Chris Marshall

NAME

PDL::IO::HDF5::Dataset - PDL::IO::HDF5 Helper Object representing HDF5 datasets.

DESCRIPTION

This is a helper-object used by PDL::IO::HDF5 to interface with HDF5 format's dataset objects. Information on the HDF5 Format can be found at the HDF Group's web site at http://www.hdfgroup.org .

SYNOPSIS

See PDL::IO::HDF5

MEMBER DATA

ID

ID number given to the dataset by the HDF5 library

name

Name of the dataset.

parent

Ref to parent object (group) that owns this dateset.

fileObj

Ref to the PDL::IO::HDF5 object that owns this object.

METHODS

####---------------------------------------------------------

new

PDL::IO::HDF5::Dataset Constructor - creates new object

Usage:

This object will usually be created using the calling format detailed in the SYNOPSIS. The following syntax is used by the PDL::IO::HDF5 object to build the object.

   $a = new PDL::IO::HDF5:Dataset( name => $name, parent => $parent, 
                                fileObj => $fileObj);
        Args:
        $name                           Name of the dataset
        $parent                         Parent Object that owns this dataset
        $fileObj                        PDL::HDF object that owns this dateset.

DESTROY

PDL::IO::HDF5::Dataset Destructor - Closes the dataset object

Usage:

   No Usage. Automatically called
   

set

Write data to the HDF5 dataset

Usage:

 $dataset->set($pdl, unlimited => 1);     # Write the array data in the dataset

     Options:
     unlimited     If present, the dataset is created with unlimited dimensions.

get

Get data from a HDF5 dataset to a PDL

Usage:

 $pdl = $dataset->get;     # Read the Array from the HDF5 dataset, create a PDL from it
                           #  and put in $pdl

                           # Assuming $dataset is three dimensional
                           # with dimensions (20,100,90)

The get method can also be used to obtain particular slices or hyperslabs of the dataset array. For example, if $dataset is three dimensional with dimensions (20,100,90) then we could do:

 $start=pdl([0,0,0]);      # We begin the slice at the very beggining
 $end=pdl([19,0,0]);       # We take the first vector of the array,
 $stride=pdl([2,1,1]);     # taking only every two values of the vector 

 $pdl = $dataset->get($start,$end,[$stride]); # Read a slice or
                           # hyperslab from the HDF5 dataset.
                           # $start, $end and optionally $stride
                           # should be PDL vectors with length the
                           # number of dimensions of the dataset.
                           # $start gives the starting coordinates
                           # in the array.
                           # $end gives the ending coordinate
                           # in the array
                           # $stride gives the steps taken from one
                           # coordinate to the next of the slice

The mapping of HDF5 datatypes in the file to PDL datatypes in memory will be according to the following table.

 HDF5 File Type                         PDL Type
 ------------------------               -----------------
 PDL::IO::HDF5::H5T_C_S1()      =>      PDL::Char Object    (Special Case for Char Strings)
 PDL::IO::HDF5::H5T_STD_I8BE()  =>      $PDL::Types::PDL_B
 PDL::IO::HDF5::H5T_STD_I8LE()  =>      $PDL::Types::PDL_B,
 PDL::IO::HDF5::H5T_STD_U8BE()  =>      $PDL::Types::PDL_S,
 PDL::IO::HDF5::H5T_STD_U8LE()  =>      $PDL::Types::PDL_S,
 PDL::IO::HDF5::H5T_STD_I16BE() =>      $PDL::Types::PDL_S,
 PDL::IO::HDF5::H5T_STD_I16LE() =>      $PDL::Types::PDL_S,
 PDL::IO::HDF5::H5T_STD_U16BE() =>      $PDL::Types::PDL_L,
 PDL::IO::HDF5::H5T_STD_U16LE() =>      $PDL::Types::PDL_L,
 PDL::IO::HDF5::H5T_STD_I32BE() =>      $PDL::Types::PDL_L,
 PDL::IO::HDF5::H5T_STD_I32LE() =>      $PDL::Types::PDL_L,
 PDL::IO::HDF5::H5T_STD_U32LE() =>      $PDL::Types::PDL_LL,
 PDL::IO::HDF5::H5T_STD_U32BE() =>      $PDL::Types::PDL_LL,
 PDL::IO::HDF5::H5T_STD_I64LE() =>      $PDL::Types::PDL_LL,
 PDL::IO::HDF5::H5T_STD_I64BE() =>      $PDL::Types::PDL_LL,
 PDL::IO::HDF5::H5T_IEEE_F32BE()=>      $PDL::Types::PDL_F,
 PDL::IO::HDF5::H5T_IEEE_F32LE()=>      $PDL::Types::PDL_F,
 PDL::IO::HDF5::H5T_IEEE_F64BE()=>      $PDL::Types::PDL_D,
 PDL::IO::HDF5::H5T_IEEE_F64LE()=>      $PDL::Types::PDL_D

For HDF5 File types not in this table, this method will attempt to map it to the default PDL type PDL_D.

If the dataset being read is a scalar reference, the referenced dataset region will be read instead.

Note:

Character arrays are returned as the special PDL::Char fixed-length string type. For fixed-length HDF5 string arrays, this is a direct mapping to the PDL::Char datatype. For HDF5 variable-length string arrays, the data is converted to a fixed-length character array, with a string size equal to the maximum size of all the strings in the array.

dims

Get the dims for a HDF5 dataset. For example, a 3 x 4 array would return a perl array (3,4);

Usage:

 @pdl = $dataset->dims;    # Get an array of dims. 

attrSet

Set the value of an attribute(s)

Attribute types supported are null-terminated strings and PDL matrices

Usage:

   $dataset->attrSet( 'attr1' => 'attr1Value',
                    'attr2' => 'attr2 value', 
                    'attr3' => $pdl,
                    .
                    .
                    .
                   );

Returns undef on failure, 1 on success.

attrDel

Delete attribute(s)

Usage:

 $dataset->attrDel( 'attr1', 
                    'attr2',
                    .
                    .
                    .
                   );

Returns undef on failure, 1 on success.

attrs

Get a list of all attribute names associated with a dataset

Usage:

   @attrs = $dataset->attrs;

attrGet

Get the value of an attribute(s)

Currently the attribute types supported are null-terminated strings and PDLs.

Usage:

   my @attrs = $dataset->attrGet( 'attr1', 'attr2');

IDget

Returns the HDF5 library ID for this object

Usage:

 my $ID = $dataSetObj->IDget;

nameGet

Returns the HDF5 Dataset Name for this object.

Usage:

 my $name = $datasetObj->nameGet;



Hosting generously
sponsored by Bytemark