefPointer<Encoder> encoder = Encoder::make(codec.value());
RefPointer<MediaPicture> picture = MediaPicture::make(width*2,height*2,
PixelFormat::PIX_FMT_YUV420P);
// set the encoder properties we need
encoder->setWidth(picture->getWidth());
encoder->setHeight(picture->getHeight());
encoder->setPixelFormat(picture->getFormat());
encoder->setProperty("b", (int64_t)400000); // bitrate
encoder->setProperty("g", (int64_t) 10); // gop
RefPointer<Rational> tb = Rational::make(1,25);
encoder->setTimeBase(tb.value());
// open the encoder
try {
// Temporarily turn down logging
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
encoder->open(0, 0);
TS_FAIL("should never get here");
} catch (std::exception & e) {
// ignore exception
}
}
// Sub-test 2
{
int32_t width=176;
int32_t height=144;
RefPointer<Codec> codec = Codec::findEncodingCodec(Codec::CODEC_ID_H264);
RefPointer<Encoder> encoder = Encoder::make(codec.value());
RefPointer<MediaPicture> picture = MediaPicture::make(width*2,height*2,
PixelFormat::PIX_FMT_YUV420P);
// set the encoder properties we need
encoder->setWidth(picture->getWidth());
encoder->setHeight(picture->getHeight());
encoder->setPixelFormat(picture->getFormat());
encoder->setProperty("b", (int64_t)400000); // bitrate
encoder->setProperty("g", (int64_t) 10); // gop
encoder->setProperty("bf", (int64_t)1); // max b frames
RefPointer<Rational> tb = Rational::make(1,25);
encoder->setTimeBase(tb.value());
// Do not open the encoder
// create an output muxer
RefPointer<Muxer> muxer = Muxer::make("EncoderTest_encodeVideo.mov", 0, 0);
// add a stream for the encoded packets
try {
// Temporarily turn down logging
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_ERROR, false);
RefPointer<MuxerStream> stream = muxer->addNewStream(encoder.value());
TS_FAIL("should never get here");
} catch (std::exception & e) {
// success
}
}
}
void
EncoderTest::testEncodeAudio() {
Logger::setGlobalIsLogging(Logger::LEVEL_TRACE, false);
LoggerStack stack;
stack.setGlobalLevel(Logger::LEVEL_INFO, false);
const bool isMemCheck = getenv("VS_TEST_MEMCHECK") ? true : false;
const int32_t sampleRate = 44100;
const int32_t maxSamples = isMemCheck ? sampleRate*0.5 : sampleRate*10;
const int32_t numSamples = 1024;
const AudioChannel::Layout channelLayout = AudioChannel::CH_LAYOUT_STEREO;
const int32_t channels = AudioChannel::getNumChannelsInLayout(channelLayout);
const AudioFormat::Type audioFormat = AudioFormat::SAMPLE_FMT_S16;
RefPointer<Codec> codec = Codec::findEncodingCodec(Codec::CODEC_ID_AAC);
RefPointer<Encoder> encoder = Encoder::make(codec.value());
RefPointer<FilterGraph> graph = FilterGraph::make();
RefPointer<MediaAudio> audio = MediaAudio::make(numSamples, sampleRate, channels, channelLayout,
audioFormat);
// set the encoder properties we need
encoder->setSampleRate(audio->getSampleRate());
encoder->setSampleFormat(audio->getFormat());
encoder->setChannelLayout(audio->getChannelLayout());
encoder->setChannels(audio->getChannels());
encoder->setProperty("b", (int64_t)64000); // bitrate
RefPointer<Rational> tb = Rational::make(1,25);
encoder->setTimeBase(tb.value());
// create an output muxer
RefPointer<Muxer> muxer = Muxer::make("EncoderTest_encodeAudio.mp4", 0, 0);
RefPointer<MuxerFormat> format = muxer->getFormat();
if (format->getFlag(MuxerFormat::GLOBAL_HEADER))
encoder->setFlag(Encoder::FLAG_GLOBAL_HEADER, true);
// open the encoder
encoder->open(0, 0);
RefPointer<FilterAudioSink> fsink = graph->addAudioSink("out", audio->getSampleRate(), audio->getChannelLayout(), audio->getFormat());
// Generate a 220 Hz sine wave with a 880 Hz beep each second, for 10 seconds.
graph->open("sine=frequency=660:beep_factor=4:duration=11[out]");
// Generate an amplitude modulated signal
//graph->open("aevalsrc=sin(10*2*PI*t)*sin(880*2*PI*t)[out]");
// add a stream for the encoded packets
{
RefPointer<MuxerStream> stream = muxer->addNewStream(encoder.value());
}
// and open the muxer
muxer->open(0, 0);
// now we're (in theory) ready to start writing data.
int32_t numFrames = 0;
RefPointer<MediaPacket> packet;
while(fsink->getAudio(audio.value()) >= 0 && audio->isComplete() && numFrames*audio->getNumSamples() < maxSamples) {
audio->setTimeStamp(numFrames*audio->getNumSamples());
// let's encode
packet = MediaPacket::make();
encoder->encodeAudio(packet.value(), audio.value());
if (packet->isComplete()) {
muxer->write(packet.value(), false);
}
++numFrames;
}
// now flush the encoder
do {
packet = MediaPacket::make();
encoder->encodeAudio(packet.value(), 0);
if (packet->isComplete()) {
muxer->write(packet.value(), false);
}
} while (packet->isComplete());
muxer->close();
}