echelon::dataset Class Reference

A handle to an echelon dataset. More...

#include <dataset.hpp>

Public Member Functions
	dataset ()=default
	Initializes the handle with its null state.
template<typename Container >
void	extend_along (std::size_t dimension_index, const Container &container)
	Extends the dataset along a given dimension. More...
std::vector< hsize_t >	shape () const
	The shape of the dataset.
std::size_t	rank () const
	The rank of the dataset.
type	datatype () const
	The value type of the dataset.
template<typename... Args>
slice	operator() (Args...args) const
	Slices the dataset. More...
object_reference	ref () const
	A echelon object reference to this dataset.
native_handle_type	native_handle () const
	The underlying HDF5 low-level handle.
	operator bool () const
	Tests the validity of the handle.
attribute_repository< dataset >	attributes () const
	The attributes, which are attached to the dataset.
dataset_dimensions	dimensions () const
	The dimensions of the dataset.

Friends
template<typename T >
void	operator<<= (dataset &sink, const T &source)
	Writes the content of a data source into the dataset. More...
template<typename T >
void	operator<<= (T &sink, const dataset &source)
	Reads the content of the dataset into a data sink. More...

Detailed Description

A handle to an echelon dataset.

Member Function Documentation

template<typename Container >

void echelon::dataset::extend_along ( std::size_t  dimension_index, const Container &  container  )

inline

Extends the dataset along a given dimension.

Template Parameters

Container

type of the container; Container must satisfy the data sink requirements.

Parameters

dimension_index	index of the extended dimension
container	container, which is used to fill the new portions of the dataset

template<typename... Args>

slice echelon::dataset::operator() ( Args...  args ) const

inline

Slices the dataset.

The boundaries of the slice within the dataset are specified by index ranges, which can be constructed using echelon::range. Currently the following index range specifiers are supported (N and M are non-negative integral values and step is a positive integral value):

index range	semantic
range(N,M)	restrict the dimension to the interval .
range(N,M,step)	restrict the dimension to the interval and only include every step-th value.
N	restrict the dimension to the single value N. This is equivalent to range(N,N+1)
_	do not restrict the dimension, but use the full range of indices

Within the range syntax the wildcard _ can be used as the lower or upper bound to specifiy, that the dimension should not be restricted through this bound.

The number of index range specifiers must match the rank of the dataset.

If an index range only contains one value, the rank of the slice is reduced by one. If none of the index range satisfies this condition, the rank of the slice equals the rank of the dataset.

The following example shows, how a slice can be used to access a portion of a dataset:

auto foo = some_group.create_dataset<double>("foo",{10,10,10});

multi_array<double> data({6,10},1.0);

// set the first dimension to 2 and clamp the second dimension between 2
// and 8. Afterwards the data is written into the 2-slice.
foo(2,range(2,8),_) <<= data;

It should be noted, that only the sliced portion of the dataset is loaded into the main memory. Therefore slicing is an efficient way to work on large datasets and even allows us to deal with datasets , which do not fit into the main memory.

At the time of writing, slice does support most operations, which are valid on a dataset.

Template Parameters

Args	types of the index range specifiers

Parameters

args	index range specifiers

Friends And Related Function Documentation

template<typename T >

void operator<<= ( dataset &  sink, const T &  source  )

friend

Writes the content of a data source into the dataset.

The shape of the data source must match the shape of the dataset.

Template Parameters

T	type of the container; T must satisfy the data source requirements.

Parameters

sink	the dataset, which is used as a sink
source	the data source

template<typename T >

void operator<<= ( T &  sink, const dataset &  source  )

friend

Reads the content of the dataset into a data sink.

Template Parameters

T	type of the container; T must satisfy the data sink requirements.

Parameters

sink	the data sink
source	the dataset, which is used as a source

---

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

• dataset.hpp
• dataset.cpp