WrittenArraySampleIDPtr
WriteStringArrayT( WrittenArraySampleMap &iMap,
hid_t iGroup,
const std::string &iName,
const AbcA::ArraySample &iSamp,
const AbcA::ArraySample::Key &iKey,
int iCompressionLevel )
{
// because strings are packed together, always write out the dimensions
Dimensions dims = iSamp.getDimensions();
ABCA_ASSERT( dims.rank() > 0,
"String type can not have a rank-0 array sample" );
std::string dimsName = iName + ".dims";
WriteDimensions( iGroup, dimsName, dims );
// See whether or not we've already stored this.
WrittenArraySampleIDPtr writeID = iMap.find( iKey );
if ( writeID )
{
CopyWrittenArray( iGroup, iName, writeID );
return writeID;
}
// Okay, need to actually store it.
// It will be a dataset with an internal attribute for storing
// the hash id.
bool hasData = dims.numPoints() > 0;
hid_t dspaceId = -1;
Dimensions wdims; // Used to store the "fake" dimensions.
std::vector<CharT> charBuffer;
// Get the dimensions, validate sample size.
if ( hasData )
{
size_t extent = iSamp.getDataType().getExtent();
size_t numStrings = dims.numPoints() * extent;
ABCA_ASSERT( dims.rank() > 0 && numStrings > 0,
"Degenerate array sample in WriteStringArrayT" );
// Get the data out of the array sample.
const StringT *strings =
reinterpret_cast<const StringT *>( iSamp.getData() );
ABCA_ASSERT( strings,
"Degenerate strings in WriteStringArrayT" );
// Compact the strings in the string array.
CompactStrings( strings, numStrings, charBuffer );
// Create the dataspace.
size_t len = charBuffer.size();
assert( len >= numStrings );
wdims = Dimensions( len );
HDimensions hdims( wdims );
dspaceId = H5Screate_simple( hdims.rank(),
hdims.rootPtr(), NULL );
}
else
{
dspaceId = H5Screate( H5S_NULL );
}
ABCA_ASSERT( dspaceId >= 0,
"WriteStringsT() Failed in dataspace constructor" );
DspaceCloser dspaceCloser( dspaceId );
hid_t dsetId = -1;
if ( iCompressionLevel >= 0 && hasData )
{
// Make a compression plist
hid_t zipPlist = DsetGzipCreatePlist( wdims,
iCompressionLevel > 9 ? 9 : iCompressionLevel );
PlistCloser plistCloser( zipPlist );
//std::cout << "Creating compressed data set named: "
// << iName << " in group named: " << iGroup.name()
// << std::endl;
// Make the dataset.
dsetId = H5Dcreate2( iGroup, iName.c_str(), GetFileDtype<CharT>(),
dspaceId, H5P_DEFAULT, zipPlist, H5P_DEFAULT );
}
else
{
//std::cout << "Creating uncompressed data set named: "
// << iName << " in group named: " << iGroup.name()
// << std::endl;
dsetId = H5Dcreate2( iGroup, iName.c_str(), GetFileDtype<CharT>(),
dspaceId, H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT );
}
DsetCloser dsetCloser(dsetId);
ABCA_ASSERT( dsetId >= 0,
"WriteArray() Failed in dataset constructor" );
// Write the data.
if ( hasData )
{
H5Dwrite( dsetId, GetNativeDtype<CharT>(), H5S_ALL,
H5S_ALL, H5P_DEFAULT, &charBuffer.front() );
}
// Write the key
WriteKey( dsetId, "key", iKey );
writeID.reset( new WrittenArraySampleID( iKey, dsetId ) );
iMap.store( writeID );
// Return the reference.
return writeID;
}
static AbcA::ArraySamplePtr
ReadStringArrayT( AbcA::ReadArraySampleCachePtr iCache,
hid_t iParent,
const std::string &iName,
const AbcA::DataType &iDataType )
{
assert( iDataType.getExtent() > 0 );
// Open the data set.
hid_t dsetId = H5Dopen( iParent, iName.c_str(), H5P_DEFAULT );
ABCA_ASSERT( dsetId >= 0, "Cannot open dataset: " << iName );
DsetCloser dsetCloser( dsetId );
// Read the digest, if there's a cache.
AbcA::ArraySample::Key key;
bool foundDigest = ReadKey( dsetId, "key", key );
// If we found a digest and there's a cache, see
// if we're in there, and return it if so.
if ( foundDigest && iCache )
{
AbcA::ReadArraySampleID found = iCache->find( key );
if ( found )
{
AbcA::ArraySamplePtr ret = found.getSample();
assert( ret );
if ( ret->getDataType() != iDataType )
{
ABCA_THROW( "ERROR: Read data type for dset: " << iName
<< ": " << ret->getDataType()
<< " does not match expected data type: "
<< iDataType );
}
// Got it!
return ret;
}
}
// Okay, we haven't found it in a cache.
// Read the data type.
// Checking code.
{
hid_t dsetFtype = H5Dget_type( dsetId );
DtypeCloser dtypeCloser( dsetFtype );
hid_t nativeDtype = GetNativeDtype<CharT>();
ABCA_ASSERT( H5Tget_class( dsetFtype ) ==
H5Tget_class( nativeDtype ) &&
H5Tget_sign( dsetFtype ) ==
H5Tget_sign( nativeDtype )
// CJH They can now be different
// sizes, because wchar_t is sometimes 16-bit,
// but we always store 32 bit.
// && H5Tget_size( dsetFtype ) ==
//H5Tget_size( nativeDtype ),
, "Invalid datatype for stringT" );
}
// String array datatypes require a "dimensions" to be stored
// externally, since the strings themselves are stored in a compacted
// array of rank 1.
// This is an attribute called "dims" that lives in the dset itself.
Dimensions realDims;
ReadDimensions( dsetId, "dims", realDims );
ABCA_ASSERT( realDims.rank() > 0,
"Degenerate rank in Dataset read" );
// Read the data space.
hid_t dspaceId = H5Dget_space( dsetId );
ABCA_ASSERT( dspaceId >= 0, "Could not get dataspace for dataSet: "
<< iName );
DspaceCloser dspaceCloser( dspaceId );
AbcA::ArraySamplePtr ret;
H5S_class_t dspaceClass = H5Sget_simple_extent_type( dspaceId );
if ( dspaceClass == H5S_SIMPLE )
{
ABCA_ASSERT( realDims.numPoints() > 0,
"Degenerate dims in Dataset read" );
size_t totalNumStrings = realDims.numPoints() * iDataType.getExtent();
// Get the dimensions
Dimensions dims;
int rank = H5Sget_simple_extent_ndims( dspaceId );
ABCA_ASSERT( rank == realDims.rank(),
"H5Sget_simple_extent_ndims() failed." );
HDimensions hdims;
hdims.setRank( rank );
rank = H5Sget_simple_extent_dims( dspaceId, hdims.rootPtr(), NULL );
ABCA_ASSERT( rank == hdims.rank(),
"H5Sget_simple_extent_dims() "
"found inconsistent ranks."
<< std::endl
<< "Expecting rank: " << hdims.rank()
<< " instead was: " << rank );
dims = hdims;
ABCA_ASSERT( dims.numPoints() > 0,
"Degenerate dims in Dataset read" );
// Create temporary char storage buffer.
size_t totalNumChars = dims.numPoints() + 1;
std::vector<CharT> charStorage( totalNumChars, ( CharT )0 );
// Read into it.
herr_t status = H5Dread( dsetId, GetNativeDtype<CharT>(),
H5S_ALL, H5S_ALL, H5P_DEFAULT,
( void * )&charStorage.front() );
ABCA_ASSERT( status >= 0,
"Could not read string array from data set. Weird." );
// Make an appropriately dimensionalized (and manageable)
// array of strings using the ArraySamples.
ret = AbcA::AllocateArraySample( iDataType,
realDims );
StringT *strings = reinterpret_cast<StringT*>(
const_cast<void*>( ret->getData() ) );
assert( strings != NULL );
// This part is hard. We have to go through the one dimensional
// array extracting each string.
ExtractStrings<StringT,CharT>( strings,
( const CharT * )&charStorage.front(),
totalNumChars,
totalNumStrings );
}
else if ( dspaceClass == H5S_NULL )
{
// Num points should be zero here.
ABCA_ASSERT( realDims.numPoints() == 0,
"Expecting zero points in dimensions" );
ret = AbcA::AllocateArraySample( iDataType, realDims );
}
else
{
ABCA_THROW( "Unexpected scalar dataspace encountered." );
}
// Store if there is a cache.
if ( foundDigest && iCache )
{
AbcA::ReadArraySampleID stored = iCache->store( key, ret );
if ( stored )
{
return stored.getSample();
}
}
// Otherwise, just leave! ArraySamplePtr returned by AllocateArraySample
// already has fancy-dan deleter built in.
// I REALLY LOVE SMART PTRS.
return ret;
}
//-*****************************************************************************
WrittenArraySampleIDPtr
WriteArray( WrittenArraySampleMap &iMap,
hid_t iGroup,
const std::string &iName,
const AbcA::ArraySample &iSamp,
const AbcA::ArraySample::Key &iKey,
hid_t iFileType,
hid_t iNativeType,
int iCompressionLevel )
{
// Dispatch to string writing utils.
const AbcA::DataType &dataType = iSamp.getDataType();
if ( dataType.getPod() == kStringPOD )
{
return WriteStringArray( iMap, iGroup, iName, iSamp, iKey,
iCompressionLevel );
}
else if ( dataType.getPod() == kWstringPOD )
{
return WriteWstringArray( iMap, iGroup, iName, iSamp, iKey,
iCompressionLevel );
}
// write the dimensions as necessary
Dimensions dims = iSamp.getDimensions();
size_t rank = dims.rank();
ABCA_ASSERT( rank > 0, "Cannot have a rank-0 array sample" );
// rank 1 is the most common case, and we can easily infer it's size
// from the dataspace for non-strings, so don't bother writing it out
if (rank > 1)
{
std::string dimsName = iName + ".dims";
WriteDimensions( iGroup, dimsName, dims );
}
// See whether or not we've already stored this.
WrittenArraySampleIDPtr writeID = iMap.find( iKey );
if ( writeID )
{
CopyWrittenArray( iGroup, iName, writeID );
return writeID;
}
// Okay, need to actually store it.
// It will be a dataset with an internal attribute for storing
// the hash id.
bool hasData = dims.numPoints() > 0;
hid_t dspaceId = -1;
if ( hasData )
{
hsize_t hdim = dims.numPoints() * dataType.getExtent();
dspaceId = H5Screate_simple( 1, &hdim, NULL );
}
else
{
dspaceId = H5Screate( H5S_NULL );
}
ABCA_ASSERT( dspaceId >= 0,
"WriteArray() Failed in dataspace construction" );
DspaceCloser dspaceCloser( dspaceId );
hid_t dsetId = -1;
if ( iCompressionLevel >= 0 && hasData )
{
// Make a compression plist
hid_t zipPlist = DsetGzipCreatePlist( dims,
iCompressionLevel > 9 ? 9 : iCompressionLevel );
PlistCloser plistCloser( zipPlist );
// Make the dataset.
dsetId = H5Dcreate2( iGroup, iName.c_str(), iFileType, dspaceId,
H5P_DEFAULT, zipPlist, H5P_DEFAULT );
}
else
{
dsetId = H5Dcreate2( iGroup, iName.c_str(),
iFileType, dspaceId,
H5P_DEFAULT, H5P_DEFAULT, H5P_DEFAULT );
}
DsetCloser dsetCloser(dsetId);
ABCA_ASSERT( dsetId >= 0,
"WriteArray() Failed in dataset constructor" );
// Write the data.
if ( hasData )
{
H5Dwrite( dsetId, iNativeType, H5S_ALL, H5S_ALL, H5P_DEFAULT,
iSamp.getData() );
}
// Write the array sample key.
WriteKey( dsetId, "key", iKey );
writeID.reset( new WrittenArraySampleID( iKey, dsetId ) );
iMap.store( writeID );
// Return the reference.
return writeID;
}
//-*****************************************************************************
AbcA::ArraySamplePtr
ReadArray( AbcA::ReadArraySampleCachePtr iCache,
hid_t iParent,
const std::string &iName,
const AbcA::DataType &iDataType,
hid_t iFileType,
hid_t iNativeType )
{
// Dispatch string stuff.
if ( iDataType.getPod() == kStringPOD )
{
return ReadStringArray( iCache, iParent, iName, iDataType );
}
else if ( iDataType.getPod() == kWstringPOD )
{
return ReadWstringArray( iCache, iParent, iName, iDataType );
}
assert( iDataType.getPod() != kStringPOD &&
iDataType.getPod() != kWstringPOD );
// Open the data set.
hid_t dsetId = H5Dopen( iParent, iName.c_str(), H5P_DEFAULT );
ABCA_ASSERT( dsetId >= 0, "Cannot open dataset: " << iName );
DsetCloser dsetCloser( dsetId );
// Read the data space.
hid_t dspaceId = H5Dget_space( dsetId );
ABCA_ASSERT( dspaceId >= 0, "Could not get dataspace for dataSet: "
<< iName );
DspaceCloser dspaceCloser( dspaceId );
AbcA::ArraySample::Key key;
bool foundDigest = false;
// if we are caching, get the key and see if it is being used
if ( iCache )
{
key.origPOD = iDataType.getPod();
key.readPOD = key.origPOD;
key.numBytes = Util::PODNumBytes( key.readPOD ) *
H5Sget_simple_extent_npoints( dspaceId );
foundDigest = ReadKey( dsetId, "key", key );
AbcA::ReadArraySampleID found = iCache->find( key );
if ( found )
{
AbcA::ArraySamplePtr ret = found.getSample();
assert( ret );
if ( ret->getDataType().getPod() != iDataType.getPod() )
{
ABCA_THROW( "ERROR: Read data type for dset: " << iName
<< ": " << ret->getDataType()
<< " does not match expected data type: "
<< iDataType );
}
// Got it!
return ret;
}
}
// Okay, we haven't found it in a cache.
// Read the data type.
hid_t dtypeId = H5Dget_type( dsetId );
ABCA_ASSERT( dtypeId >= 0, "Could not get datatype for dataSet: "
<< iName );
DtypeCloser dtypeCloser( dtypeId );
ABCA_ASSERT( EquivalentDatatypes( iFileType, dtypeId ),
"File DataType clash for array dataset: "
<< iName );
AbcA::ArraySamplePtr ret;
H5S_class_t dspaceClass = H5Sget_simple_extent_type( dspaceId );
if ( dspaceClass == H5S_SIMPLE )
{
// Get the dimensions
int rank = H5Sget_simple_extent_ndims( dspaceId );
ABCA_ASSERT( rank == 1,
"H5Sget_simple_extent_ndims() must be 1." );
hsize_t hdim = 0;
rank = H5Sget_simple_extent_dims( dspaceId, &hdim, NULL );
Dimensions dims;
std::string dimName = iName + ".dims";
if ( H5Aexists( iParent, dimName.c_str() ) )
{
ReadDimensions( iParent, dimName, dims );
}
else
{
dims.setRank(1);
dims[0] = hdim / iDataType.getExtent();
}
ABCA_ASSERT( dims.numPoints() > 0,
"Degenerate dims in Dataset read" );
// Create a buffer into which we shall read.
ret = AbcA::AllocateArraySample( iDataType, dims );
assert( ret->getData() );
// And... read into it.
herr_t status = H5Dread( dsetId, iNativeType,
H5S_ALL, H5S_ALL, H5P_DEFAULT,
const_cast<void*>( ret->getData() ) );
ABCA_ASSERT( status >= 0, "H5Dread() failed." );
}
else if ( dspaceClass == H5S_NULL )
{
Dimensions dims;
std::string dimName = iName + ".dims";
if ( H5Aexists( iParent, dimName.c_str() ) )
{
ReadDimensions( iParent, dimName, dims );
ABCA_ASSERT( dims.rank() > 0,
"Degenerate rank in Dataset read" );
// Num points should be zero here.
ABCA_ASSERT( dims.numPoints() == 0,
"Expecting zero points in dimensions" );
}
else
{
dims.setRank(1);
dims[0] = 0;
}
ret = AbcA::AllocateArraySample( iDataType, dims );
}
else
{
ABCA_THROW( "Unexpected scalar dataspace encountered." );
}
// Store if there is a cache.
if ( foundDigest && iCache )
{
AbcA::ReadArraySampleID stored = iCache->store( key, ret );
if ( stored )
{
return stored.getSample();
}
}
// Otherwise, just leave! ArraySamplePtr returned by AllocateArraySample
// already has fancy-dan deleter built in.
// I REALLY LOVE SMART PTRS.
return ret;
}