/*
Example: AAC, 1024 frames/packet, 2112 frame offset
2112
|
Absolute frames: 0 1024 2048 | 3072
+---------+---------+--|------+---------+---------+
Packets: | 0 | 1 | | 2 | 3 | 4 |
+---------+---------+--|------+---------+---------+
Client frames: -2112 -1088 -64 | 960 SeekToFrame, TellFrame
|
0
* Offset between absolute and client frames is mFrame0Offset.
*** mFrameMark is in client frames ***
Examples:
clientFrame 0 960 1000 1024
absoluteFrame 2112 3072 3112 3136
packet 0 0 0 1
tempFrameMark* -2112 -2112 -2112 -1088
mFramesToSkipFollowingSeek 2112 3072 3112 2112
*/
void CAAudioFile::Seek(SInt64 clientFrame)
{
if (clientFrame == mFrameMark)
return; // already there! don't reset converter
//SInt64 absoluteFrame = clientFrame + mFrame0Offset;
XThrowIf(mMode != kReading || clientFrame < 0 || !mClientDataFormat.IsPCM(), kExtAudioFileError_InvalidSeek, "seek to frame in audio file");
#if VERBOSE_IO
SInt64 prevFrameMark = mFrameMark;
#endif
SInt64 packet;
packet = FrameToPacket(clientFrame);
if (packet < 0)
packet = 0;
SeekToPacket(packet);
// this will have backed up mFrameMark to match the beginning of the packet
mFramesToSkipFollowingSeek = std::max(UInt32(clientFrame - mFrameMark), UInt32(0));
mFrameMark = clientFrame;
#if VERBOSE_IO
printf("CAAudioFile::SeekToFrame: frame %qd (from %qd), packet %qd, skip %ld frames\n", mFrameMark, prevFrameMark, packet, mFramesToSkipFollowingSeek);
#endif
}
// _______________________________________________________________________________________
//
void CAAudioFile::Write(UInt32 numPackets, const AudioBufferList *data)
{
if (mIOBufferList.mBuffers[0].mData == NULL) {
#if DEBUG
printf("warning: CAAudioFile::AllocateBuffers called from WritePackets\n");
#endif
AllocateBuffers();
}
if (mMode == kPreparingToWrite)
mMode = kWriting;
else
XThrowIf(mMode != kWriting, kExtAudioFileError_InvalidOperationOrder, "can't write to this file");
if (mConverter != NULL) {
mWritePackets = numPackets;
mWriteBufferList->SetFrom(data);
WritePacketsFromCallback(WriteInputProc, this);
} else {
StartTiming(this, write);
XThrowIfError(AudioFileWritePackets(mAudioFile, mUseCache, data->mBuffers[0].mDataByteSize,
NULL, mPacketMark, &numPackets, data->mBuffers[0].mData),
"write audio file");
ElapsedTime(this, write, mTicksInIO);
#if VERBOSE_IO
printf("CAAudioFile::WritePackets: wrote %ld packets at %qd, %ld bytes\n", numPackets, mPacketMark, data->mBuffers[0].mDataByteSize);
#endif
//mNumberPackets =
mPacketMark += numPackets;
if (mFileDataFormat.mFramesPerPacket > 0)
mFrameMark += numPackets * mFileDataFormat.mFramesPerPacket;
// else: shouldn't happen since we're only called when there's no converter
}
}
void CAAudioFileConverter::OpenOutputFile(const CAStreamBasicDescription &srcFormat, const CAStreamBasicDescription &destFormat, FSRef &destFSRef, CAAudioChannelLayout &destFileLayout)
{
const ConversionParameters ¶ms = mParams;
// output file
if (params.output.filePath == NULL) {
GenerateOutputFileName(params.input.filePath, srcFormat,
destFormat, params.output.fileType, mOutName);
} else
strcpy(mOutName, params.output.filePath);
// deal with pre-existing output file
if (FSPathMakeRef((UInt8 *)mOutName, &destFSRef, NULL) == noErr) {
XThrowIf(!(params.flags & kOpt_OverwriteOutputFile), 1, "overwrite output file");
// not allowed to overwrite
// output file exists - delete it
XThrowIfError(FSDeleteObject(&destFSRef), "delete output file");
}
// get FSRef/CFStringRef for output file
FSRef outFolderFSRef;
CFStringRef outFileName;
XThrowIfError(PosixPathToParentFSRefAndName(mOutName, outFolderFSRef, outFileName), "locate output audio file");
// create the output file
mDestFile.CreateNew(outFolderFSRef, outFileName, params.output.fileType, destFormat, &destFileLayout.Layout());
CFRelease(outFileName);
}
// _______________________________________________________________________________________
//
void CAAudioFile::Open(const FSRef &fsref)
{
LOG_FUNCTION("CAAudioFile::Open", "%p", this);
XThrowIf(mMode != kClosed, kExtAudioFileError_InvalidOperationOrder, "file already open");
mFSRef = fsref;
XThrowIfError(AudioFileOpen(&mFSRef, fsRdPerm, 0, &mAudioFile), "open audio file");
mOwnOpenFile = true;
mMode = kReading;
GetExistingFileInfo();
}
// _______________________________________________________________________________________
//
void CAAudioFile::Wrap(AudioFileID fileID, bool forWriting)
{
LOG_FUNCTION("CAAudioFile::Wrap", "%p", this);
XThrowIf(mMode != kClosed, kExtAudioFileError_InvalidOperationOrder, "file already open");
mAudioFile = fileID;
mOwnOpenFile = false;
mMode = forWriting ? kPreparingToWrite : kReading;
GetExistingFileInfo();
if (forWriting)
FileFormatChanged();
}
// _______________________________________________________________________________________
//
void CAAudioFile::SetConverterChannelLayout(bool output, const CAAudioChannelLayout &layout)
{
LOG_FUNCTION("CAAudioFile::SetConverterChannelLayout", "%p", this);
OSStatus err;
if (layout.IsValid()) {
#if VERBOSE_CHANNELMAP
printf("Setting converter's %s channel layout: %s\n", output ? "output" : "input",
CAChannelLayouts::ConstantToString(mFileChannelLayout.Tag()));
#endif
if (output) {
err = AudioConverterSetProperty(mConverter, kAudioConverterOutputChannelLayout,
layout.Size(), &layout.Layout());
XThrowIf(err && err != kAudioConverterErr_OperationNotSupported, err, "couldn't set converter's output channel layout");
} else {
err = AudioConverterSetProperty(mConverter, kAudioConverterInputChannelLayout,
layout.Size(), &layout.Layout());
XThrowIf(err && err != kAudioConverterErr_OperationNotSupported, err, "couldn't set converter's input channel layout");
}
if (mMode == kPreparingToWrite)
FileFormatChanged();
}
}
// _______________________________________________________________________________________
//
void CAAudioFile::CreateNew(const FSRef &parentDir, CFStringRef filename, AudioFileTypeID filetype, const AudioStreamBasicDescription &dataFormat, const AudioChannelLayout *layout)
{
LOG_FUNCTION("CAAudioFile::CreateNew", "%p", this);
XThrowIf(mMode != kClosed, kExtAudioFileError_InvalidOperationOrder, "file already open");
mFileDataFormat = dataFormat;
if (layout) {
mFileChannelLayout = layout;
#if VERBOSE_CHANNELMAP
printf("PrepareNew passed channel layout: %s\n", CAChannelLayouts::ConstantToString(mFileChannelLayout.Tag()));
#endif
}
mMode = kPreparingToCreate;
FileFormatChanged(&parentDir, filename, filetype);
}
// _______________________________________________________________________________________
//
void CAAudioFile::WritePacketsFromCallback(
AudioConverterComplexInputDataProc inInputDataProc,
void * inInputDataProcUserData)
{
while (true) {
// keep writing until we exhaust the input (temporary stop), or produce no output (EOF)
UInt32 numEncodedPackets = mIOBufferSizePackets;
mIOBufferList.mBuffers[0].mDataByteSize = mIOBufferSizeBytes;
#if CAAUDIOFILE_PROFILE
mInConverter = true;
#endif
StartTiming(this, fill);
OSStatus err = AudioConverterFillComplexBuffer(mConverter, inInputDataProc, inInputDataProcUserData,
&numEncodedPackets, &mIOBufferList, mPacketDescs);
ElapsedTime(this, fill, mTicksInConverter);
#if CAAUDIOFILE_PROFILE
mInConverter = false;
#endif
XThrowIf(err != 0 && err != kNoMoreInputRightNow, err, "convert audio packets (write)");
if (numEncodedPackets == 0)
break;
Byte *buf = (Byte *)mIOBufferList.mBuffers[0].mData;
#if VERBOSE_IO
printf("CAAudioFile::WritePacketsFromCallback: wrote %ld packets, %ld bytes\n", numEncodedPackets, mIOBufferList.mBuffers[0].mDataByteSize);
if (mPacketDescs) {
for (UInt32 i = 0; i < numEncodedPackets; ++i) {
printf(" write packet %qd : offset %qd, length %ld\n", mPacketMark + i, mPacketDescs[i].mStartOffset, mPacketDescs[i].mDataByteSize);
#if VERBOSE_IO >= 2
hexdump(buf + mPacketDescs[i].mStartOffset, mPacketDescs[i].mDataByteSize);
#endif
}
}
#endif
StartTiming(this, write);
XThrowIfError(AudioFileWritePackets(mAudioFile, mUseCache, mIOBufferList.mBuffers[0].mDataByteSize, mPacketDescs, mPacketMark, &numEncodedPackets, buf), "write audio file");
ElapsedTime(this, write, mTicksInIO);
mPacketMark += numEncodedPackets;
//mNumberPackets += numEncodedPackets;
if (mFileDataFormat.mFramesPerPacket > 0)
mFrameMark += numEncodedPackets * mFileDataFormat.mFramesPerPacket;
else {
for (UInt32 i = 0; i < numEncodedPackets; ++i)
mFrameMark += mPacketDescs[i].mVariableFramesInPacket;
}
if (err == kNoMoreInputRightNow)
break;
}
}
void CAAudioFile::SeekToPacket(SInt64 packetNumber)
{
#if VERBOSE_IO
printf("CAAudioFile::SeekToPacket: %qd\n", packetNumber);
#endif
XThrowIf(mMode != kReading || packetNumber < 0 /*|| packetNumber >= mNumberPackets*/ , kExtAudioFileError_InvalidSeek, "seek to packet in audio file");
if (mPacketMark == packetNumber)
return; // already there! don't reset converter
mPacketMark = packetNumber;
mFrameMark = PacketToFrame(packetNumber) - mFrame0Offset;
mFramesToSkipFollowingSeek = 0;
if (mConverter)
// must reset -- if we reached end of stream. converter will no longer work otherwise
AudioConverterReset(mConverter);
}
// _______________________________________________________________________________________
// Allocates: mIOBufferList, mIOBufferSizePackets, mPacketDescs
// Dependent on: mFileMaxPacketSize, mIOBufferSizeBytes
void CAAudioFile::AllocateBuffers(bool okToFail)
{
LOG_FUNCTION("CAAudioFile::AllocateBuffers", "%p", this);
if (mFileMaxPacketSize == 0) {
if (okToFail)
return;
XThrowIf(true, kExtAudioFileError_MaxPacketSizeUnknown, "file's maximum packet size is 0");
}
UInt32 bufferSizeBytes = mIOBufferSizeBytes = std::max(mIOBufferSizeBytes, mFileMaxPacketSize);
// must be big enough for at least one maximum size packet
if (mIOBufferList.mBuffers[0].mDataByteSize != bufferSizeBytes) {
mIOBufferList.mNumberBuffers = 1;
mIOBufferList.mBuffers[0].mNumberChannels = mFileDataFormat.mChannelsPerFrame;
if (!mClientOwnsIOBuffer) {
//printf("reallocating I/O buffer\n");
delete[] (Byte *)mIOBufferList.mBuffers[0].mData;
mIOBufferList.mBuffers[0].mData = new Byte[bufferSizeBytes];
}
mIOBufferList.mBuffers[0].mDataByteSize = bufferSizeBytes;
mIOBufferSizePackets = bufferSizeBytes / mFileMaxPacketSize;
}
UInt32 propertySize = sizeof(UInt32);
UInt32 externallyFramed;
XThrowIfError(AudioFormatGetProperty(kAudioFormatProperty_FormatIsExternallyFramed,
sizeof(AudioStreamBasicDescription), &mFileDataFormat, &propertySize, &externallyFramed),
"is format externally framed");
if (mNumPacketDescs != (externallyFramed ? mIOBufferSizePackets : 0)) {
delete[] mPacketDescs;
mPacketDescs = NULL;
mNumPacketDescs = 0;
if (externallyFramed) {
//printf("reallocating packet descs\n");
mPacketDescs = new AudioStreamPacketDescription[mIOBufferSizePackets];
mNumPacketDescs = mIOBufferSizePackets;
}
}
}
// _______________________________________________________________________________________
//
void CAAudioFile::Read(UInt32 &ioNumPackets, AudioBufferList *ioData)
// May read fewer packets than requested if:
// buffer is not big enough
// file does not contain that many more packets
// Note that eofErr is not fatal, just results in 0 packets returned
// ioData's buffer sizes may be shortened
{
XThrowIf(mClientMaxPacketSize == 0, kExtAudioFileError_MaxPacketSizeUnknown, "client maximum packet size is 0");
if (mIOBufferList.mBuffers[0].mData == NULL) {
#if DEBUG
printf("warning: CAAudioFile::AllocateBuffers called from ReadPackets\n");
#endif
AllocateBuffers();
}
UInt32 bufferSizeBytes = ioData->mBuffers[0].mDataByteSize;
UInt32 maxNumPackets = bufferSizeBytes / mClientMaxPacketSize;
// older versions of AudioConverterFillComplexBuffer don't do this, so do our own sanity check
UInt32 nPackets = std::min(ioNumPackets, maxNumPackets);
mMaxPacketsToRead = ~0UL;
if (mClientDataFormat.mFramesPerPacket == 1) { // PCM or equivalent
while (mFramesToSkipFollowingSeek > 0) {
UInt32 skipFrames = std::min(mFramesToSkipFollowingSeek, maxNumPackets);
UInt32 framesPerPacket;
if ((framesPerPacket=mFileDataFormat.mFramesPerPacket) > 0)
mMaxPacketsToRead = (skipFrames + framesPerPacket - 1) / framesPerPacket;
if (mConverter == NULL) {
XThrowIfError(ReadInputProc(NULL, &skipFrames, ioData, NULL, this), "read audio file");
} else {
#if CAAUDIOFILE_PROFILE
mInConverter = true;
#endif
StartTiming(this, fill);
XThrowIfError(AudioConverterFillComplexBuffer(mConverter, ReadInputProc, this, &skipFrames, ioData, NULL), "convert audio packets (pcm read)");
ElapsedTime(this, fill, mTicksInConverter);
#if CAAUDIOFILE_PROFILE
mInConverter = false;
#endif
}
if (skipFrames == 0) { // hit EOF
ioNumPackets = 0;
return;
}
mFrameMark += skipFrames;
#if VERBOSE_IO
printf("CAAudioFile::ReadPackets: skipped %ld frames\n", skipFrames);
#endif
mFramesToSkipFollowingSeek -= skipFrames;
// restore mDataByteSize
for (int i = ioData->mNumberBuffers; --i >= 0 ; )
ioData->mBuffers[i].mDataByteSize = bufferSizeBytes;
}
}
if (mFileDataFormat.mFramesPerPacket > 0)
// don't read more packets than we are being asked to produce
mMaxPacketsToRead = nPackets / mFileDataFormat.mFramesPerPacket + 1;
if (mConverter == NULL) {
XThrowIfError(ReadInputProc(NULL, &nPackets, ioData, NULL, this), "read audio file");
} else {
#if CAAUDIOFILE_PROFILE
mInConverter = true;
#endif
StartTiming(this, fill);
XThrowIfError(AudioConverterFillComplexBuffer(mConverter, ReadInputProc, this, &nPackets, ioData, NULL), "convert audio packets (read)");
ElapsedTime(this, fill, mTicksInConverter);
#if CAAUDIOFILE_PROFILE
mInConverter = false;
#endif
}
if (mClientDataFormat.mFramesPerPacket == 1)
mFrameMark += nPackets;
ioNumPackets = nPackets;
}
// _______________________________________________________________________________________
//
void CAAudioFile::SetClientFormat(const CAStreamBasicDescription &dataFormat, const CAAudioChannelLayout *layout)
{
LOG_FUNCTION("CAAudioFile::SetClientFormat", "%p", this);
XThrowIf(!dataFormat.IsPCM(), kExtAudioFileError_NonPCMClientFormat, "non-PCM client format on audio file");
bool dataFormatChanging = (mClientDataFormat.mFormatID == 0 || mClientDataFormat != dataFormat);
if (dataFormatChanging) {
CloseConverter();
if (mWriteBufferList) {
delete mWriteBufferList;
mWriteBufferList = NULL;
}
mClientDataFormat = dataFormat;
}
if (layout && layout->IsValid()) {
XThrowIf(layout->NumberChannels() != mClientDataFormat.NumberChannels(), kExtAudioFileError_InvalidChannelMap, "inappropriate channel map");
mClientChannelLayout = *layout;
}
bool differentLayouts;
if (mClientChannelLayout.IsValid()) {
if (mFileChannelLayout.IsValid()) {
differentLayouts = mClientChannelLayout.Tag() != mFileChannelLayout.Tag();
#if VERBOSE_CHANNELMAP
printf("two valid layouts, %s\n", differentLayouts ? "different" : "same");
#endif
} else {
differentLayouts = false;
#if VERBOSE_CHANNELMAP
printf("valid client layout, unknown file layout\n");
#endif
}
} else {
differentLayouts = false;
#if VERBOSE_CHANNELMAP
if (mFileChannelLayout.IsValid())
printf("valid file layout, unknown client layout\n");
else
printf("two invalid layouts\n");
#endif
}
if (mClientDataFormat != mFileDataFormat || differentLayouts) {
// We need an AudioConverter.
if (mMode == kReading) {
// file -> client (decode)
//mFileDataFormat.PrintFormat( stdout, "", "File: ");
//mClientDataFormat.PrintFormat(stdout, "", "Client: ");
if (mConverter == NULL)
XThrowIfError(AudioConverterNew(&mFileDataFormat, &mClientDataFormat, &mConverter),
"create audio converter");
#if VERBOSE_CONVERTER
printf("CAAudioFile %p -- created converter\n", this);
CAShow(mConverter);
#endif
// set the magic cookie, if any (for decode)
if (mMagicCookie)
SetConverterProperty(kAudioConverterDecompressionMagicCookie, mMagicCookieSize, mMagicCookie, mFileDataFormat.IsPCM());
// we get cookies from some AIFF's but the converter barfs on them,
// so we set canFail to true for PCM
SetConverterChannelLayout(false, mFileChannelLayout);
SetConverterChannelLayout(true, mClientChannelLayout);
// propagate leading/trailing frame counts
if (mFileDataFormat.mBitsPerChannel == 0) {
UInt32 propertySize;
OSStatus err;
AudioFilePacketTableInfo pti;
propertySize = sizeof(pti);
err = AudioFileGetProperty(mAudioFile, kAudioFilePropertyPacketTableInfo, &propertySize, &pti);
if (err == noErr && (pti.mPrimingFrames > 0 || pti.mRemainderFrames > 0)) {
AudioConverterPrimeInfo primeInfo;
primeInfo.leadingFrames = pti.mPrimingFrames;
primeInfo.trailingFrames = pti.mRemainderFrames;
/* ignore any error. better to play it at all than not. */
/*err = */AudioConverterSetProperty(mConverter, kAudioConverterPrimeInfo, sizeof(primeInfo), &primeInfo);
//XThrowIfError(err, "couldn't set prime info on converter");
}
}
} else if (mMode == kPreparingToCreate || mMode == kPreparingToWrite) {
// client -> file (encode)
if (mConverter == NULL)
XThrowIfError(AudioConverterNew(&mClientDataFormat, &mFileDataFormat, &mConverter), "create audio converter");
mWriteBufferList = CABufferList::New("", mClientDataFormat);
SetConverterChannelLayout(false, mClientChannelLayout);
SetConverterChannelLayout(true, mFileChannelLayout);
if (mMode == kPreparingToWrite)
FileFormatChanged();
} else
XThrowIfError(kExtAudioFileError_InvalidOperationOrder, "audio file format not yet known");
}
UpdateClientMaxPacketSize();
}
void Interleave(int nInputs, const char *infilenames[], const char *outfilename, const CAAudioChannelLayout *layout)
{
const UInt32 kBufferSizeFrames = 0x8000;
const UInt32 kBufferSizeBytes = kBufferSizeFrames * sizeof(Float32);
class FileAndBuffer : public CAAudioFile {
public:
FileAndBuffer() : mBuf(NULL), mPtrs(NULL) { }
~FileAndBuffer() { delete mBuf; delete mPtrs; }
CABufferList * mBuf;
CABufferList * mPtrs;
};
FileAndBuffer *infiles = new FileAndBuffer[nInputs], *file;
FileAndBuffer outfile;
int i;
UInt32 outputChannels = 0;
double sampleRate = 0.;
UInt32 maxBitDepth = 0;
CAStreamBasicDescription clientFormat;
bool outFileCreated = false;
try {
// set up input files
for (i = 0; i < nInputs; ++i) {
file = &infiles[i];
file->Open(infilenames[i]);
const CAStreamBasicDescription &fmt = file->GetFileDataFormat();
//fmt.PrintFormat(stdout, "", "input file");
XThrowIf(fmt.mFormatID != kAudioFormatLinearPCM, -1, "input files must be PCM");
outputChannels += fmt.mChannelsPerFrame;
if (sampleRate == 0.)
sampleRate = fmt.mSampleRate;
else
XThrowIf(fmt.mSampleRate != sampleRate, -1, "input files must have the same sample rate");
if (fmt.mBitsPerChannel > maxBitDepth)
maxBitDepth = fmt.mBitsPerChannel;
clientFormat.mSampleRate = sampleRate;
clientFormat.SetCanonical(fmt.mChannelsPerFrame, false); // deinterleaved
file->SetClientFormat(clientFormat, NULL);
file->mBuf = CABufferList::New("readbuf", clientFormat);
file->mBuf->AllocateBuffers(kBufferSizeBytes);
file->mPtrs = CABufferList::New("readptrs", clientFormat);
//clientFormat.PrintFormat(stdout, "", "input client");
}
if (layout != NULL) {
if (AudioChannelLayoutTag_GetNumberOfChannels(layout->Tag()) != outputChannels) {
fprintf(stderr, "Channel layout tag '%s' is inappropriate for %u channels of audio -- aborting\n",
CAChannelLayouts::ConstantToString(layout->Tag()), (unsigned)outputChannels);
exit(2);
}
}
// prepare output file format
CAStreamBasicDescription outfmt;
outfmt.mSampleRate = sampleRate;
outfmt.mFormatID = kAudioFormatLinearPCM;
outfmt.mFormatFlags = kLinearPCMFormatFlagIsBigEndian | kLinearPCMFormatFlagIsSignedInteger | kLinearPCMFormatFlagIsPacked;
outfmt.mBitsPerChannel = maxBitDepth;
outfmt.mChannelsPerFrame = outputChannels;
outfmt.mBytesPerPacket = outfmt.mBytesPerFrame = outputChannels * (maxBitDepth >> 3);
outfmt.mFramesPerPacket = 1;
//outfmt.PrintFormat(stdout, "", "output file");
unlink(outfilename);
FSRef parentDir;
CFStringRef outName;
XThrowIfError(PosixPathToParentFSRefAndName(outfilename, parentDir, outName), "Couldn't locate output directory");
outfile.CreateNew(parentDir, outName, kAudioFileAIFFType, outfmt, layout ? &layout->Layout() : NULL);
outFileCreated = true;
// create the output file and buffers
clientFormat.mSampleRate = sampleRate;
clientFormat.SetCanonical(outputChannels, false);
outfile.SetClientFormat(clientFormat, NULL);
//clientFormat.PrintFormat(stdout, "", "output client");
outfile.mPtrs = CABufferList::New("writeptrs", clientFormat);
AudioBufferList &writebufs = outfile.mPtrs->GetModifiableBufferList();
while (true) {
UInt32 maxFramesRead = 0;
UInt32 nframes;
int outbuf = 0;
for (i = 0; i < nInputs; ++i) {
file = &infiles[i];
file->mPtrs->SetFrom(file->mBuf);
nframes = kBufferSizeFrames;
AudioBufferList &readbufs = file->mPtrs->GetModifiableBufferList();
file->Read(nframes, &readbufs);
//CAShowAudioBufferList(&readbufs, 8, 0);
if (nframes > maxFramesRead)
maxFramesRead = nframes;
if (nframes < kBufferSizeFrames)
file->mPtrs->PadWithZeroes(kBufferSizeBytes);
memcpy(&writebufs.mBuffers[outbuf], &readbufs.mBuffers[0],
readbufs.mNumberBuffers * sizeof(AudioBuffer));
outbuf += readbufs.mNumberBuffers;
}
if (maxFramesRead == 0)
break;
if (maxFramesRead < kBufferSizeFrames)
outfile.mPtrs->SetNumBytes(maxFramesRead * sizeof(Float32));
//CAShowAudioBufferList(&writebufs, 8, 0);
outfile.Write(maxFramesRead, &writebufs);
if (maxFramesRead < kBufferSizeFrames)
break;
}
}
catch (...) {
if (outFileCreated)
unlink(outfilename);
delete[] infiles;
throw;
}
outfile.Close();
// input files are closed from destructors
delete[] infiles;
}
// _______________________________________________________________________________________
//
// called to create the file -- or update its format/channel layout/properties based on an encoder
// setting change
void CAAudioFile::FileFormatChanged(const FSRef *parentDir, CFStringRef filename, AudioFileTypeID filetype)
{
LOG_FUNCTION("CAAudioFile::FileFormatChanged", "%p", this);
XThrowIf(mMode != kPreparingToCreate && mMode != kPreparingToWrite, kExtAudioFileError_InvalidOperationOrder, "new file not prepared");
UInt32 propertySize;
OSStatus err;
AudioStreamBasicDescription saveFileDataFormat = mFileDataFormat;
#if VERBOSE_CONVERTER
mFileDataFormat.PrintFormat(stdout, "", "Specified file data format");
#endif
// Find out the actual format the converter will produce. This is necessary in
// case the bitrate has forced a lower sample rate, which needs to be set correctly
// in the stream description passed to AudioFileCreate.
if (mConverter != NULL) {
propertySize = sizeof(AudioStreamBasicDescription);
Float64 origSampleRate = mFileDataFormat.mSampleRate;
XThrowIfError(AudioConverterGetProperty(mConverter, kAudioConverterCurrentOutputStreamDescription, &propertySize, &mFileDataFormat), "get audio converter's output stream description");
// do the same for the channel layout being output by the converter
#if VERBOSE_CONVERTER
mFileDataFormat.PrintFormat(stdout, "", "Converter output");
#endif
if (fiszero(mFileDataFormat.mSampleRate))
mFileDataFormat.mSampleRate = origSampleRate;
err = AudioConverterGetPropertyInfo(mConverter, kAudioConverterOutputChannelLayout, &propertySize, NULL);
if (err == noErr && propertySize > 0) {
AudioChannelLayout *layout = static_cast<AudioChannelLayout *>(malloc(propertySize));
err = AudioConverterGetProperty(mConverter, kAudioConverterOutputChannelLayout, &propertySize, layout);
if (err) {
free(layout);
XThrow(err, "couldn't get audio converter's output channel layout");
}
mFileChannelLayout = layout;
#if VERBOSE_CHANNELMAP
printf("got new file's channel layout from converter: %s\n", CAChannelLayouts::ConstantToString(mFileChannelLayout.Tag()));
#endif
free(layout);
}
}
// create the output file
if (mMode == kPreparingToCreate) {
CAStreamBasicDescription newFileDataFormat = mFileDataFormat;
if (fiszero(newFileDataFormat.mSampleRate))
newFileDataFormat.mSampleRate = 44100; // just make something up for now
#if VERBOSE_CONVERTER
newFileDataFormat.PrintFormat(stdout, "", "Applied to new file");
#endif
XThrowIfError(AudioFileCreate(parentDir, filename, filetype, &newFileDataFormat, 0, &mFSRef, &mAudioFile), "create audio file");
mMode = kPreparingToWrite;
mOwnOpenFile = true;
} else if (saveFileDataFormat != mFileDataFormat || fnotequal(saveFileDataFormat.mSampleRate, mFileDataFormat.mSampleRate)) {
// second check must be explicit since operator== on ASBD treats SR of zero as "don't care"
if (fiszero(mFileDataFormat.mSampleRate))
mFileDataFormat.mSampleRate = mClientDataFormat.mSampleRate;
#if VERBOSE_CONVERTER
mFileDataFormat.PrintFormat(stdout, "", "Applied to new file");
#endif
XThrowIf(fiszero(mFileDataFormat.mSampleRate), kExtAudioFileError_InvalidDataFormat, "file's sample rate is 0");
XThrowIfError(AudioFileSetProperty(mAudioFile, kAudioFilePropertyDataFormat, sizeof(AudioStreamBasicDescription), &mFileDataFormat), "couldn't update file's data format");
}
UInt32 deferSizeUpdates = 1;
err = AudioFileSetProperty(mAudioFile, kAudioFilePropertyDeferSizeUpdates, sizeof(UInt32), &deferSizeUpdates);
if (mConverter != NULL) {
// encoder
// get the magic cookie, if any, from the converter
delete[] mMagicCookie; mMagicCookie = NULL;
mMagicCookieSize = 0;
err = AudioConverterGetPropertyInfo(mConverter, kAudioConverterCompressionMagicCookie, &propertySize, NULL);
// we can get a noErr result and also a propertySize == 0
// -- if the file format does support magic cookies, but this file doesn't have one.
if (err == noErr && propertySize > 0) {
mMagicCookie = new Byte[propertySize];
XThrowIfError(AudioConverterGetProperty(mConverter, kAudioConverterCompressionMagicCookie, &propertySize, mMagicCookie), "get audio converter's magic cookie");
mMagicCookieSize = propertySize; // the converter lies and tell us the wrong size
// now set the magic cookie on the output file
UInt32 willEatTheCookie = false;
// the converter wants to give us one; will the file take it?
err = AudioFileGetPropertyInfo(mAudioFile, kAudioFilePropertyMagicCookieData,
NULL, &willEatTheCookie);
if (err == noErr && willEatTheCookie) {
#if VERBOSE_CONVERTER
printf("Setting cookie on encoded file\n");
#endif
XThrowIfError(AudioFileSetProperty(mAudioFile, kAudioFilePropertyMagicCookieData, mMagicCookieSize, mMagicCookie), "set audio file's magic cookie");
}
}
// get maximum packet size
propertySize = sizeof(UInt32);
XThrowIfError(AudioConverterGetProperty(mConverter, kAudioConverterPropertyMaximumOutputPacketSize, &propertySize, &mFileMaxPacketSize), "get audio converter's maximum output packet size");
AllocateBuffers(true /* okToFail */);
} else {
InitFileMaxPacketSize();
}
if (mFileChannelLayout.IsValid() && mFileChannelLayout.NumberChannels() > 2) {
// don't bother tagging mono/stereo files
UInt32 isWritable;
err = AudioFileGetPropertyInfo(mAudioFile, kAudioFilePropertyChannelLayout, NULL, &isWritable);
if (!err && isWritable) {
#if VERBOSE_CHANNELMAP
printf("writing file's channel layout: %s\n", CAChannelLayouts::ConstantToString(mFileChannelLayout.Tag()));
#endif
err = AudioFileSetProperty(mAudioFile, kAudioFilePropertyChannelLayout,
mFileChannelLayout.Size(), &mFileChannelLayout.Layout());
if (err)
CAXException::Warning("could not set the file's channel layout", err);
} else {
#if VERBOSE_CHANNELMAP
printf("file won't accept a channel layout (write)\n");
#endif
}
}
UpdateClientMaxPacketSize(); // also sets mFrame0Offset
mPacketMark = 0;
mFrameMark = 0;
}
// _______________________________________________________________________________________
//
void CAAudioFile::SetNumberFrames(SInt64 nFrames)
{
XThrowIf(mFileDataFormat.mFramesPerPacket != 1, kExtAudioFileError_InvalidDataFormat, "SetNumberFrames only supported for PCM");
XThrowIfError(AudioFileSetProperty(mAudioFile, kAudioFilePropertyAudioDataPacketCount, sizeof(SInt64), &nFrames), "Couldn't set number of packets on audio file");
}
void CAAudioFileConverter::ConvertFile(const ConversionParameters &_params)
{
FSRef destFSRef;
UInt32 propertySize;
CAStreamBasicDescription destFormat;
CAAudioChannelLayout origSrcFileLayout, srcFileLayout, destFileLayout;
bool openedSourceFile = false, createdOutputFile = false;
mParams = _params;
mReadBuffer = NULL;
mReadPtrs = NULL;
CABufferList *writeBuffer = NULL;
CABufferList *writePtrs = NULL;
PrepareConversion();
try {
if (TaggedDecodingFromCAF())
ReadCAFInfo();
OpenInputFile();
openedSourceFile = true;
// get input file's format
const CAStreamBasicDescription &srcFormat = mSrcFile.GetFileDataFormat();
if (mParams.flags & kOpt_Verbose) {
printf("Input file: %s, %qd frames\n", mParams.input.filePath ? basename(mParams.input.filePath) : "?",
mSrcFile.GetNumberFrames());
}
mSrcFormat = srcFormat;
bool encoding = !destFormat.IsPCM();
bool decoding = !srcFormat.IsPCM();
// prepare output file's format
destFormat = mParams.output.dataFormat;
if (!encoding && destFormat.mSampleRate == 0.)
// on encode, it's OK to have a 0 sample rate; ExtAudioFile will get the SR from the converter and set it on the file.
// on decode or PCM->PCM, a sample rate of 0 is interpreted as using the source sample rate
destFormat.mSampleRate = srcFormat.mSampleRate;
// source channel layout
srcFileLayout = mSrcFile.GetFileChannelLayout();
origSrcFileLayout = srcFileLayout;
if (mParams.input.channelLayoutTag != 0) {
XThrowIf(AudioChannelLayoutTag_GetNumberOfChannels(mParams.input.channelLayoutTag)
!= srcFormat.mChannelsPerFrame, -1, "input channel layout has wrong number of channels for file");
srcFileLayout = CAAudioChannelLayout(mParams.input.channelLayoutTag);
mSrcFile.SetFileChannelLayout(srcFileLayout);
}
// destination channel layout
int outChannels = mParams.output.channels;
if (mParams.output.channelLayoutTag != 0) {
// use the one specified by caller, if any
destFileLayout = CAAudioChannelLayout(mParams.output.channelLayoutTag);
} else if (srcFileLayout.IsValid()) {
// otherwise, assume the same as the source, if any
destFileLayout = srcFileLayout;
}
if (destFileLayout.IsValid()) {
// the output channel layout specifies the number of output channels
if (outChannels != -1)
XThrowIf((unsigned)outChannels != destFileLayout.NumberChannels(), -1,
"output channel layout has wrong number of channels");
else
outChannels = destFileLayout.NumberChannels();
}
if (!(mParams.flags & kOpt_NoSanitizeOutputFormat)) {
// adjust the output format's channels; output.channels overrides the channels
if (outChannels == -1)
outChannels = srcFormat.mChannelsPerFrame;
if (outChannels > 0) {
destFormat.mChannelsPerFrame = outChannels;
destFormat.mBytesPerPacket *= outChannels;
destFormat.mBytesPerFrame *= outChannels;
}
// use AudioFormat API to clean up the output format
propertySize = sizeof(AudioStreamBasicDescription);
XThrowIfError(AudioFormatGetProperty(kAudioFormatProperty_FormatInfo, 0, NULL, &propertySize, &destFormat),
"get destination format info");
}
OpenOutputFile(srcFormat, destFormat, destFSRef, destFileLayout);
createdOutputFile = true;
mDestFormat = destFormat;
// set up client formats
CAStreamBasicDescription srcClientFormat, destClientFormat;
{
CAAudioChannelLayout srcClientLayout, destClientLayout;
if (encoding) {
if (decoding) {
// transcoding
// XThrowIf(encoding && decoding, -1, "transcoding not currently supported");
if (srcFormat.mChannelsPerFrame > 2 || destFormat.mChannelsPerFrame > 2)
CAXException::Warning("Transcoding multichannel audio may not handle channel layouts correctly", 0);
srcClientFormat.SetCanonical(std::min(srcFormat.mChannelsPerFrame, destFormat.mChannelsPerFrame), true);
srcClientFormat.mSampleRate = std::max(srcFormat.mSampleRate, destFormat.mSampleRate);
mSrcFile.SetClientFormat(srcClientFormat, NULL);
destClientFormat = srcClientFormat;
} else {
// encoding
srcClientFormat = srcFormat;
destClientFormat = srcFormat;
}
// by here, destClientFormat will have a valid sample rate
destClientLayout = srcFileLayout.IsValid() ? srcFileLayout : destFileLayout;
mDestFile.SetClientFormat(destClientFormat, &destClientLayout);
} else {
// decoding or PCM->PCM
if (destFormat.mSampleRate == 0.)
destFormat.mSampleRate = srcFormat.mSampleRate;
destClientFormat = destFormat;
srcClientFormat = destFormat;
srcClientLayout = destFileLayout;
mSrcFile.SetClientFormat(srcClientFormat, &srcClientLayout);
}
}
XThrowIf(srcClientFormat.mBytesPerPacket == 0, -1, "source client format not PCM");
XThrowIf(destClientFormat.mBytesPerPacket == 0, -1, "dest client format not PCM");
if (encoding) {
// set the bitrate
if (mParams.output.bitRate != -1) {
if (mParams.flags & kOpt_Verbose)
printf("bitrate = %ld\n", mParams.output.bitRate);
mDestFile.SetConverterProperty(kAudioConverterEncodeBitRate, sizeof(UInt32), &mParams.output.bitRate);
}
// set the codec quality
if (mParams.output.codecQuality != -1) {
if (mParams.flags & kOpt_Verbose)
printf("codec quality = %ld\n", mParams.output.codecQuality);
mDestFile.SetConverterProperty(kAudioConverterCodecQuality, sizeof(UInt32), &mParams.output.codecQuality);
}
// set the bitrate strategy -- called bitrate format in the codecs since it had already shipped
if (mParams.output.strategy != -1) {
if (mParams.flags & kOpt_Verbose)
printf("strategy = %ld\n", mParams.output.strategy);
mDestFile.SetConverterProperty(kAudioCodecBitRateFormat, sizeof(UInt32), &mParams.output.strategy);
}
}
// set the SRC quality
if (mParams.output.srcQuality != -1) {
if (srcFormat.mSampleRate != 0. && destFormat.mSampleRate != 0. && srcFormat.mSampleRate != destFormat.mSampleRate) {
if (mParams.flags & kOpt_Verbose)
printf("SRC quality = %ld\n", mParams.output.srcQuality);
if (encoding)
mDestFile.SetConverterProperty(kAudioConverterSampleRateConverterQuality, sizeof(UInt32), &mParams.output.srcQuality);
else
mSrcFile.SetConverterProperty(kAudioConverterSampleRateConverterQuality, sizeof(UInt32), &mParams.output.srcQuality);
}
}
if (decoding) {
if (mParams.output.primeMethod != -1)
mSrcFile.SetConverterProperty(kAudioConverterPrimeMethod, sizeof(UInt32), &mParams.output.primeMethod);
}
PrintFormats(&origSrcFileLayout);
// prepare I/O buffers
UInt32 bytesToRead = 0x10000;
UInt32 framesToRead = bytesToRead; // OK, ReadPackets will limit as appropriate
ComputeReadSize(srcFormat, destFormat, bytesToRead, framesToRead);
// const SInt64 totalFrames = mSrcFile.GetNumberFrames();
//#warning "GetNumberFrames() can be prohibitively slow for some formats"
mReadBuffer = CABufferList::New("readbuf", srcClientFormat);
mReadBuffer->AllocateBuffers(bytesToRead);
mReadPtrs = CABufferList::New("readptrs", srcClientFormat);
BeginConversion();
while (true) {
//XThrowIf(Progress(mSrcFile.Tell(), totalFrames), userCanceledErr, "user stopped");
// this was commented out for awhile -- performance? make it optional?
UInt32 nFrames = framesToRead;
mReadPtrs->SetFrom(mReadBuffer);
AudioBufferList *readbuf = &mReadPtrs->GetModifiableBufferList();
mSrcFile.Read(nFrames, readbuf);
//printf("read %ld of %ld frames\n", nFrames, framesToRead);
if (nFrames == 0)
break;
mDestFile.Write(nFrames, readbuf);
if (ShouldTerminateConversion())
break;
}
if (decoding) {
// fix up the destination file's length if necessary and possible
SInt64 nframes = mSrcFile.GetNumberFrames();
if (nframes != 0) {
// only shorten, don't try to lengthen
nframes = SInt64(ceil(nframes * destFormat.mSampleRate / srcFormat.mSampleRate));
if (nframes < mDestFile.GetNumberFrames()) {
mDestFile.SetNumberFrames(nframes);
}
}
}
EndConversion();
}
catch (...) {
delete mReadBuffer;
delete mReadPtrs;
delete writeBuffer;
delete writePtrs;
if (!createdOutputFile)
PrintFormats(&origSrcFileLayout);
try { mSrcFile.Close(); } catch (...) { }
try { mDestFile.Close(); } catch (...) { }
if (createdOutputFile)
unlink(mOutName);
throw;
}
delete mReadBuffer;
delete mReadPtrs;
delete writeBuffer;
delete writePtrs;
mSrcFile.Close();
mDestFile.Close();
if (TaggedEncodingToCAF())
WriteCAFInfo();
if (mParams.flags & kOpt_Verbose) {
// must close to flush encoder; GetNumberFrames() not necessarily valid until afterwards but then
// the file is closed
CAAudioFile temp;
FSRef destFSRef;
if (FSPathMakeRef((UInt8 *)mOutName, &destFSRef, NULL) == noErr) {
temp.Open(destFSRef);
printf("Output file: %s, %qd frames\n", basename(mOutName), temp.GetNumberFrames());
}
}
}
CAAudioFileRecorder::CAAudioFileRecorder(int nBuffers, UInt32 bufferSizeFrames) :
CAAudioFileWriter(nBuffers, bufferSizeFrames),
mInputUnit(NULL),
mAudioInputPtrs(NULL)
{
// open input unit
Component comp;
ComponentDescription desc;
desc.componentType = kAudioUnitType_Output;
desc.componentSubType = kAudioUnitSubType_HALOutput;
desc.componentManufacturer = kAudioUnitManufacturer_Apple;
desc.componentFlags = 0;
desc.componentFlagsMask = 0;
comp = FindNextComponent(NULL, &desc);
XThrowIf(comp == NULL, -1, "find audio input unit");
XThrowIfError(OpenAComponent(comp, &mInputUnit), "open audio input unit");
UInt32 enableIO;
UInt32 propSize;
enableIO = 0;
XThrowIfError(AudioUnitSetProperty(mInputUnit,
kAudioOutputUnitProperty_EnableIO,
kAudioUnitScope_Output,
0,
&enableIO,
sizeof(enableIO)), "failed to disable output");
enableIO = 1;
XThrowIfError(AudioUnitSetProperty(mInputUnit,
kAudioOutputUnitProperty_EnableIO,
kAudioUnitScope_Input,
1,
&enableIO,
sizeof(enableIO)), "failed to enable input");
// select the default input device
propSize = sizeof(AudioDeviceID);
AudioDeviceID inputDevice;
XThrowIfError(
AudioHardwareGetProperty(kAudioHardwarePropertyDefaultInputDevice, &propSize, &inputDevice),
"failed to get default input device");
XThrowIfError(
AudioUnitSetProperty(mInputUnit, kAudioOutputUnitProperty_CurrentDevice, kAudioUnitScope_Global, 0, &inputDevice, sizeof(inputDevice)),
"failed to select input device");
// set render callback
AURenderCallbackStruct input;
input.inputProc = InputProc;
input.inputProcRefCon = this;
XThrowIfError(AudioUnitSetProperty(
mInputUnit,
kAudioOutputUnitProperty_SetInputCallback,
kAudioUnitScope_Global,
0,
&input,
sizeof(input)), "connect input proc to output unit");
XThrowIfError(AudioUnitInitialize(mInputUnit), "initialize audio input unit");
// get the hardware format
propSize = sizeof(mInputDataFormat);
XThrowIfError(AudioUnitGetProperty(mInputUnit, kAudioUnitProperty_StreamFormat, kAudioUnitScope_Input, 1, &mInputDataFormat, &propSize), "couldn't get input unit's stream format");
}
