tagparser/mp4/mp4track.cpp

2014 lines
90 KiB
C++

#include "./mp4track.h"
#include "./mp4atom.h"
#include "./mp4container.h"
#include "./mp4ids.h"
#include "./mpeg4descriptor.h"
#include "../avc/avcconfiguration.h"
#include "../mpegaudio/mpegaudioframe.h"
#include "../mpegaudio/mpegaudioframestream.h"
#include "../exceptions.h"
#include "../mediaformat.h"
#include <c++utilities/conversion/stringbuilder.h>
#include <c++utilities/io/binaryreader.h>
#include <c++utilities/io/binarywriter.h>
#include <c++utilities/io/bitreader.h>
#include <c++utilities/io/catchiofailure.h>
#include <cmath>
#include <locale>
using namespace std;
using namespace IoUtilities;
using namespace ConversionUtilities;
using namespace ChronoUtilities;
namespace TagParser {
/*!
* \brief The TrackHeaderInfo struct holds information about the present track header (tkhd atom) and
* information for making a new track header based on it.
* \sa TrackHeaderInfo Mp4Track::verifyPresentTrackHeader() for obtaining an instance.
* \remarks The struct is only used internally by the Mp4Track class.
*/
struct TrackHeaderInfo {
friend class Mp4Track;
private:
constexpr TrackHeaderInfo();
/// \brief Specifies the size which is required for <i>making a new</i> track header based one the existing one.
uint64 requiredSize;
/// \brief Specifies whether there actually a track header exists and whether it can be used as basis for a new one.
bool canUseExisting;
/// \brief Specifies whether the existing track header is truncated.
bool truncated;
/// \brief Specifies the version of the existing track header.
byte version;
/// \brief Specifies whether the version of the existing track header is unknown (and assumed to be 1).
bool versionUnknown;
/// \brief Specifies the additional data offset of the existing header. Unspecified if canUseExisting is false.
byte additionalDataOffset;
/// \brief Specifies whether the buffered header data should be discarded when making a new track header.
bool discardBuffer;
};
constexpr TrackHeaderInfo::TrackHeaderInfo()
: requiredSize(100)
, canUseExisting(false)
, truncated(false)
, version(0)
, versionUnknown(false)
, additionalDataOffset(0)
, discardBuffer(false)
{
}
/// \brief Dates within MP4 tracks are expressed as the number of seconds since this date.
const DateTime startDate = DateTime::fromDate(1904, 1, 1);
/*!
* \class Mpeg4AudioSpecificConfig
* \brief The Mpeg4AudioSpecificConfig class holds MPEG-4 audio specific config parsed using Mp4Track::parseAudioSpecificConfig().
* \remarks Is part of Mpeg4ElementaryStreamInfo (audio streams only).
*/
Mpeg4AudioSpecificConfig::Mpeg4AudioSpecificConfig()
: audioObjectType(0)
, sampleFrequencyIndex(0xF)
, sampleFrequency(0)
, channelConfiguration(0)
, extensionAudioObjectType(0)
, sbrPresent(false)
, psPresent(false)
, extensionSampleFrequencyIndex(0xF)
, extensionSampleFrequency(0)
, extensionChannelConfiguration(0)
, frameLengthFlag(false)
, dependsOnCoreCoder(false)
, coreCoderDelay(0)
, extensionFlag(0)
, layerNr(0)
, numOfSubFrame(0)
, layerLength(0)
, resilienceFlags(0)
, epConfig(0)
{
}
/*!
* \class Mpeg4VideoSpecificConfig
* \brief The Mpeg4VideoSpecificConfig class holds MPEG-4 video specific config parsed using Mp4Track::parseVideoSpecificConfig().
* \remarks
* - Is part of Mpeg4ElementaryStreamInfo (video streams only).
* - AVC configuration is another thing and covered by the AvcConfiguration class.
*/
Mpeg4VideoSpecificConfig::Mpeg4VideoSpecificConfig()
: profile(0)
{
}
/*!
* \class Mpeg4ElementaryStreamInfo
* \brief The Mpeg4ElementaryStreamInfo class holds MPEG-4 elementary stream info parsed using Mp4Track::parseMpeg4ElementaryStreamInfo().
*/
/*!
* \class TagParser::Mp4Track
* \brief Implementation of TagParser::AbstractTrack for the MP4 container.
*/
/*!
* \brief Constructs a new track for the specified \a trakAtom.
*
* "trak"-atoms are stored in the top-level atom "move". Each "trak"-atom holds
* header information for one track in the MP4 file.
*/
Mp4Track::Mp4Track(Mp4Atom &trakAtom)
: AbstractTrack(trakAtom.stream(), trakAtom.startOffset())
, m_trakAtom(&trakAtom)
, m_tkhdAtom(nullptr)
, m_mdiaAtom(nullptr)
, m_mdhdAtom(nullptr)
, m_hdlrAtom(nullptr)
, m_minfAtom(nullptr)
, m_stblAtom(nullptr)
, m_stsdAtom(nullptr)
, m_stscAtom(nullptr)
, m_stcoAtom(nullptr)
, m_stszAtom(nullptr)
,
//m_codecConfigAtom(nullptr),
//m_esDescAtom(nullptr),
m_framesPerSample(1)
, m_chunkOffsetSize(4)
, m_chunkCount(0)
, m_sampleToChunkEntryCount(0)
{
}
/*!
* \brief Destroys the track.
*/
Mp4Track::~Mp4Track()
{
}
TrackType Mp4Track::type() const
{
return TrackType::Mp4Track;
}
/*!
* \brief Reads the chunk offsets from the stco atom and fragments if \a parseFragments is true.
* \returns Returns the chunk offset table for the track.
* \throws Throws InvalidDataException when
* - there is no stream assigned.
* - the header has been considered as invalid when parsing the header information.
* - the determined chunk offset size is invalid.
* \throws Throws std::ios_base::failure when an IO error occurs.
* \sa readChunkSizes();
*/
std::vector<uint64> Mp4Track::readChunkOffsets(bool parseFragments, Diagnostics &diag)
{
static const string context("reading chunk offset table of MP4 track");
if (!isHeaderValid() || !m_istream) {
diag.emplace_back(DiagLevel::Critical, "Track has not been parsed.", context);
throw InvalidDataException();
}
vector<uint64> offsets;
if (m_stcoAtom) {
// verify integrity of the chunk offset table
uint64 actualTableSize = m_stcoAtom->dataSize();
if (actualTableSize < (8 + chunkOffsetSize())) {
diag.emplace_back(DiagLevel::Critical, "The stco atom is truncated. There are no chunk offsets present.", context);
throw InvalidDataException();
} else {
actualTableSize -= 8;
}
uint32 actualChunkCount = chunkCount();
uint64 calculatedTableSize = chunkCount() * chunkOffsetSize();
if (calculatedTableSize < actualTableSize) {
diag.emplace_back(
DiagLevel::Critical, "The stco atom stores more chunk offsets as denoted. The additional chunk offsets will be ignored.", context);
} else if (calculatedTableSize > actualTableSize) {
diag.emplace_back(DiagLevel::Critical, "The stco atom is truncated. It stores less chunk offsets as denoted.", context);
actualChunkCount = floor(static_cast<double>(actualTableSize) / static_cast<double>(chunkOffsetSize()));
}
// read the table
offsets.reserve(actualChunkCount);
m_istream->seekg(m_stcoAtom->dataOffset() + 8);
switch (chunkOffsetSize()) {
case 4:
for (uint32 i = 0; i < actualChunkCount; ++i) {
offsets.push_back(reader().readUInt32BE());
}
break;
case 8:
for (uint32 i = 0; i < actualChunkCount; ++i) {
offsets.push_back(reader().readUInt64BE());
}
break;
default:
diag.emplace_back(DiagLevel::Critical, "The determined chunk offset size is invalid.", context);
throw InvalidDataException();
}
}
// read sample offsets of fragments
if (parseFragments) {
uint64 totalDuration = 0;
for (Mp4Atom *moofAtom = m_trakAtom->container().firstElement()->siblingByIdIncludingThis(Mp4AtomIds::MovieFragment, diag); moofAtom;
moofAtom = moofAtom->siblingById(Mp4AtomIds::MovieFragment, diag)) {
moofAtom->parse(diag);
for (Mp4Atom *trafAtom = moofAtom->childById(Mp4AtomIds::TrackFragment, diag); trafAtom;
trafAtom = trafAtom->siblingById(Mp4AtomIds::TrackFragment, diag)) {
trafAtom->parse(diag);
for (Mp4Atom *tfhdAtom = trafAtom->childById(Mp4AtomIds::TrackFragmentHeader, diag); tfhdAtom;
tfhdAtom = tfhdAtom->siblingById(Mp4AtomIds::TrackFragmentHeader, diag)) {
tfhdAtom->parse(diag);
uint32 calculatedDataSize = 0;
if (tfhdAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "tfhd atom is truncated.", context);
} else {
inputStream().seekg(tfhdAtom->dataOffset() + 1);
const uint32 flags = reader().readUInt24BE();
if (m_id == reader().readUInt32BE()) { // check track ID
if (flags & 0x000001) { // base-data-offset present
calculatedDataSize += 8;
}
if (flags & 0x000002) { // sample-description-index present
calculatedDataSize += 4;
}
if (flags & 0x000008) { // default-sample-duration present
calculatedDataSize += 4;
}
if (flags & 0x000010) { // default-sample-size present
calculatedDataSize += 4;
}
if (flags & 0x000020) { // default-sample-flags present
calculatedDataSize += 4;
}
// some variables are currently skipped because they are currently not interesting
//uint64 baseDataOffset = moofAtom->startOffset();
//uint32 defaultSampleDescriptionIndex = 0;
uint32 defaultSampleDuration = 0;
uint32 defaultSampleSize = 0;
//uint32 defaultSampleFlags = 0;
if (tfhdAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "tfhd atom is truncated (presence of fields denoted).", context);
} else {
if (flags & 0x000001) { // base-data-offset present
//baseDataOffset = reader.readUInt64();
inputStream().seekg(8, ios_base::cur);
}
if (flags & 0x000002) { // sample-description-index present
//defaultSampleDescriptionIndex = reader.readUInt32();
inputStream().seekg(4, ios_base::cur);
}
if (flags & 0x000008) { // default-sample-duration present
defaultSampleDuration = reader().readUInt32BE();
//inputStream().seekg(4, ios_base::cur);
}
if (flags & 0x000010) { // default-sample-size present
defaultSampleSize = reader().readUInt32BE();
}
if (flags & 0x000020) { // default-sample-flags present
//defaultSampleFlags = reader().readUInt32BE();
inputStream().seekg(4, ios_base::cur);
}
}
for (Mp4Atom *trunAtom = trafAtom->childById(Mp4AtomIds::TrackFragmentRun, diag); trunAtom;
trunAtom = trunAtom->siblingById(Mp4AtomIds::TrackFragmentRun, diag)) {
uint32 calculatedDataSize = 8;
if (trunAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "trun atom is truncated.", context);
} else {
inputStream().seekg(trunAtom->dataOffset() + 1);
uint32 flags = reader().readUInt24BE();
uint32 sampleCount = reader().readUInt32BE();
m_sampleCount += sampleCount;
if (flags & 0x000001) { // data offset present
calculatedDataSize += 4;
}
if (flags & 0x000004) { // first-sample-flags present
calculatedDataSize += 4;
}
uint32 entrySize = 0;
if (flags & 0x000100) { // sample-duration present
entrySize += 4;
}
if (flags & 0x000200) { // sample-size present
entrySize += 4;
}
if (flags & 0x000400) { // sample-flags present
entrySize += 4;
}
if (flags & 0x000800) { // sample-composition-time-offsets present
entrySize += 4;
}
calculatedDataSize += entrySize * sampleCount;
if (trunAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "trun atom is truncated (presence of fields denoted).", context);
} else {
if (flags & 0x000001) { // data offset present
inputStream().seekg(4, ios_base::cur);
//int32 dataOffset = reader().readInt32BE();
}
if (flags & 0x000004) { // first-sample-flags present
inputStream().seekg(4, ios_base::cur);
}
for (uint32 i = 0; i < sampleCount; ++i) {
if (flags & 0x000100) { // sample-duration present
totalDuration += reader().readUInt32BE();
} else {
totalDuration += defaultSampleDuration;
}
if (flags & 0x000200) { // sample-size present
m_sampleSizes.push_back(reader().readUInt32BE());
m_size += m_sampleSizes.back();
} else {
m_size += defaultSampleSize;
}
if (flags & 0x000400) { // sample-flags present
inputStream().seekg(4, ios_base::cur);
}
if (flags & 0x000800) { // sample-composition-time-offsets present
inputStream().seekg(4, ios_base::cur);
}
}
}
}
}
if (m_sampleSizes.empty() && defaultSampleSize) {
m_sampleSizes.push_back(defaultSampleSize);
}
}
}
}
}
}
}
return offsets;
}
/*!
* \brief Accumulates \a count sample sizes from the specified \a sampleSizeTable starting at the specified \a sampleIndex.
* \remarks This helper function is used by the addChunkSizeEntries() method.
*/
uint64 Mp4Track::accumulateSampleSizes(size_t &sampleIndex, size_t count, Diagnostics &diag)
{
if (sampleIndex + count <= m_sampleSizes.size()) {
uint64 sum = 0;
for (size_t end = sampleIndex + count; sampleIndex < end; ++sampleIndex) {
sum += m_sampleSizes[sampleIndex];
}
return sum;
} else if (m_sampleSizes.size() == 1) {
sampleIndex += count;
return static_cast<uint64>(m_sampleSizes.front()) * count;
} else {
diag.emplace_back(DiagLevel::Critical, "There are not as many sample size entries as samples.", "reading chunk sizes of MP4 track");
throw InvalidDataException();
}
}
/*!
* \brief Adds chunks size entries to the specified \a chunkSizeTable.
* \param chunkSizeTable Specifies the chunk size table. The chunks sizes will be added to this table.
* \param count Specifies the number of chunks to be added. The size of \a chunkSizeTable is increased this value.
* \param sampleIndex Specifies the index of the first sample in the \a sampleSizeTable; is increased by \a count * \a sampleCount.
* \param sampleSizeTable Specifies the table holding the sample sizes.
* \remarks This helper function is used by the readChunkSizes() method.
*/
void Mp4Track::addChunkSizeEntries(std::vector<uint64> &chunkSizeTable, size_t count, size_t &sampleIndex, uint32 sampleCount, Diagnostics &diag)
{
for (size_t i = 0; i < count; ++i) {
chunkSizeTable.push_back(accumulateSampleSizes(sampleIndex, sampleCount, diag));
}
}
/*!
* \brief Verifies the present track header (tkhd atom) and returns relevant information for making a new track header
* based on it.
*/
TrackHeaderInfo Mp4Track::verifyPresentTrackHeader() const
{
TrackHeaderInfo info;
// return the default TrackHeaderInfo in case there is no track header prsent
if (!m_tkhdAtom) {
return info;
}
// ensure the tkhd atom is buffered but mark the buffer to be discarded again if it has not been present
info.discardBuffer = m_tkhdAtom->buffer() == nullptr;
if (info.discardBuffer) {
m_tkhdAtom->makeBuffer();
}
// check the version of the existing tkhd atom to determine where additional data starts
switch (info.version = static_cast<byte>(m_tkhdAtom->buffer()[m_tkhdAtom->headerSize()])) {
case 0:
info.additionalDataOffset = 32;
break;
case 1:
info.additionalDataOffset = 44;
break;
default:
info.additionalDataOffset = 44;
info.versionUnknown = true;
}
// check whether the existing tkhd atom is not truncated
if (info.additionalDataOffset + 48u <= m_tkhdAtom->dataSize()) {
info.canUseExisting = true;
} else {
info.truncated = true;
info.canUseExisting = info.additionalDataOffset < m_tkhdAtom->dataSize();
if (!info.canUseExisting && info.discardBuffer) {
m_tkhdAtom->discardBuffer();
}
}
// determine required size
info.requiredSize = m_tkhdAtom->dataSize() + 8;
if (info.version == 0) {
// add 12 byte to size if the existing version is 0 because we always write version 1 which takes 12 byte more space
info.requiredSize += 12;
}
if (info.requiredSize > numeric_limits<uint32>::max()) {
// add 8 byte to the size because it must be denoted using a 64-bit integer
info.requiredSize += 8;
}
return info;
}
/*!
* \brief Reads the sample to chunk table.
* \returns Returns a vector with the table entries wrapped using the tuple container. The first value
* is an integer that gives the first chunk that share the same samples count and sample description index.
* The second value is sample cound and the third value the sample description index.
* \remarks The table is not validated.
*/
vector<tuple<uint32, uint32, uint32>> Mp4Track::readSampleToChunkTable(Diagnostics &diag)
{
static const string context("reading sample to chunk table of MP4 track");
if (!isHeaderValid() || !m_istream || !m_stscAtom) {
diag.emplace_back(DiagLevel::Critical, "Track has not been parsed or is invalid.", context);
throw InvalidDataException();
}
// verify integrity of the sample to chunk table
uint64 actualTableSize = m_stscAtom->dataSize();
if (actualTableSize < 20) {
diag.emplace_back(DiagLevel::Critical, "The stsc atom is truncated. There are no \"sample to chunk\" entries present.", context);
throw InvalidDataException();
} else {
actualTableSize -= 8;
}
uint32 actualSampleToChunkEntryCount = sampleToChunkEntryCount();
uint64 calculatedTableSize = actualSampleToChunkEntryCount * 12;
if (calculatedTableSize < actualTableSize) {
diag.emplace_back(DiagLevel::Critical, "The stsc atom stores more entries as denoted. The additional entries will be ignored.", context);
} else if (calculatedTableSize > actualTableSize) {
diag.emplace_back(DiagLevel::Critical, "The stsc atom is truncated. It stores less entries as denoted.", context);
// FIXME: floor and cast actually required?
actualSampleToChunkEntryCount = floor(static_cast<double>(actualTableSize) / 12.0);
}
// prepare reading
vector<tuple<uint32, uint32, uint32>> sampleToChunkTable;
sampleToChunkTable.reserve(actualSampleToChunkEntryCount);
m_istream->seekg(m_stscAtom->dataOffset() + 8);
for (uint32 i = 0; i < actualSampleToChunkEntryCount; ++i) {
// read entry
uint32 firstChunk = reader().readUInt32BE();
uint32 samplesPerChunk = reader().readUInt32BE();
uint32 sampleDescriptionIndex = reader().readUInt32BE();
sampleToChunkTable.emplace_back(firstChunk, samplesPerChunk, sampleDescriptionIndex);
}
return sampleToChunkTable;
}
/*!
* \brief Reads the chunk sizes from the stsz (sample sizes) and stsc (samples per chunk) atom.
* \returns Returns the chunk sizes for the track.
*
* \throws Throws InvalidDataException when
* - there is no stream assigned.
* - the header has been considered as invalid when parsing the header information.
* - the determined chunk offset size is invalid.
* \throws Throws std::ios_base::failure when an IO error occurs.
*
* \sa readChunkOffsets();
*/
vector<uint64> Mp4Track::readChunkSizes(Diagnostics &diag)
{
static const string context("reading chunk sizes of MP4 track");
if (!isHeaderValid() || !m_istream || !m_stcoAtom) {
diag.emplace_back(DiagLevel::Critical, "Track has not been parsed or is invalid.", context);
throw InvalidDataException();
}
// read sample to chunk table
const auto sampleToChunkTable = readSampleToChunkTable(diag);
// accumulate chunk sizes from the table
vector<uint64> chunkSizes;
if (!sampleToChunkTable.empty()) {
// prepare reading
auto tableIterator = sampleToChunkTable.cbegin();
chunkSizes.reserve(m_chunkCount);
// read first entry
size_t sampleIndex = 0;
uint32 previousChunkIndex = get<0>(*tableIterator); // the first chunk has the index 1 and not zero!
if (previousChunkIndex != 1) {
diag.emplace_back(DiagLevel::Critical, "The first chunk of the first \"sample to chunk\" entry must be 1.", context);
previousChunkIndex = 1; // try to read the entry anyway
}
uint32 samplesPerChunk = get<1>(*tableIterator);
// read the following entries
++tableIterator;
for (const auto tableEnd = sampleToChunkTable.cend(); tableIterator != tableEnd; ++tableIterator) {
uint32 firstChunkIndex = get<0>(*tableIterator);
if (firstChunkIndex > previousChunkIndex && firstChunkIndex <= m_chunkCount) {
addChunkSizeEntries(chunkSizes, firstChunkIndex - previousChunkIndex, sampleIndex, samplesPerChunk, diag);
} else {
diag.emplace_back(DiagLevel::Critical,
"The first chunk index of a \"sample to chunk\" entry must be greather than the first chunk of the previous entry and not "
"greather than the chunk count.",
context);
throw InvalidDataException();
}
previousChunkIndex = firstChunkIndex;
samplesPerChunk = get<1>(*tableIterator);
}
if (m_chunkCount >= previousChunkIndex) {
addChunkSizeEntries(chunkSizes, m_chunkCount + 1 - previousChunkIndex, sampleIndex, samplesPerChunk, diag);
}
}
return chunkSizes;
}
/*!
* \brief Reads the MPEG-4 elementary stream descriptor for the track.
* \remarks
* - Notifications might be added.
* \sa mpeg4ElementaryStreamInfo()
*/
std::unique_ptr<Mpeg4ElementaryStreamInfo> Mp4Track::parseMpeg4ElementaryStreamInfo(
IoUtilities::BinaryReader &reader, Mp4Atom *esDescAtom, Diagnostics &diag)
{
static const string context("parsing MPEG-4 elementary stream descriptor");
using namespace Mpeg4ElementaryStreamObjectIds;
unique_ptr<Mpeg4ElementaryStreamInfo> esInfo;
if (esDescAtom->dataSize() >= 12) {
reader.stream()->seekg(esDescAtom->dataOffset());
// read version/flags
if (reader.readUInt32BE() != 0) {
diag.emplace_back(DiagLevel::Warning, "Unknown version/flags.", context);
}
// read extended descriptor
Mpeg4Descriptor esDesc(esDescAtom->container(), reader.stream()->tellg(), esDescAtom->dataSize() - 4);
try {
esDesc.parse(diag);
// check ID
if (esDesc.id() != Mpeg4DescriptorIds::ElementaryStreamDescr) {
diag.emplace_back(DiagLevel::Critical, "Invalid descriptor found.", context);
throw Failure();
}
// read stream info
reader.stream()->seekg(esDesc.dataOffset());
esInfo = make_unique<Mpeg4ElementaryStreamInfo>();
esInfo->id = reader.readUInt16BE();
esInfo->esDescFlags = reader.readByte();
if (esInfo->dependencyFlag()) {
esInfo->dependsOnId = reader.readUInt16BE();
}
if (esInfo->urlFlag()) {
esInfo->url = reader.readString(reader.readByte());
}
if (esInfo->ocrFlag()) {
esInfo->ocrId = reader.readUInt16BE();
}
for (Mpeg4Descriptor *esDescChild = esDesc.denoteFirstChild(static_cast<uint64>(reader.stream()->tellg()) - esDesc.startOffset());
esDescChild; esDescChild = esDescChild->nextSibling()) {
esDescChild->parse(diag);
switch (esDescChild->id()) {
case Mpeg4DescriptorIds::DecoderConfigDescr:
// read decoder config descriptor
reader.stream()->seekg(esDescChild->dataOffset());
esInfo->objectTypeId = reader.readByte();
esInfo->decCfgDescFlags = reader.readByte();
esInfo->bufferSize = reader.readUInt24BE();
esInfo->maxBitrate = reader.readUInt32BE();
esInfo->averageBitrate = reader.readUInt32BE();
for (Mpeg4Descriptor *decCfgDescChild = esDescChild->denoteFirstChild(esDescChild->headerSize() + 13); decCfgDescChild;
decCfgDescChild = decCfgDescChild->nextSibling()) {
decCfgDescChild->parse(diag);
switch (decCfgDescChild->id()) {
case Mpeg4DescriptorIds::DecoderSpecificInfo:
// read decoder specific info
switch (esInfo->objectTypeId) {
case Aac:
case Mpeg2AacMainProfile:
case Mpeg2AacLowComplexityProfile:
case Mpeg2AacScaleableSamplingRateProfile:
case Mpeg2Audio:
case Mpeg1Audio:
esInfo->audioSpecificConfig
= parseAudioSpecificConfig(*reader.stream(), decCfgDescChild->dataOffset(), decCfgDescChild->dataSize(), diag);
break;
case Mpeg4Visual:
esInfo->videoSpecificConfig
= parseVideoSpecificConfig(reader, decCfgDescChild->dataOffset(), decCfgDescChild->dataSize(), diag);
break;
default:; // TODO: cover more object types
}
break;
}
}
break;
case Mpeg4DescriptorIds::SlConfigDescr:
// uninteresting
break;
}
}
} catch (const Failure &) {
diag.emplace_back(DiagLevel::Critical, "The MPEG-4 descriptor element structure is invalid.", context);
}
} else {
diag.emplace_back(DiagLevel::Warning, "Elementary stream descriptor atom (esds) is truncated.", context);
}
return esInfo;
}
/*!
* \brief Parses the audio specific configuration for the track.
* \remarks
* - Notifications might be added.
* \sa mpeg4ElementaryStreamInfo()
*/
unique_ptr<Mpeg4AudioSpecificConfig> Mp4Track::parseAudioSpecificConfig(istream &stream, uint64 startOffset, uint64 size, Diagnostics &diag)
{
static const string context("parsing MPEG-4 audio specific config from elementary stream descriptor");
using namespace Mpeg4AudioObjectIds;
// read config into buffer and construct BitReader for bitwise reading
stream.seekg(startOffset);
auto buff = make_unique<char[]>(size);
stream.read(buff.get(), size);
BitReader bitReader(buff.get(), size);
auto audioCfg = make_unique<Mpeg4AudioSpecificConfig>();
try {
// read audio object type
auto getAudioObjectType = [&bitReader] {
byte objType = bitReader.readBits<byte>(5);
if (objType == 31) {
objType = 32 + bitReader.readBits<byte>(6);
}
return objType;
};
audioCfg->audioObjectType = getAudioObjectType();
// read sampling frequency
if ((audioCfg->sampleFrequencyIndex = bitReader.readBits<byte>(4)) == 0xF) {
audioCfg->sampleFrequency = bitReader.readBits<uint32>(24);
}
// read channel config
audioCfg->channelConfiguration = bitReader.readBits<byte>(4);
// read extension header
switch (audioCfg->audioObjectType) {
case Sbr:
case Ps:
audioCfg->extensionAudioObjectType = audioCfg->audioObjectType;
audioCfg->sbrPresent = true;
if ((audioCfg->extensionSampleFrequencyIndex = bitReader.readBits<byte>(4)) == 0xF) {
audioCfg->extensionSampleFrequency = bitReader.readBits<uint32>(24);
}
if ((audioCfg->audioObjectType = getAudioObjectType()) == ErBsac) {
audioCfg->extensionChannelConfiguration = bitReader.readBits<byte>(4);
}
break;
}
switch (audioCfg->extensionAudioObjectType) {
case Ps:
audioCfg->psPresent = true;
audioCfg->extensionChannelConfiguration = Mpeg4ChannelConfigs::FrontLeftFrontRight;
break;
}
// read GA specific config
switch (audioCfg->audioObjectType) {
case AacMain:
case AacLc:
case AacLtp:
case AacScalable:
case TwinVq:
case ErAacLc:
case ErAacLtp:
case ErAacScalable:
case ErTwinVq:
case ErBsac:
case ErAacLd:
audioCfg->frameLengthFlag = bitReader.readBits<byte>(1);
if ((audioCfg->dependsOnCoreCoder = bitReader.readBit())) {
audioCfg->coreCoderDelay = bitReader.readBits<byte>(14);
}
audioCfg->extensionFlag = bitReader.readBit();
if (audioCfg->channelConfiguration == 0) {
throw NotImplementedException(); // TODO: parse program_config_element
}
switch (audioCfg->audioObjectType) {
case AacScalable:
case ErAacScalable:
audioCfg->layerNr = bitReader.readBits<byte>(3);
break;
default:;
}
if (audioCfg->extensionFlag == 1) {
switch (audioCfg->audioObjectType) {
case ErBsac:
audioCfg->numOfSubFrame = bitReader.readBits<byte>(5);
audioCfg->layerLength = bitReader.readBits<uint16>(11);
break;
case ErAacLc:
case ErAacLtp:
case ErAacScalable:
case ErAacLd:
audioCfg->resilienceFlags = bitReader.readBits<byte>(3);
break;
default:;
}
if (bitReader.readBit() == 1) { // extension flag 3
throw NotImplementedException(); // TODO
}
}
break;
default:
throw NotImplementedException(); // TODO: cover remaining object types
}
// read error specific config
switch (audioCfg->audioObjectType) {
case ErAacLc:
case ErAacLtp:
case ErAacScalable:
case ErTwinVq:
case ErBsac:
case ErAacLd:
case ErCelp:
case ErHvxc:
case ErHiln:
case ErParametric:
case ErAacEld:
switch (audioCfg->epConfig = bitReader.readBits<byte>(2)) {
case 2:
break;
case 3:
bitReader.skipBits(1);
break;
default:
throw NotImplementedException(); // TODO
}
break;
}
if (audioCfg->extensionAudioObjectType != Sbr && audioCfg->extensionAudioObjectType != Ps && bitReader.bitsAvailable() >= 16) {
uint16 syncExtensionType = bitReader.readBits<uint16>(11);
if (syncExtensionType == 0x2B7) {
if ((audioCfg->extensionAudioObjectType = getAudioObjectType()) == Sbr) {
if ((audioCfg->sbrPresent = bitReader.readBit())) {
if ((audioCfg->extensionSampleFrequencyIndex = bitReader.readBits<byte>(4)) == 0xF) {
audioCfg->extensionSampleFrequency = bitReader.readBits<uint32>(24);
}
if (bitReader.bitsAvailable() >= 12) {
if ((syncExtensionType = bitReader.readBits<uint16>(11)) == 0x548) {
audioCfg->psPresent = bitReader.readBits<byte>(1);
}
}
}
} else if (audioCfg->extensionAudioObjectType == ErBsac) {
if ((audioCfg->sbrPresent = bitReader.readBit())) {
if ((audioCfg->extensionSampleFrequencyIndex = bitReader.readBits<byte>(4)) == 0xF) {
audioCfg->extensionSampleFrequency = bitReader.readBits<uint32>(24);
}
}
audioCfg->extensionChannelConfiguration = bitReader.readBits<byte>(4);
}
} else if (syncExtensionType == 0x548) {
audioCfg->psPresent = bitReader.readBit();
}
}
} catch (const NotImplementedException &) {
diag.emplace_back(DiagLevel::Information, "Not implemented for the format of audio track.", context);
} catch (...) {
const char *what = catchIoFailure();
if (stream.fail()) {
// IO error caused by input stream
throwIoFailure(what);
} else {
// IO error caused by bitReader
diag.emplace_back(DiagLevel::Critical, "Audio specific configuration is truncated.", context);
}
}
return audioCfg;
}
/*!
* \brief Parses the video specific configuration for the track.
* \remarks
* - Notifications might be added.
* \sa mpeg4ElementaryStreamInfo()
*/
std::unique_ptr<Mpeg4VideoSpecificConfig> Mp4Track::parseVideoSpecificConfig(BinaryReader &reader, uint64 startOffset, uint64 size, Diagnostics &diag)
{
static const string context("parsing MPEG-4 video specific config from elementary stream descriptor");
using namespace Mpeg4AudioObjectIds;
auto videoCfg = make_unique<Mpeg4VideoSpecificConfig>();
// seek to start
reader.stream()->seekg(startOffset);
if (size > 3 && (reader.readUInt24BE() == 1)) {
size -= 3;
uint32 buff1;
while (size) {
--size;
switch (reader.readByte()) { // read start code
case Mpeg4VideoCodes::VisualObjectSequenceStart:
if (size) {
videoCfg->profile = reader.readByte();
--size;
}
break;
case Mpeg4VideoCodes::VideoObjectLayerStart:
break;
case Mpeg4VideoCodes::UserDataStart:
buff1 = 0;
while (size >= 3) {
if ((buff1 = reader.readUInt24BE()) != 1) {
reader.stream()->seekg(-2, ios_base::cur);
videoCfg->userData.push_back(buff1 >> 16);
--size;
} else {
size -= 3;
break;
}
}
if (buff1 != 1 && size > 0) {
videoCfg->userData += reader.readString(size);
size = 0;
}
break;
default:;
}
// skip remainging values to get the start of the next video object
while (size >= 3) {
if (reader.readUInt24BE() != 1) {
reader.stream()->seekg(-2, ios_base::cur);
--size;
} else {
size -= 3;
break;
}
}
}
} else {
diag.emplace_back(DiagLevel::Critical, "\"Visual Object Sequence Header\" not found.", context);
}
return videoCfg;
}
/*!
* \brief Updates the chunk offsets of the track. This is necessary when the "mdat"-atom
* (which contains the actual chunk data) is moved.
* \param oldMdatOffsets Specifies a vector holding the old offsets of the "mdat"-atoms.
* \param newMdatOffsets Specifies a vector holding the new offsets of the "mdat"-atoms.
*
* \throws Throws InvalidDataException when
* - there is no stream assigned.
* - the header has been considered as invalid when parsing the header information.
* - \a oldMdatOffsets holds not the same number of offsets as \a newMdatOffsets.
* - there is no atom holding these offsets.
* - the ID of the atom holding these offsets is not "stco" or "co64"
*
* \throws Throws std::ios_base::failure when an IO error occurs.
*
* \remarks This method needs to be fixed.
*/
void Mp4Track::updateChunkOffsets(const vector<int64> &oldMdatOffsets, const vector<int64> &newMdatOffsets)
{
if (!isHeaderValid() || !m_ostream || !m_istream || !m_stcoAtom) {
throw InvalidDataException();
}
if (oldMdatOffsets.size() == 0 || oldMdatOffsets.size() != newMdatOffsets.size()) {
throw InvalidDataException();
}
static const unsigned int stcoDataBegin = 8;
uint64 startPos = m_stcoAtom->dataOffset() + stcoDataBegin;
uint64 endPos = startPos + m_stcoAtom->dataSize() - stcoDataBegin;
m_istream->seekg(startPos);
m_ostream->seekp(startPos);
vector<int64>::size_type i;
vector<int64>::size_type size;
uint64 currentPos = m_istream->tellg();
switch (m_stcoAtom->id()) {
case Mp4AtomIds::ChunkOffset: {
uint32 off;
while ((currentPos + 4) <= endPos) {
off = m_reader.readUInt32BE();
for (i = 0, size = oldMdatOffsets.size(); i < size; ++i) {
if (off > static_cast<uint64>(oldMdatOffsets[i])) {
off += (newMdatOffsets[i] - oldMdatOffsets[i]);
break;
}
}
m_ostream->seekp(currentPos);
m_writer.writeUInt32BE(off);
currentPos += m_istream->gcount();
}
break;
}
case Mp4AtomIds::ChunkOffset64: {
uint64 off;
while ((currentPos + 8) <= endPos) {
off = m_reader.readUInt64BE();
for (i = 0, size = oldMdatOffsets.size(); i < size; ++i) {
if (off > static_cast<uint64>(oldMdatOffsets[i])) {
off += (newMdatOffsets[i] - oldMdatOffsets[i]);
break;
}
}
m_ostream->seekp(currentPos);
m_writer.writeUInt64BE(off);
currentPos += m_istream->gcount();
}
break;
}
default:
throw InvalidDataException();
}
}
/*!
* \brief Updates the chunk offsets of the track. This is necessary when the "mdat"-atom
* (which contains the actual chunk data) is moved.
* \param chunkOffsets Specifies the new chunk offset table.
*
* \throws Throws InvalidDataException when
* - there is no stream assigned.
* - the header has been considered as invalid when parsing the header information.
* - the size of \a chunkOffsets does not match chunkCount().
* - there is no atom holding these offsets.
* - the ID of the atom holding these offsets is not "stco" or "co64".
*/
void Mp4Track::updateChunkOffsets(const std::vector<uint64> &chunkOffsets)
{
if (!isHeaderValid() || !m_ostream || !m_istream || !m_stcoAtom) {
throw InvalidDataException();
}
if (chunkOffsets.size() != chunkCount()) {
throw InvalidDataException();
}
m_ostream->seekp(m_stcoAtom->dataOffset() + 8);
switch (m_stcoAtom->id()) {
case Mp4AtomIds::ChunkOffset:
for (auto offset : chunkOffsets) {
m_writer.writeUInt32BE(offset);
}
break;
case Mp4AtomIds::ChunkOffset64:
for (auto offset : chunkOffsets) {
m_writer.writeUInt64BE(offset);
}
default:
throw InvalidDataException();
}
}
/*!
* \brief Updates a particular chunk offset.
* \param chunkIndex Specifies the index of the chunk offset to be updated.
* \param offset Specifies the new chunk offset.
* \remarks This method seems to be obsolete.
* \throws Throws InvalidDataException when
* - there is no stream assigned.
* - the header has been considered as invalid when parsing the header information.
* - \a chunkIndex is not less than chunkCount().
* - there is no atom holding these offsets.
* - the ID of the atom holding these offsets is not "stco" or "co64".
*/
void Mp4Track::updateChunkOffset(uint32 chunkIndex, uint64 offset)
{
if (!isHeaderValid() || !m_istream || !m_stcoAtom || chunkIndex >= m_chunkCount) {
throw InvalidDataException();
}
m_ostream->seekp(m_stcoAtom->dataOffset() + 8 + chunkOffsetSize() * chunkIndex);
switch (chunkOffsetSize()) {
case 4:
writer().writeUInt32BE(offset);
break;
case 8:
writer().writeUInt64BE(offset);
break;
default:
throw InvalidDataException();
}
}
/*!
* \brief Adds the information from the specified \a avcConfig to the specified \a track.
*/
void Mp4Track::addInfo(const AvcConfiguration &avcConfig, AbstractTrack &track)
{
if (!avcConfig.spsInfos.empty()) {
const SpsInfo &spsInfo = avcConfig.spsInfos.back();
track.m_format.sub = spsInfo.profileIndication;
track.m_version = static_cast<double>(spsInfo.levelIndication) / 10;
track.m_cropping = spsInfo.cropping;
track.m_pixelSize = spsInfo.pictureSize;
switch (spsInfo.chromaFormatIndication) {
case 0:
track.m_chromaFormat = "monochrome";
break;
case 1:
track.m_chromaFormat = "YUV 4:2:0";
break;
case 2:
track.m_chromaFormat = "YUV 4:2:2";
break;
case 3:
track.m_chromaFormat = "YUV 4:4:4";
break;
default:;
}
track.m_pixelAspectRatio = spsInfo.pixelAspectRatio;
} else {
track.m_format.sub = avcConfig.profileIndication;
track.m_version = static_cast<double>(avcConfig.levelIndication) / 10;
}
}
/*!
* \brief Buffers all atoms required by the makeTrack() method.
*
* This allows to invoke makeTrack() also when the input stream is going to be
* modified (eg. to apply changed tags without rewriting the file).
*/
void Mp4Track::bufferTrackAtoms(Diagnostics &diag)
{
if (m_tkhdAtom) {
m_tkhdAtom->makeBuffer();
}
if (Mp4Atom *trefAtom = m_trakAtom->childById(Mp4AtomIds::TrackReference, diag)) {
trefAtom->makeBuffer();
}
if (Mp4Atom *edtsAtom = m_trakAtom->childById(Mp4AtomIds::Edit, diag)) {
edtsAtom->makeBuffer();
}
if (m_minfAtom) {
if (Mp4Atom *vmhdAtom = m_minfAtom->childById(Mp4AtomIds::VideoMediaHeader, diag)) {
vmhdAtom->makeBuffer();
}
if (Mp4Atom *smhdAtom = m_minfAtom->childById(Mp4AtomIds::SoundMediaHeader, diag)) {
smhdAtom->makeBuffer();
}
if (Mp4Atom *hmhdAtom = m_minfAtom->childById(Mp4AtomIds::HintMediaHeader, diag)) {
hmhdAtom->makeBuffer();
}
if (Mp4Atom *nmhdAtom = m_minfAtom->childById(Mp4AtomIds::NullMediaHeaderBox, diag)) {
nmhdAtom->makeBuffer();
}
if (Mp4Atom *dinfAtom = m_minfAtom->childById(Mp4AtomIds::DataInformation, diag)) {
dinfAtom->makeBuffer();
}
if (Mp4Atom *stblAtom = m_minfAtom->childById(Mp4AtomIds::SampleTable, diag)) {
stblAtom->makeBuffer();
}
}
}
/*!
* \brief Returns the number of bytes written when calling makeTrack().
*/
uint64 Mp4Track::requiredSize(Diagnostics &diag) const
{
// add size of
// ... trak header
uint64 size = 8;
// ... tkhd atom (TODO: buffer TrackHeaderInfo in v7)
size += verifyPresentTrackHeader().requiredSize;
// ... tref atom (if one exists)
if (Mp4Atom *trefAtom = m_trakAtom->childById(Mp4AtomIds::TrackReference, diag)) {
size += trefAtom->totalSize();
}
// ... edts atom (if one exists)
if (Mp4Atom *edtsAtom = m_trakAtom->childById(Mp4AtomIds::Edit, diag)) {
size += edtsAtom->totalSize();
}
// ... mdia header + mdhd total size + hdlr total size + minf header
size += 8 + 44 + (33 + m_name.size()) + 8;
// ... minf childs
bool dinfAtomWritten = false;
if (m_minfAtom) {
if (Mp4Atom *vmhdAtom = m_minfAtom->childById(Mp4AtomIds::VideoMediaHeader, diag)) {
size += vmhdAtom->totalSize();
}
if (Mp4Atom *smhdAtom = m_minfAtom->childById(Mp4AtomIds::SoundMediaHeader, diag)) {
size += smhdAtom->totalSize();
}
if (Mp4Atom *hmhdAtom = m_minfAtom->childById(Mp4AtomIds::HintMediaHeader, diag)) {
size += hmhdAtom->totalSize();
}
if (Mp4Atom *nmhdAtom = m_minfAtom->childById(Mp4AtomIds::NullMediaHeaderBox, diag)) {
size += nmhdAtom->totalSize();
}
if (Mp4Atom *dinfAtom = m_minfAtom->childById(Mp4AtomIds::DataInformation, diag)) {
size += dinfAtom->totalSize();
dinfAtomWritten = true;
}
if (Mp4Atom *stblAtom = m_minfAtom->childById(Mp4AtomIds::SampleTable, diag)) {
size += stblAtom->totalSize();
}
}
if (!dinfAtomWritten) {
size += 36;
}
return size;
}
/*!
* \brief Makes the track entry ("trak"-atom) for the track.
*
* The data is written to the assigned output stream at the current position. Note that this method
* uses the assigned input stream to copy some parts from the source file. Hence the input stream must
* still be valid when calling this method. To avoid this limitation call bufferTrackAtoms() before
* invalidating the input stream.
*/
void Mp4Track::makeTrack(Diagnostics &diag)
{
// write header
ostream::pos_type trakStartOffset = outputStream().tellp();
m_writer.writeUInt32BE(0); // write size later
m_writer.writeUInt32BE(Mp4AtomIds::Track);
// write tkhd atom
makeTrackHeader(diag);
// write tref atom (if one exists)
if (Mp4Atom *trefAtom = trakAtom().childById(Mp4AtomIds::TrackReference, diag)) {
trefAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
}
// write edts atom (if one exists)
if (Mp4Atom *edtsAtom = trakAtom().childById(Mp4AtomIds::Edit, diag)) {
edtsAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
}
// write mdia atom
makeMedia(diag);
// write size (of trak atom)
Mp4Atom::seekBackAndWriteAtomSize(outputStream(), trakStartOffset, diag);
}
/*!
* \brief Makes the track header (tkhd atom) for the track. The data is written to the assigned output stream
* at the current position.
*/
void Mp4Track::makeTrackHeader(Diagnostics &diag)
{
// verify the existing track header to make the new one based on it (if possible)
const TrackHeaderInfo info(verifyPresentTrackHeader());
// add notifications in case the present track header could not be parsed
if (info.versionUnknown) {
diag.emplace_back(DiagLevel::Critical,
argsToString("The version of the present \"tkhd\"-atom (", info.version, ") is unknown. Assuming version 1."),
argsToString("making \"tkhd\"-atom of track ", m_id));
}
if (info.truncated) {
diag.emplace_back(
DiagLevel::Critical, argsToString("The present \"tkhd\"-atom is truncated."), argsToString("making \"tkhd\"-atom of track ", m_id));
}
// make size and element ID
if (info.requiredSize > numeric_limits<uint32>::max()) {
writer().writeUInt32BE(1);
writer().writeUInt32BE(Mp4AtomIds::TrackHeader);
writer().writeUInt64BE(info.requiredSize);
} else {
writer().writeUInt32BE(static_cast<uint32>(info.requiredSize));
writer().writeUInt32BE(Mp4AtomIds::TrackHeader);
}
// make version and flags
writer().writeByte(1);
uint32 flags = 0;
if (m_enabled) {
flags |= 0x000001;
}
if (m_usedInPresentation) {
flags |= 0x000002;
}
if (m_usedWhenPreviewing) {
flags |= 0x000004;
}
writer().writeUInt24BE(flags);
// make creation and modification time
writer().writeUInt64BE(static_cast<uint64>((m_creationTime - startDate).totalSeconds()));
writer().writeUInt64BE(static_cast<uint64>((m_modificationTime - startDate).totalSeconds()));
// make track ID and duration
writer().writeUInt32BE(static_cast<uint32>(m_id));
writer().writeUInt32BE(0); // reserved
writer().writeUInt64BE(static_cast<uint64>(m_duration.totalSeconds() * m_timeScale));
writer().writeUInt32BE(0); // reserved
writer().writeUInt32BE(0); // reserved
// make further values, either from existing tkhd atom or just some defaults
if (info.canUseExisting) {
// write all bytes after the previously determined additionalDataOffset
m_ostream->write(m_tkhdAtom->buffer().get() + m_tkhdAtom->headerSize() + info.additionalDataOffset,
static_cast<streamoff>(m_tkhdAtom->dataSize() - info.additionalDataOffset));
// discard the buffer again if it wasn't present before
if (info.discardBuffer) {
m_tkhdAtom->discardBuffer();
}
} else {
// write default values
writer().writeInt16BE(0); // layer
writer().writeInt16BE(0); // alternate group
writer().writeFixed8BE(1.0); // volume (fixed 8.8 - 2 byte)
writer().writeUInt16BE(0); // reserved
for (const int32 value : { 0x00010000, 0, 0, 0, 0x00010000, 0, 0, 0, 0x40000000 }) { // unity matrix
writer().writeInt32BE(value);
}
writer().writeFixed16BE(1.0); // width
writer().writeFixed16BE(1.0); // height
}
}
/*!
* \brief Makes the media information (mdia atom) for the track. The data is written to the assigned output stream
* at the current position.
*/
void Mp4Track::makeMedia(Diagnostics &diag)
{
ostream::pos_type mdiaStartOffset = outputStream().tellp();
writer().writeUInt32BE(0); // write size later
writer().writeUInt32BE(Mp4AtomIds::Media);
// write mdhd atom
writer().writeUInt32BE(44); // size
writer().writeUInt32BE(Mp4AtomIds::MediaHeader);
writer().writeByte(1); // version
writer().writeUInt24BE(0); // flags
writer().writeUInt64BE(static_cast<uint64>((m_creationTime - startDate).totalSeconds()));
writer().writeUInt64BE(static_cast<uint64>((m_modificationTime - startDate).totalSeconds()));
writer().writeUInt32BE(m_timeScale);
writer().writeUInt64BE(static_cast<uint64>(m_duration.totalSeconds() * m_timeScale));
// convert and write language
uint16 language = 0;
for (size_t charIndex = 0; charIndex != 3; ++charIndex) {
const char langChar = charIndex < m_language.size() ? m_language[charIndex] : 0;
if (langChar >= 'a' && langChar <= 'z') {
language |= static_cast<uint16>(langChar - 0x60) << (0xA - charIndex * 0x5);
} else { // invalid character
if (!m_language.empty()) {
diag.emplace_back(
DiagLevel::Warning, "Assigned language \"" % m_language + "\" is of an invalid format and will be ignored.", "making mdhd atom");
}
language = 0x55C4; // und
break;
}
}
if (m_language.size() > 3) {
diag.emplace_back(
DiagLevel::Warning, "Assigned language \"" % m_language + "\" is longer than 3 byte and hence will be truncated.", "making mdhd atom");
}
writer().writeUInt16BE(language);
writer().writeUInt16BE(0); // pre defined
// write hdlr atom
writer().writeUInt32BE(33 + m_name.size()); // size
writer().writeUInt32BE(Mp4AtomIds::HandlerReference);
writer().writeUInt64BE(0); // version, flags, pre defined
switch (m_mediaType) {
case MediaType::Video:
outputStream().write("vide", 4);
break;
case MediaType::Audio:
outputStream().write("soun", 4);
break;
case MediaType::Hint:
outputStream().write("hint", 4);
break;
case MediaType::Text:
outputStream().write("meta", 4);
break;
default:
diag.emplace_back(DiagLevel::Critical, "Media type is invalid; The media type video is assumed.", "making hdlr atom");
outputStream().write("vide", 4);
break;
}
for (int i = 0; i < 3; ++i)
writer().writeUInt32BE(0); // reserved
writer().writeTerminatedString(m_name);
// write minf atom
makeMediaInfo(diag);
// write size (of mdia atom)
Mp4Atom::seekBackAndWriteAtomSize(outputStream(), mdiaStartOffset, diag);
}
/*!
* \brief Makes a media information (minf atom) for the track. The data is written to the assigned output stream
* at the current position.
*/
void Mp4Track::makeMediaInfo(Diagnostics &diag)
{
ostream::pos_type minfStartOffset = outputStream().tellp();
writer().writeUInt32BE(0); // write size later
writer().writeUInt32BE(Mp4AtomIds::MediaInformation);
bool dinfAtomWritten = false;
if (m_minfAtom) {
// copy existing vmhd atom
if (Mp4Atom *vmhdAtom = m_minfAtom->childById(Mp4AtomIds::VideoMediaHeader, diag)) {
vmhdAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
}
// copy existing smhd atom
if (Mp4Atom *smhdAtom = m_minfAtom->childById(Mp4AtomIds::SoundMediaHeader, diag)) {
smhdAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
}
// copy existing hmhd atom
if (Mp4Atom *hmhdAtom = m_minfAtom->childById(Mp4AtomIds::HintMediaHeader, diag)) {
hmhdAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
}
// copy existing nmhd atom
if (Mp4Atom *nmhdAtom = m_minfAtom->childById(Mp4AtomIds::NullMediaHeaderBox, diag)) {
nmhdAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
}
// copy existing dinf atom
if (Mp4Atom *dinfAtom = m_minfAtom->childById(Mp4AtomIds::DataInformation, diag)) {
dinfAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
dinfAtomWritten = true;
}
}
// write dinf atom if not written yet
if (!dinfAtomWritten) {
writer().writeUInt32BE(36); // size
writer().writeUInt32BE(Mp4AtomIds::DataInformation);
// write dref atom
writer().writeUInt32BE(28); // size
writer().writeUInt32BE(Mp4AtomIds::DataReference);
writer().writeUInt32BE(0); // version and flags
writer().writeUInt32BE(1); // entry count
// write url atom
writer().writeUInt32BE(12); // size
writer().writeUInt32BE(Mp4AtomIds::DataEntryUrl);
writer().writeByte(0); // version
writer().writeUInt24BE(0x000001); // flags (media data is in the same file as the movie box)
}
// write stbl atom
// -> just copy existing stbl atom because makeSampleTable() is not fully implemented (yet)
bool stblAtomWritten = false;
if (m_minfAtom) {
if (Mp4Atom *stblAtom = m_minfAtom->childById(Mp4AtomIds::SampleTable, diag)) {
stblAtom->copyPreferablyFromBuffer(outputStream(), diag, nullptr);
stblAtomWritten = true;
}
}
if (!stblAtomWritten) {
diag.emplace_back(DiagLevel::Critical,
"Source track does not contain mandatory stbl atom and the tagparser lib is unable to make one from scratch.", "making stbl atom");
}
// write size (of minf atom)
Mp4Atom::seekBackAndWriteAtomSize(outputStream(), minfStartOffset, diag);
}
/*!
* \brief Makes the sample table (stbl atom) for the track. The data is written to the assigned output stream
* at the current position.
* \remarks Not fully implemented yet.
*/
void Mp4Track::makeSampleTable(Diagnostics &diag)
{
ostream::pos_type stblStartOffset = outputStream().tellp();
writer().writeUInt32BE(0); // write size later
writer().writeUInt32BE(Mp4AtomIds::SampleTable);
Mp4Atom *stblAtom = m_minfAtom ? m_minfAtom->childById(Mp4AtomIds::SampleTable, diag) : nullptr;
// write stsd atom
if (m_stsdAtom) {
// copy existing stsd atom
m_stsdAtom->copyEntirely(outputStream(), diag, nullptr);
} else {
diag.emplace_back(DiagLevel::Critical, "Unable to make stsd atom from scratch.", "making stsd atom");
throw NotImplementedException();
}
// write stts and ctts atoms
Mp4Atom *sttsAtom = stblAtom ? stblAtom->childById(Mp4AtomIds::DecodingTimeToSample, diag) : nullptr;
if (sttsAtom) {
// copy existing stts atom
sttsAtom->copyEntirely(outputStream(), diag, nullptr);
} else {
diag.emplace_back(DiagLevel::Critical, "Unable to make stts atom from scratch.", "making stts atom");
throw NotImplementedException();
}
Mp4Atom *cttsAtom = stblAtom ? stblAtom->childById(Mp4AtomIds::CompositionTimeToSample, diag) : nullptr;
if (cttsAtom) {
// copy existing ctts atom
cttsAtom->copyEntirely(outputStream(), diag, nullptr);
}
// write stsc atom (sample-to-chunk table)
throw NotImplementedException();
// write stsz atom (sample sizes)
// write stz2 atom (compact sample sizes)
// write stco/co64 atom (chunk offset table)
// write stss atom (sync sample table)
// write stsh atom (shadow sync sample table)
// write padb atom (sample padding bits)
// write stdp atom (sample degradation priority)
// write sdtp atom (independent and disposable samples)
// write sbgp atom (sample group description)
// write sbgp atom (sample-to-group)
// write sgpd atom (sample group description)
// write subs atom (sub-sample information)
// write size (of stbl atom)
Mp4Atom::seekBackAndWriteAtomSize(outputStream(), stblStartOffset, diag);
}
void Mp4Track::internalParseHeader(Diagnostics &diag)
{
static const string context("parsing MP4 track");
using namespace Mp4AtomIds;
if (!m_trakAtom) {
diag.emplace_back(DiagLevel::Critical, "\"trak\"-atom is null.", context);
throw InvalidDataException();
}
// get atoms
try {
if (!(m_tkhdAtom = m_trakAtom->childById(TrackHeader, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"tkhd\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_mdiaAtom = m_trakAtom->childById(Media, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"mdia\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_mdhdAtom = m_mdiaAtom->childById(MediaHeader, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"mdhd\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_hdlrAtom = m_mdiaAtom->childById(HandlerReference, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"hdlr\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_minfAtom = m_mdiaAtom->childById(MediaInformation, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"minf\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_stblAtom = m_minfAtom->childById(SampleTable, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"stbl\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_stsdAtom = m_stblAtom->childById(SampleDescription, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"stsd\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_stcoAtom = m_stblAtom->childById(ChunkOffset, diag)) && !(m_stcoAtom = m_stblAtom->childById(ChunkOffset64, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"stco\"/\"co64\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_stscAtom = m_stblAtom->childById(SampleToChunk, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"stsc\"-atom found.", context);
throw InvalidDataException();
}
if (!(m_stszAtom = m_stblAtom->childById(SampleSize, diag)) && !(m_stszAtom = m_stblAtom->childById(CompactSampleSize, diag))) {
diag.emplace_back(DiagLevel::Critical, "No \"stsz\"/\"stz2\"-atom found.", context);
throw InvalidDataException();
}
} catch (const Failure &) {
diag.emplace_back(DiagLevel::Critical, "Unable to parse relevant atoms.", context);
throw InvalidDataException();
}
BinaryReader &reader = m_trakAtom->reader();
// read tkhd atom
m_istream->seekg(m_tkhdAtom->startOffset() + 8); // seek to beg, skip size and name
byte atomVersion = reader.readByte(); // read version
uint32 flags = reader.readUInt24BE();
m_enabled = flags & 0x000001;
m_usedInPresentation = flags & 0x000002;
m_usedWhenPreviewing = flags & 0x000004;
switch (atomVersion) {
case 0:
m_creationTime = startDate + TimeSpan::fromSeconds(reader.readUInt32BE());
m_modificationTime = startDate + TimeSpan::fromSeconds(reader.readUInt32BE());
m_id = reader.readUInt32BE();
break;
case 1:
m_creationTime = startDate + TimeSpan::fromSeconds(reader.readUInt64BE());
m_modificationTime = startDate + TimeSpan::fromSeconds(reader.readUInt64BE());
m_id = reader.readUInt32BE();
break;
default:
diag.emplace_back(DiagLevel::Critical,
"Version of \"tkhd\"-atom not supported. It will be ignored. Track ID, creation time and modification time might not be be determined.",
context);
m_creationTime = DateTime();
m_modificationTime = DateTime();
m_id = 0;
}
// read mdhd atom
m_istream->seekg(m_mdhdAtom->dataOffset()); // seek to beg, skip size and name
atomVersion = reader.readByte(); // read version
m_istream->seekg(3, ios_base::cur); // skip flags
switch (atomVersion) {
case 0:
m_creationTime = startDate + TimeSpan::fromSeconds(reader.readUInt32BE());
m_modificationTime = startDate + TimeSpan::fromSeconds(reader.readUInt32BE());
m_timeScale = reader.readUInt32BE();
m_duration = TimeSpan::fromSeconds(static_cast<double>(reader.readUInt32BE()) / static_cast<double>(m_timeScale));
break;
case 1:
m_creationTime = startDate + TimeSpan::fromSeconds(reader.readUInt64BE());
m_modificationTime = startDate + TimeSpan::fromSeconds(reader.readUInt64BE());
m_timeScale = reader.readUInt32BE();
m_duration = TimeSpan::fromSeconds(static_cast<double>(reader.readUInt64BE()) / static_cast<double>(m_timeScale));
break;
default:
diag.emplace_back(DiagLevel::Warning,
"Version of \"mdhd\"-atom not supported. It will be ignored. Creation time, modification time, time scale and duration might not be "
"determined.",
context);
m_timeScale = 0;
m_duration = TimeSpan();
}
uint16 tmp = reader.readUInt16BE();
if (tmp) {
const char buff[] = {
static_cast<char>(((tmp & 0x7C00) >> 0xA) + 0x60),
static_cast<char>(((tmp & 0x03E0) >> 0x5) + 0x60),
static_cast<char>(((tmp & 0x001F) >> 0x0) + 0x60),
};
m_language = string(buff, 3);
} else {
m_language.clear();
}
// read hdlr atom
// -> seek to begin skipping size, name, version, flags and reserved bytes
m_istream->seekg(m_hdlrAtom->dataOffset() + 8);
// -> track type
switch (reader.readUInt32BE()) {
case 0x76696465:
m_mediaType = MediaType::Video;
break;
case 0x736F756E:
m_mediaType = MediaType::Audio;
break;
case 0x68696E74:
m_mediaType = MediaType::Hint;
break;
case 0x6D657461:
case 0x74657874:
m_mediaType = MediaType::Text;
break;
default:
m_mediaType = MediaType::Unknown;
}
// -> name
m_istream->seekg(12, ios_base::cur); // skip reserved bytes
if ((tmp = m_istream->peek()) == m_hdlrAtom->dataSize() - 12 - 4 - 8 - 1) {
// assume size prefixed string (seems to appear in QuickTime files)
m_istream->seekg(1, ios_base::cur);
m_name = reader.readString(tmp);
} else {
// assume null terminated string (appears in MP4 files)
m_name = reader.readTerminatedString(m_hdlrAtom->dataSize() - 12 - 4 - 8, 0);
}
// read stco atom (only chunk count)
m_chunkOffsetSize = (m_stcoAtom->id() == Mp4AtomIds::ChunkOffset64) ? 8 : 4;
m_istream->seekg(m_stcoAtom->dataOffset() + 4);
m_chunkCount = reader.readUInt32BE();
// read stsd atom
m_istream->seekg(m_stsdAtom->dataOffset() + 4); // seek to beg, skip size, name, version and flags
uint32 entryCount = reader.readUInt32BE();
Mp4Atom *esDescParentAtom = nullptr;
if (entryCount) {
try {
for (Mp4Atom *codecConfigContainerAtom = m_stsdAtom->firstChild(); codecConfigContainerAtom;
codecConfigContainerAtom = codecConfigContainerAtom->nextSibling()) {
codecConfigContainerAtom->parse(diag);
// parse FOURCC
m_formatId = interpretIntegerAsString<uint32>(codecConfigContainerAtom->id());
m_format = FourccIds::fourccToMediaFormat(codecConfigContainerAtom->id());
// parse codecConfigContainerAtom
m_istream->seekg(codecConfigContainerAtom->dataOffset());
switch (codecConfigContainerAtom->id()) {
case FourccIds::Mpeg4Audio:
case FourccIds::AmrNarrowband:
case FourccIds::Amr:
case FourccIds::Drms:
case FourccIds::Alac:
case FourccIds::WindowsMediaAudio:
case FourccIds::Ac3:
case FourccIds::EAc3:
case FourccIds::DolbyMpl:
case FourccIds::Dts:
case FourccIds::DtsH:
case FourccIds::DtsE:
m_istream->seekg(6 + 2, ios_base::cur); // skip reserved bytes, data reference index
tmp = reader.readUInt16BE(); // read sound version
m_istream->seekg(6, ios_base::cur);
m_channelCount = reader.readUInt16BE();
m_bitsPerSample = reader.readUInt16BE();
m_istream->seekg(4, ios_base::cur); // skip reserved bytes (again)
if (!m_samplingFrequency) {
m_samplingFrequency = reader.readUInt32BE() >> 16;
if (codecConfigContainerAtom->id() != FourccIds::DolbyMpl) {
m_samplingFrequency >>= 16;
}
} else {
m_istream->seekg(4, ios_base::cur);
}
if (codecConfigContainerAtom->id() != FourccIds::WindowsMediaAudio) {
switch (tmp) {
case 1:
codecConfigContainerAtom->denoteFirstChild(codecConfigContainerAtom->headerSize() + 28 + 16);
break;
case 2:
codecConfigContainerAtom->denoteFirstChild(codecConfigContainerAtom->headerSize() + 28 + 32);
break;
default:
codecConfigContainerAtom->denoteFirstChild(codecConfigContainerAtom->headerSize() + 28);
}
if (!esDescParentAtom) {
esDescParentAtom = codecConfigContainerAtom;
}
}
break;
case FourccIds::Mpeg4Video:
case FourccIds::H263Quicktime:
case FourccIds::H2633GPP:
case FourccIds::Avc1:
case FourccIds::Avc2:
case FourccIds::Avc3:
case FourccIds::Avc4:
case FourccIds::Drmi:
case FourccIds::Hevc1:
case FourccIds::Hevc2:
m_istream->seekg(6 + 2 + 16, ios_base::cur); // skip reserved bytes, data reference index, and reserved bytes (again)
m_pixelSize.setWidth(reader.readUInt16BE());
m_pixelSize.setHeight(reader.readUInt16BE());
m_resolution.setWidth(static_cast<uint32>(reader.readFixed16BE()));
m_resolution.setHeight(static_cast<uint32>(reader.readFixed16BE()));
m_istream->seekg(4, ios_base::cur); // skip reserved bytes
m_framesPerSample = reader.readUInt16BE();
tmp = reader.readByte();
m_compressorName = reader.readString(31);
if (tmp == 0) {
m_compressorName.clear();
} else if (tmp < 32) {
m_compressorName.resize(tmp);
}
m_depth = reader.readUInt16BE(); // 24: color without alpha
codecConfigContainerAtom->denoteFirstChild(codecConfigContainerAtom->headerSize() + 78);
if (!esDescParentAtom) {
esDescParentAtom = codecConfigContainerAtom;
}
break;
case FourccIds::Mpeg4Sample:
// skip reserved bytes and data reference index
codecConfigContainerAtom->denoteFirstChild(codecConfigContainerAtom->headerSize() + 8);
if (!esDescParentAtom) {
esDescParentAtom = codecConfigContainerAtom;
}
break;
case Mp4AtomIds::PixalAspectRatio:
break; // TODO
case Mp4AtomIds::CleanAperature:
break; // TODO
default:;
}
}
if (esDescParentAtom) {
// parse AVC configuration
if (Mp4Atom *avcConfigAtom = esDescParentAtom->childById(Mp4AtomIds::AvcConfiguration, diag)) {
m_istream->seekg(avcConfigAtom->dataOffset());
m_avcConfig = make_unique<TagParser::AvcConfiguration>();
try {
m_avcConfig->parse(reader, avcConfigAtom->dataSize());
addInfo(*m_avcConfig, *this);
} catch (const TruncatedDataException &) {
diag.emplace_back(DiagLevel::Critical, "AVC configuration is truncated.", context);
} catch (const Failure &) {
diag.emplace_back(DiagLevel::Critical, "AVC configuration is invalid.", context);
}
}
// parse MPEG-4 elementary stream descriptor
Mp4Atom *esDescAtom = esDescParentAtom->childById(Mp4FormatExtensionIds::Mpeg4ElementaryStreamDescriptor, diag);
if (!esDescAtom) {
esDescAtom = esDescParentAtom->childById(Mp4FormatExtensionIds::Mpeg4ElementaryStreamDescriptor2, diag);
}
if (esDescAtom) {
try {
if ((m_esInfo = parseMpeg4ElementaryStreamInfo(m_reader, esDescAtom, diag))) {
m_format += Mpeg4ElementaryStreamObjectIds::streamObjectTypeFormat(m_esInfo->objectTypeId);
m_bitrate = static_cast<double>(m_esInfo->averageBitrate) / 1000;
m_maxBitrate = static_cast<double>(m_esInfo->maxBitrate) / 1000;
if (m_esInfo->audioSpecificConfig) {
// check the audio specific config for useful information
m_format += Mpeg4AudioObjectIds::idToMediaFormat(m_esInfo->audioSpecificConfig->audioObjectType,
m_esInfo->audioSpecificConfig->sbrPresent, m_esInfo->audioSpecificConfig->psPresent);
if (m_esInfo->audioSpecificConfig->sampleFrequencyIndex == 0xF) {
m_samplingFrequency = m_esInfo->audioSpecificConfig->sampleFrequency;
} else if (m_esInfo->audioSpecificConfig->sampleFrequencyIndex < sizeof(mpeg4SamplingFrequencyTable)) {
m_samplingFrequency = mpeg4SamplingFrequencyTable[m_esInfo->audioSpecificConfig->sampleFrequencyIndex];
} else {
diag.emplace_back(DiagLevel::Warning, "Audio specific config has invalid sample frequency index.", context);
}
if (m_esInfo->audioSpecificConfig->extensionSampleFrequencyIndex == 0xF) {
m_extensionSamplingFrequency = m_esInfo->audioSpecificConfig->extensionSampleFrequency;
} else if (m_esInfo->audioSpecificConfig->extensionSampleFrequencyIndex < sizeof(mpeg4SamplingFrequencyTable)) {
m_extensionSamplingFrequency
= mpeg4SamplingFrequencyTable[m_esInfo->audioSpecificConfig->extensionSampleFrequencyIndex];
} else {
diag.emplace_back(
DiagLevel::Warning, "Audio specific config has invalid extension sample frequency index.", context);
}
m_channelConfig = m_esInfo->audioSpecificConfig->channelConfiguration;
m_extensionChannelConfig = m_esInfo->audioSpecificConfig->extensionChannelConfiguration;
}
if (m_esInfo->videoSpecificConfig) {
// check the video specific config for useful information
if (m_format.general == GeneralMediaFormat::Mpeg4Video && m_esInfo->videoSpecificConfig->profile) {
m_format.sub = m_esInfo->videoSpecificConfig->profile;
if (!m_esInfo->videoSpecificConfig->userData.empty()) {
m_formatId += " / ";
m_formatId += m_esInfo->videoSpecificConfig->userData;
}
}
}
// check the stream data for missing information
switch (m_format.general) {
case GeneralMediaFormat::Mpeg1Audio:
case GeneralMediaFormat::Mpeg2Audio: {
MpegAudioFrame frame;
m_istream->seekg(m_stcoAtom->dataOffset() + 8);
m_istream->seekg(m_chunkOffsetSize == 8 ? reader.readUInt64BE() : reader.readUInt32BE());
frame.parseHeader(reader);
MpegAudioFrameStream::addInfo(frame, *this);
break;
}
default:;
}
}
} catch (const Failure &) {
}
}
}
} catch (const Failure &) {
diag.emplace_back(DiagLevel::Critical, "Unable to parse child atoms of \"stsd\"-atom.", context);
}
}
// read stsz atom which holds the sample size table
m_sampleSizes.clear();
m_size = m_sampleCount = 0;
uint64 actualSampleSizeTableSize = m_stszAtom->dataSize();
if (actualSampleSizeTableSize < 12) {
diag.emplace_back(DiagLevel::Critical,
"The stsz atom is truncated. There are no sample sizes present. The size of the track can not be determined.", context);
} else {
actualSampleSizeTableSize -= 12; // subtract size of version and flags
m_istream->seekg(m_stszAtom->dataOffset() + 4); // seek to beg, skip size, name, version and flags
uint32 fieldSize;
uint32 constantSize;
if (m_stszAtom->id() == Mp4AtomIds::CompactSampleSize) {
constantSize = 0;
m_istream->seekg(3, ios_base::cur); // seek reserved bytes
fieldSize = reader.readByte();
m_sampleCount = reader.readUInt32BE();
} else {
constantSize = reader.readUInt32BE();
m_sampleCount = reader.readUInt32BE();
fieldSize = 32;
}
if (constantSize) {
m_sampleSizes.push_back(constantSize);
m_size = constantSize * m_sampleCount;
} else {
uint64 actualSampleCount = m_sampleCount;
uint64 calculatedSampleSizeTableSize = ceil((0.125 * fieldSize) * m_sampleCount);
if (calculatedSampleSizeTableSize < actualSampleSizeTableSize) {
diag.emplace_back(
DiagLevel::Critical, "The stsz atom stores more entries as denoted. The additional entries will be ignored.", context);
} else if (calculatedSampleSizeTableSize > actualSampleSizeTableSize) {
diag.emplace_back(DiagLevel::Critical, "The stsz atom is truncated. It stores less entries as denoted.", context);
actualSampleCount = floor(static_cast<double>(actualSampleSizeTableSize) / (0.125 * fieldSize));
}
m_sampleSizes.reserve(actualSampleCount);
uint32 i = 1;
switch (fieldSize) {
case 4:
for (; i <= actualSampleCount; i += 2) {
byte val = reader.readByte();
m_sampleSizes.push_back(val >> 4);
m_sampleSizes.push_back(val & 0xF0);
m_size += (val >> 4) + (val & 0xF0);
}
if (i <= actualSampleCount + 1) {
m_sampleSizes.push_back(reader.readByte() >> 4);
m_size += m_sampleSizes.back();
}
break;
case 8:
for (; i <= actualSampleCount; ++i) {
m_sampleSizes.push_back(reader.readByte());
m_size += m_sampleSizes.back();
}
break;
case 16:
for (; i <= actualSampleCount; ++i) {
m_sampleSizes.push_back(reader.readUInt16BE());
m_size += m_sampleSizes.back();
}
break;
case 32:
for (; i <= actualSampleCount; ++i) {
m_sampleSizes.push_back(reader.readUInt32BE());
m_size += m_sampleSizes.back();
}
break;
default:
diag.emplace_back(DiagLevel::Critical,
"The fieldsize used to store the sample sizes is not supported. The sample count and size of the track can not be determined.",
context);
}
}
}
// no sample sizes found, search for trun atoms
uint64 totalDuration = 0;
for (Mp4Atom *moofAtom = m_trakAtom->container().firstElement()->siblingByIdIncludingThis(MovieFragment, diag); moofAtom;
moofAtom = moofAtom->siblingById(MovieFragment, diag)) {
moofAtom->parse(diag);
for (Mp4Atom *trafAtom = moofAtom->childById(TrackFragment, diag); trafAtom; trafAtom = trafAtom->siblingById(TrackFragment, diag)) {
trafAtom->parse(diag);
for (Mp4Atom *tfhdAtom = trafAtom->childById(TrackFragmentHeader, diag); tfhdAtom;
tfhdAtom = tfhdAtom->siblingById(TrackFragmentHeader, diag)) {
tfhdAtom->parse(diag);
uint32 calculatedDataSize = 0;
if (tfhdAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "tfhd atom is truncated.", context);
} else {
m_istream->seekg(tfhdAtom->dataOffset() + 1);
uint32 flags = reader.readUInt24BE();
if (m_id == reader.readUInt32BE()) { // check track ID
if (flags & 0x000001) { // base-data-offset present
calculatedDataSize += 8;
}
if (flags & 0x000002) { // sample-description-index present
calculatedDataSize += 4;
}
if (flags & 0x000008) { // default-sample-duration present
calculatedDataSize += 4;
}
if (flags & 0x000010) { // default-sample-size present
calculatedDataSize += 4;
}
if (flags & 0x000020) { // default-sample-flags present
calculatedDataSize += 4;
}
//uint64 baseDataOffset = moofAtom->startOffset();
//uint32 defaultSampleDescriptionIndex = 0;
uint32 defaultSampleDuration = 0;
uint32 defaultSampleSize = 0;
//uint32 defaultSampleFlags = 0;
if (tfhdAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "tfhd atom is truncated (presence of fields denoted).", context);
} else {
if (flags & 0x000001) { // base-data-offset present
//baseDataOffset = reader.readUInt64();
m_istream->seekg(8, ios_base::cur);
}
if (flags & 0x000002) { // sample-description-index present
//defaultSampleDescriptionIndex = reader.readUInt32();
m_istream->seekg(4, ios_base::cur);
}
if (flags & 0x000008) { // default-sample-duration present
defaultSampleDuration = reader.readUInt32BE();
//m_istream->seekg(4, ios_base::cur);
}
if (flags & 0x000010) { // default-sample-size present
defaultSampleSize = reader.readUInt32BE();
}
if (flags & 0x000020) { // default-sample-flags present
//defaultSampleFlags = reader.readUInt32BE();
m_istream->seekg(4, ios_base::cur);
}
}
for (Mp4Atom *trunAtom = trafAtom->childById(TrackFragmentRun, diag); trunAtom;
trunAtom = trunAtom->siblingById(TrackFragmentRun, diag)) {
uint32 calculatedDataSize = 8;
if (trunAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "trun atom is truncated.", context);
} else {
m_istream->seekg(trunAtom->dataOffset() + 1);
uint32 flags = reader.readUInt24BE();
uint32 sampleCount = reader.readUInt32BE();
m_sampleCount += sampleCount;
if (flags & 0x000001) { // data offset present
calculatedDataSize += 4;
}
if (flags & 0x000004) { // first-sample-flags present
calculatedDataSize += 4;
}
uint32 entrySize = 0;
if (flags & 0x000100) { // sample-duration present
entrySize += 4;
}
if (flags & 0x000200) { // sample-size present
entrySize += 4;
}
if (flags & 0x000400) { // sample-flags present
entrySize += 4;
}
if (flags & 0x000800) { // sample-composition-time-offsets present
entrySize += 4;
}
calculatedDataSize += entrySize * sampleCount;
if (trunAtom->dataSize() < calculatedDataSize) {
diag.emplace_back(DiagLevel::Critical, "trun atom is truncated (presence of fields denoted).", context);
} else {
if (flags & 0x000001) { // data offset present
m_istream->seekg(4, ios_base::cur);
//int32 dataOffset = reader.readInt32();
}
if (flags & 0x000004) { // first-sample-flags present
m_istream->seekg(4, ios_base::cur);
}
for (uint32 i = 0; i < sampleCount; ++i) {
if (flags & 0x000100) { // sample-duration present
totalDuration += reader.readUInt32BE();
} else {
totalDuration += defaultSampleDuration;
}
if (flags & 0x000200) { // sample-size present
m_sampleSizes.push_back(reader.readUInt32BE());
m_size += m_sampleSizes.back();
} else {
m_size += defaultSampleSize;
}
if (flags & 0x000400) { // sample-flags present
m_istream->seekg(4, ios_base::cur);
}
if (flags & 0x000800) { // sample-composition-time-offsets present
m_istream->seekg(4, ios_base::cur);
}
}
}
}
}
if (m_sampleSizes.empty() && defaultSampleSize) {
m_sampleSizes.push_back(defaultSampleSize);
}
}
}
}
}
}
// set duration from "trun-information" if the duration has not been determined yet
if (m_duration.isNull() && totalDuration) {
uint32 timeScale = m_timeScale;
if (!timeScale) {
timeScale = trakAtom().container().timeScale();
}
if (timeScale) {
m_duration = TimeSpan::fromSeconds(static_cast<double>(totalDuration) / static_cast<double>(timeScale));
}
}
// caluculate average bitrate
if (m_bitrate < 0.01 && m_bitrate > -0.01) {
m_bitrate = (static_cast<double>(m_size) * 0.0078125) / m_duration.totalSeconds();
}
// read stsc atom (only number of entries)
m_istream->seekg(m_stscAtom->dataOffset() + 4);
m_sampleToChunkEntryCount = reader.readUInt32BE();
}
} // namespace TagParser