1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201
|
/*==============================================================================
Program: Visualization Toolkit
Module: ExampleDataArrayRangeDispatch.cxx
Copyright (c) Ken Martin, Will Schroeder, Bill Lorensen
All rights reserved.
See Copyright.txt or http://www.kitware.com/Copyright.htm for details.
This software is distributed WITHOUT ANY WARRANTY; without even
the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
PURPOSE. See the above copyright notice for more information.
==============================================================================*/
// This file provides some examples of how to use the ArrayDispatch system
// with the DataArrayRange utilities to create high performance algorithms that
// will work with all vtkDataArrays.
//
// ArrayDispatch provides a mechanism to automatically downcast a vtkDataArray
// object to a subclass (such as vtkFloatArray, vtkSOADataArrayTemplate<int>,
// etc). Using a subclass instead of a vtkDataArray allows the array data
// to be accessed directly in memory, while using the vtkDataArray API requires
// virtual methods and type conversions.
//
// The vtk::DataArrayRange utilities provide STL-style iterators that abstract
// the details of reading/writing values in a vtkDataArray. When used with a
// specific subclass of vtkDataArray, optimized memory access are used. When
// used with a vtkDataArray, the virtual interface is used. This allows a
// single algorithm to be written using these Range objects that will serve
// as both a fast-path for common array types, while also functioning as a
// slower fallback path for uncommon array types.
//
// This example fills an array with values increasing from [0, size), sums the
// values in the array, and then copies data into another array.
//
// The fill is performed using the STL function `std::iota` to show how the
// Ranges interact with the standard library.
//
// The sum is performed using `std::accumulate`.
//
// The copy is performed using a C++11-style for-range loop to retrieve the
// source values, and a raw iterator is used to write the output to the
// destination array. This demonstrates the similarity between using the Range
// iterators and the more familiar pointer-style algorithms.
#include "vtkDataArrayRange.h"
#include "vtkArrayDispatch.h"
#include "vtkFloatArray.h"
#include "vtkNew.h"
#include <algorithm>
#include <numeric>
namespace
{
// Fills the supplied array with increasing values, starting from 0.
struct FillImpl
{
template <typename ArrayType>
void operator()(ArrayType* array) const
{
// We know this is a single component array, so providing a template
// parameter of `1` enables additional optimizations. The template
// parameter may be omitted if the tuple size is unknown:
auto range = vtk::DataArrayValueRange<1>(array);
// The begin/end methods return STL-style iterators, similar to std::vector.
std::iota(range.begin(), range.end(), 0);
}
};
// Adds all values in Range into Sum.
struct SumImpl
{
double Sum{ 0.0 };
template <typename ArrayType>
void operator()(ArrayType* array)
{
const auto range = vtk::DataArrayValueRange<1>(array);
// The cbegin/cend methods return constant iterators that prevent the data
// they point at from being modified.
this->Sum = std::accumulate(range.cbegin(), range.cend(), 0.0);
}
};
// Copies from the src range to the dst range.
struct CopyImpl
{
template <typename SrcArray, typename DstArray>
void operator()(SrcArray* src, DstArray* dst) const
{
// The vtk::GetAPIType<ArrayType> helper provides the API type of the
// data array (e.g. `float` for vtkFloatArray, `double` for the virtual
// vtkDataArray API).
using SrcValueType = vtk::GetAPIType<SrcArray>;
using DstValueType = vtk::GetAPIType<DstArray>;
// Create range object for the arrays. These work whether the arrays are
// downcasted AOS arrays, SOA array, etc or plain vtkDataArrays.
const auto srcRange = vtk::DataArrayValueRange<1>(src);
auto dstRange = vtk::DataArrayValueRange<1>(dst);
// Ensure that the ranges are the same size:
assert(srcRange.size() == dstRange.size());
// We use a C++11-style ranged-for loop to retrieve the values from src,
// and write to an iterator obtained from dst. These iterators behave
// like pointers.
auto dstIter = dstRange.begin();
for (SrcValueType srcVal : srcRange)
{
*dstIter++ = static_cast<DstValueType>(srcVal);
}
}
};
// Create an AOS<float> array and return a plain vtkDataArray pointer.
vtkDataArray* CreateArray()
{
vtkNew<vtkFloatArray> aosArray;
aosArray->SetNumberOfComponents(1);
aosArray->SetNumberOfTuples(1024);
aosArray->Register(nullptr); // Add a reference before we return.
return aosArray;
}
// std::equal predicate to do "close enough" equality comparisons to work
// around floating point issues.
struct CloseEnough
{
bool operator()(double a, double b) { return std::fabs(a - b) < 1e-5; }
};
} // end anon namespace
int ExampleDataArrayRangeDispatch(int, char*[])
{
// Create the array. The auto type here is vtkSmartPointer<vtkDataArray>.
auto array = vtk::TakeSmartPointer(CreateArray());
// Attempt to downcast the array to a known type and call the FillImpl
// functor with the downcasted array:
FillImpl fillFunctor;
if (!vtkArrayDispatch::Dispatch::Execute(array, fillFunctor))
{
// If Execute returns false, the input array type was not determined because
// the array was not in the set of supported array types. No worries, we
// can reuse the functor with the vtkDataArray pointer as a fallback:
fillFunctor(array.Get());
}
// SumImpl:
SumImpl sumFunctor;
if (!vtkArrayDispatch::Dispatch::Execute(array, sumFunctor))
{
sumFunctor(array.Get());
}
// Verify the sum:
double expected = (1024. * 1023.) / 2.;
if (std::fabs(sumFunctor.Sum - expected) > 1e-5)
{
std::cerr << "Sum was not as expected: " << sumFunctor.Sum << " (expected: " << expected
<< ")\n";
return EXIT_FAILURE;
}
// CopyImpl:
{
// Create an array of ints, and copy the other array's data into it.
// The auto type here is vtkSmartPointer<vtkDataArray>.
auto intArray = vtk::TakeSmartPointer(vtkDataArray::CreateDataArray(VTK_INT));
intArray->SetNumberOfComponents(array->GetNumberOfComponents());
intArray->SetNumberOfTuples(array->GetNumberOfTuples());
// Dispatch2 dispatches two arrays at once. Various other dispatchers are
// documented in the vtkArrayDispatch header.
CopyImpl copyFunctor;
if (!vtkArrayDispatch::Dispatch2::Execute(array, intArray, copyFunctor))
{
copyFunctor(array.Get(), intArray.Get());
}
// Check the arrays are equal using the vtkDataArray double interface via
// range iterators and std::equal:
const auto srcRange = vtk::DataArrayValueRange<1>(array);
const auto dstRange = vtk::DataArrayValueRange<1>(intArray);
if (!std::equal(srcRange.cbegin(), srcRange.cend(), dstRange.cbegin(), CloseEnough{}))
{
std::cerr << "Copied values do not match!\n";
return EXIT_FAILURE;
}
}
return EXIT_SUCCESS;
}
|