HDF5File Class Reference

Import/Export of Images and Arrays in HDF5 Format

Access to HDF5 files. More...

#include <vigra/hdf5impex.hxx>

Public Types
enum	OpenMode
	Set how a file is opened. More...

Public Member Functions
void	cd (std::string groupName)
	Change the current group. Both absolute and relative group names are allowed.

void	cd_mk (std::string groupName)
	Change the current group; create it if necessary. If the first character is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

bool	cd_up ()
	Change the current group to its parent group. Returns true if successful, false otherwise. If unsuccessful, the group will not change.

bool	cd_up (int levels)
	Change the current group to its parent group. Returns true if successful, false otherwise. If unsuccessful, the group will not change.

void	close ()
	Close the current file.

template<int N, class T >
HDF5HandleShared	createDataset (std::string datasetName, TinyVector< MultiArrayIndex, N > const &shape, typename detail::HDF5TypeTraits< T >::value_type init=typename detail::HDF5TypeTraits< T >::value_type(), TinyVector< MultiArrayIndex, N > const &chunkSize=(TinyVector< MultiArrayIndex, N >()), int compressionParameter=0)
	Create a new dataset. This function can be used to create a dataset filled with a default value init, for example before writing data into it using writeBlock(). More...

bool	existsAttribute (std::string object_name, std::string attribute_name)
	Test if attribute exists.

bool	existsDataset (std::string datasetName) const
	Check if given datasetName exists.

std::string	filename () const
	Get the name of the associated file.

void	flushToDisk ()
	Immediately write all data to disk.

std::string	get_absolute_path (std::string path) const
	takes any path and converts it into an absolute path in the current file. More...

HDF5Handle	getAttributeHandle (std::string dataset_name, std::string attribute_name) const
	Obtain the HDF5 handle of a attribute.

ArrayVector< hsize_t >	getChunkShape (std::string datasetName) const
	Get the shape of chunks along each dimension of a certain dataset. More...

hssize_t	getDatasetDimensions (std::string datasetName) const
	Get the number of dimensions of a certain dataset If the first character is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

HDF5Handle	getDatasetHandle (std::string const &datasetName) const
	Obtain the HDF5 handle of a dataset.

HDF5HandleShared	getDatasetHandleShared (std::string const &datasetName) const
	Obtain a shared HDF5 handle of a dataset.

ArrayVector< hsize_t >	getDatasetShape (std::string datasetName) const
	Get the shape of each dimension of a certain dataset. More...

std::string	getDatasetType (std::string const &datasetName) const

HDF5Handle	getGroupHandle (std::string group_name, std::string function_name="HDF5File::getGroupHandle()")
	Obtain the HDF5 handle of a group (create the group if it doesn't exist).

	HDF5File ()
	Default constructor. More...

	HDF5File (bool track_creation_times)
	Construct with time tagging of datasets enabled. More...

	HDF5File (std::string filePath, OpenMode mode=ReadOnly, bool track_creation_times=false)
	Open or create an HDF5File object. More...

	HDF5File (char const *filePath, OpenMode mode=ReadOnly, bool track_creation_times=false)
	Open or create an HDF5File object. More...

	HDF5File (HDF5HandleShared const &fileHandle, const std::string &pathname="", bool read_only=false)
	Initialize an HDF5File object from HDF5 file handle. More...

	HDF5File (HDF5File const &other)
	Copy a HDF5File object. More...

std::vector< std::string >	listAttributes (std::string const &group_or_dataset) const
	List the attribute names of the given group or dataset. More...

template<class Container >
void	listAttributes (std::string const &group_or_dataset, Container &container) const
	Insert the attribute names of the given group or dataset into the given container by calling `container.insert(std::string)`. More...

std::vector< std::string >	ls () const
	List the contents of the current group. The function returns a vector of strings holding the entries of the current group. Only datasets and groups are listed, other objects (e.g. datatypes) are ignored. Group names always have a trailing "/".

template<class Container >
void	ls (Container &cont) const
	List the contents of the current group into a container-like object via insert(). More...

void	mkdir (std::string groupName)
	Create a new group. If the first character is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

void	open (std::string filePath, OpenMode mode)
	Open or create the given file in the given mode and set the group to "/". If another file is currently open, it is first closed.

HDF5File &	operator= (HDF5File const &other)
	Assign a HDF5File object. More...

std::string	pwd () const
	Get the path of the current group.

template<unsigned int N, class T , class Stride >
void	read (std::string datasetName, MultiArrayView< N, T, Stride > array)
	Read data into a multi array. If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group. More...

template<class T >
void	read (const std::string &datasetName, ArrayVectorView< T > array)
	Read data into an array vector. If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

void	read (std::string datasetName, char &data)
	Read a single value. Specialization of the read function for simple datatypes.

template<unsigned int N, class T , class Alloc >
void	readAndResize (std::string datasetName, MultiArray< N, T, Alloc > &array)
	Read data into a MultiArray. Resize MultiArray to the correct size. If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group. More...

template<class T >
void	readAndResize (std::string datasetName, ArrayVector< T > &array)
	Read data into an array vector. Resize the array vector to the correct size. If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

template<unsigned int N, class T , class Stride >
void	readAttribute (std::string object_name, std::string attribute_name, MultiArrayView< N, T, Stride > array)
	Read MultiArray Attributes. In contrast to datasets, subarray access is not available.

void	readAttribute (std::string object_name, std::string attribute_name, char &data)
	Read a single value. Specialization of the read function for simple datatypes.

template<unsigned int N, class T , class Stride >
void	readBlock (std::string datasetName, typename MultiArrayShape< N >::type blockOffset, typename MultiArrayShape< N >::type blockShape, MultiArrayView< N, T, Stride > array)
	Read a block of data into a multi array. This function allows to read a small block out of a larger volume stored in an HDF5 dataset. More...

void	root ()
	Change current group to "/".

template<unsigned int N, class T , class Stride >
void	write (std::string datasetName, const MultiArrayView< N, T, Stride > &array, int iChunkSize=0, int compression=0)
	Write multi arrays. More...

template<unsigned int N, class T , class Stride >
void	write (std::string datasetName, const MultiArrayView< N, T, Stride > &array, typename MultiArrayShape< N >::type chunkSize, int compression=0)
	Write multi arrays. Chunks can be activated by providing a MultiArrayShape as chunkSize. chunkSize must have equal dimension as array. More...

template<class T >
void	write (const std::string &datasetName, const ArrayVectorView< T > &array, int compression=0)
	Write array vectors. More...

void	write (std::string datasetName, char data)
	Write a single value. Specialization of the write function for simple datatypes.

template<unsigned int N, class T , class Stride >
void	writeAttribute (std::string object_name, std::string attribute_name, const MultiArrayView< N, T, Stride > &array)
	Write MultiArray Attributes. In contrast to datasets, subarray access, chunks and compression are not available.

void	writeAttribute (std::string object_name, std::string attribute_name, char data)
	Write a single value. Specialization of the write function for simple datatypes.

template<unsigned int N, class T , class Stride >
void	writeBlock (std::string datasetName, typename MultiArrayShape< N >::type blockOffset, const MultiArrayView< N, T, Stride > &array)
	Write a multi array into a larger volume. blockOffset determines the position, where array is written. More...

	~HDF5File ()
	The destructor flushes and closes the file.

Detailed Description

Access to HDF5 files.

HDF5File provides a convenient way of accessing data in HDF5 files. vigra::MultiArray structures of any dimension can be stored to / loaded from HDF5 files. Typical HDF5 features like subvolume access, chunks and data compression are available, string attributes can be attached to any dataset or group. Group- or dataset-handles are encapsulated in the class and managed automatically. The internal file-system like structure can be accessed by functions like "cd()" or "mkdir()".

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This means that a VIGRA MultiArray, whose indices represent the 'x'-, 'y'-, and 'z'-axis in that order, is reversed upon writing to an HDF5 file, i.e. in the file the axis order is 'z', 'y', 'x'. Likewise, the order is reversed upon reading.

Example: Write the MultiArray out_multi_array to file. Change the current directory to "/group" and read in the same MultiArray as in_multi_array.

HDF5File file("/path/to/file",HDF5File::New);
file.mkdir("group");
file.write("/group/dataset", out_multi_array);
file.cd("/group");
file.read("dataset", in_multi_array);

#include <vigra/hdf5impex.hxx>
Namespace: vigra

Member Enumeration Documentation

enum OpenMode

Set how a file is opened.

OpenMode::New creates a new file. If the file already exists, it is overwritten.

OpenMode::ReadWrite opens a file for reading/writing. The file will be created if it doesn't exist.

OpenMode::ReadOnly opens a file for reading. The file as well as any dataset to be accessed must already exist.

Constructor & Destructor Documentation

HDF5File ( )

Default constructor.

A file can later be opened via the open() function. Time tagging of datasets is disabled.

HDF5File ( bool track_creation_times )

explicit

Construct with time tagging of datasets enabled.

If track_creation_times is 'true', time tagging of datasets will be enabled.

HDF5File	(	std::string	filePath,
		OpenMode	mode = `ReadOnly`,
		bool	track_creation_times = `false`
	)

explicit

Open or create an HDF5File object.

Creates or opens HDF5 file with given filename. The current group is set to "/". By default, the files is opened in read-only mode.

HDF5File	(	char const *	filePath,
		OpenMode	mode = `ReadOnly`,
		bool	track_creation_times = `false`
	)

explicit

Open or create an HDF5File object.

Creates or opens HDF5 file with given filename. The current group is set to "/". By default, the files is opened in read-only mode.

HDF5File	(	HDF5HandleShared const &	fileHandle,
		const std::string &	pathname = `""`,
		bool	read_only = `false`
	)

explicit

Initialize an HDF5File object from HDF5 file handle.

Initializes an HDF5File object corresponding to the HDF5 file opened elsewhere. If fileHandle is constructed with a NULL destructor, ownership is not transferred to the new HDF5File object, and you must ensure that the file is not closed while the new HDF5File object is in use. Otherwise, ownership will be shared.

The current group is set to the specified pathname. If read_only is 'true', you cannot create new datasets or overwrite data.

Warning: When VIRA is linked against a different HDF5 library than the one used to open the file with the given id, this method will lead to crashes.

HDF5File ( HDF5File const & other )

Copy a HDF5File object.

The new object will refer to the same file and group as other.

Member Function Documentation

HDF5File& operator= ( HDF5File const & other )

Assign a HDF5File object.

Calls close() on the present file and refers to the same file and group as other afterwards.

void ls ( Container & cont ) const

List the contents of the current group into a container-like object via insert().

Only datasets and groups are inserted, other objects (e.g., datatypes) are ignored. Group names always have a trailing "/".

The argument cont is presumably an associative container, however, only its member function cont.insert(std::string) will be called.

Parameters

cont	reference to a container supplying a member function `insert(const i_type &)`, where `i_type` is convertible to `std::string`.

ArrayVector<hsize_t> getDatasetShape ( std::string datasetName ) const

Get the shape of each dimension of a certain dataset.

Normally, this function is called after determining the dimension of the dataset using getDatasetDimensions(). If the first character is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This function therefore reverses the axis order relative to the file contents. That is, when the axes in the file are ordered as 'z', 'y', 'x', this function will return the shape in the order 'x', 'y', 'z'.

ArrayVector<hsize_t> getChunkShape ( std::string datasetName ) const

Get the shape of chunks along each dimension of a certain dataset.

Normally, this function is called after determining the dimension of the dataset using getDatasetDimensions(). If the first character is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This function therefore reverses the axis order relative to the file contents. That is, when the axes in the file are ordered as 'z', 'y', 'x', this function will return the shape in the order 'x', 'y', 'z'.

std::string getDatasetType ( std::string const & datasetName ) const

Query the pixel type of the dataset.

Possible values are:

"INT8": 8-bit signed integer (unsigned char)
"UINT8": 8-bit unsigned integer (unsigned char)
"INT16": 16-bit signed integer (short)
"UINT16": 16-bit unsigned integer (unsigned short)
"INT32": 32-bit signed integer (long)
"UINT32": 32-bit unsigned integer (unsigned long)
"INT64": 64-bit signed integer (long long)
"UINT64": 64-bit unsigned integer (unsigned long long)
"FLOAT": 32-bit floating point (float)
"DOUBLE": 64-bit floating point (double)
"UNKNOWN": any other type

std::vector<std::string> listAttributes ( std::string const & group_or_dataset ) const

List the attribute names of the given group or dataset.

If group_or_dataset is empty or "." (a dot), the command refers to the current group of this file object.

void listAttributes	(	std::string const &	group_or_dataset,
		Container &	container
	)		const

Insert the attribute names of the given group or dataset into the given container by calling container.insert(std::string).

If group_or_dataset is empty or "." (a dot), the command refers to the current group of this file object.

void write	(	std::string	datasetName,
		const MultiArrayView< N, T, Stride > &	array,
		int	iChunkSize = `0`,
		int	compression = `0`
	)

Write multi arrays.

Chunks can be activated by setting

iChunkSize = size; //size > 0

. The chunks will be hypercubes with edge length size. When iChunkSize == 0 (default), the behavior depends on the compression setting: If no compression is requested, the data is written without chunking. Otherwise, chuning is required, and the chunk size is automatically selected such that each chunk contains about 300k pixels.

Compression can be activated by setting

compression = parameter; // 0 < parameter <= 9

where 0 stands for no compression and 9 for maximum compression.

If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This means that a VIGRA MultiArray, whose indices represent the 'x'-, 'y'-, and 'z'-axis in that order, is reversed upon writing to an HDF5 file, i.e. in the file the axis order is 'z', 'y', 'x'.

void write	(	std::string	datasetName,
		const MultiArrayView< N, T, Stride > &	array,
		typename MultiArrayShape< N >::type	chunkSize,
		int	compression = `0`
	)

Write multi arrays. Chunks can be activated by providing a MultiArrayShape as chunkSize. chunkSize must have equal dimension as array.

Compression can be activated by setting

compression = parameter; // 0 < parameter <= 9

where 0 stands for no compression and 9 for maximum compression.

If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This means that a VIGRA MultiArray, whose indices represent the 'x'-, 'y'-, and 'z'-axis in that order, is reversed upon writing to an HDF5 file, i.e. in the file the axis order is 'z', 'y', 'x'.

void writeBlock	(	std::string	datasetName,
		typename MultiArrayShape< N >::type	blockOffset,
		const MultiArrayView< N, T, Stride > &	array
	)

Write a multi array into a larger volume. blockOffset determines the position, where array is written.

If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This means that a VIGRA MultiArray, whose indices represent the 'x'-, 'y'-, and 'z'-axis in that order, is reversed upon writing to an HDF5 file, i.e. in the file the axis order is 'z', 'y', 'x'.

void write	(	const std::string &	datasetName,
		const ArrayVectorView< T > &	array,
		int	compression = `0`
	)

Write array vectors.

Compression can be activated by setting

compression = parameter; // 0 < parameter <= 9

where 0 stands for no compression and 9 for maximum compression.

If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

void read	(	std::string	datasetName,
		MultiArrayView< N, T, Stride >	array
	)

Read data into a multi array. If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This means that a HDF5 dataset, whose indices represent the 'z'-, 'y'-, and 'x'-axis in that order, is reversed upon reading into a MultiArrayView, i.e. in the array axis order must be 'x', 'y', 'z'.

void readAndResize	(	std::string	datasetName,
		MultiArray< N, T, Alloc > &	array
	)

Read data into a MultiArray. Resize MultiArray to the correct size. If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This means that a HDF5 dataset, whose indices represent the 'z'-, 'y'-, and 'x'-axis in that order, is reversed upon reading into a MultiArray, i.e. in the array axis order will be 'x', 'y', 'z'.

void readBlock	(	std::string	datasetName,
		typename MultiArrayShape< N >::type	blockOffset,
		typename MultiArrayShape< N >::type	blockShape,
		MultiArrayView< N, T, Stride >	array
	)

Read a block of data into a multi array. This function allows to read a small block out of a larger volume stored in an HDF5 dataset.

blockOffset determines the position of the block. blockSize determines the size in each dimension of the block.

If the first character of datasetName is a "/", the path will be interpreted as absolute path, otherwise it will be interpreted as path relative to the current group.

Note that the memory order between VIGRA and HDF5 files differs: VIGRA uses Fortran-order, while HDF5 uses C-order. This means that a HDF5 dataset, whose indices represent the 'z'-, 'y'-, and 'x'-axis in that order, is reversed upon reading into a MultiArray, i.e. in the array axis order will be 'x', 'y', 'z'.

std::string get_absolute_path ( std::string path ) const

takes any path and converts it into an absolute path in the current file.

Elements like "." and ".." are treated as expected. Links are not supported or resolved.

The documentation for this class was generated from the following file:

vigra/hdf5impex.hxx

Public Types

Public Member Functions

Detailed Description

Member Enumeration Documentation

Constructor & Destructor Documentation

Member Function Documentation