C++ Utilities  5.10.5
Useful C++ classes and routines such as argument parser, IO and conversion utilities
levenshtein.cpp
Go to the documentation of this file.
1 #include "./levenshtein.h"
2 
3 #include "./math.h"
4 #include "./multiarray.h"
5 
6 #include <iostream>
7 #include <limits>
8 #include <memory>
9 
10 using namespace std;
11 
12 namespace CppUtilities {
13 
15 
17 using DistanceArray = MultiArray<size_t, NoneOwningMultiArray, size_t, size_t>;
18 
31 void initDistanceArray(DistanceArray &distanceArray, const size_t size1, const size_t size2)
32 {
33  const auto maxDistance(size1 + size2);
34  distanceArray.at(0, 0) = maxDistance;
35  for (size_t i = 0; i <= size1; ++i) {
36  distanceArray.at(i + 1, 1) = i;
37  distanceArray.at(i + 1, 0) = maxDistance;
38  }
39  for (size_t i = 0; i <= size2; ++i) {
40  distanceArray.at(1, i + 1) = i;
41  distanceArray.at(0, i + 1) = maxDistance;
42  }
43 }
44 
47 size_t performDamerauLevenshteinAlgorithm(
48  DistanceArray &distanceArray, const char *const str1, const size_t size1, const char *const str2, const size_t size2)
49 {
50  size_t dist1[std::numeric_limits<unsigned char>::max() + 1] = { 0 };
51  for (size_t index1 = 1; index1 <= size1; ++index1) {
52  size_t dist2 = 0;
53  for (size_t index2 = 1; index2 <= size2; ++index2) {
54  const size_t substitution((str1[index1 - 1] == str2[index2 - 1]) ? 0 : 1);
55  const size_t transposition1(dist1[static_cast<unsigned char>(str2[index2 - 1])]);
56  const size_t transposition2(dist2);
57  if (!substitution) {
58  dist2 = index2;
59  }
60  // clang-format off
61  distanceArray.at(index1 + 1, index2 + 1) = CppUtilities::min(
62  distanceArray.at(index1, index2) + substitution, // substitution
63  distanceArray.at(index1 + 1, index2) + 1, // insertion
64  distanceArray.at(index1, index2 + 1) + 1, // deletion
65  distanceArray.at(transposition1, transposition2) + (index1 - transposition1 - 1) + 1 + (index2 - transposition2 - 1) // transposition
66  );
67  // clang-format on
68  }
69  dist1[static_cast<int>(str1[index1 - 1])] = index1;
70  }
71  return distanceArray.at(size1 + 1, size2 + 1);
72 }
73 
75 template <typename DistanceArray>
76 size_t performDamerauLevenshteinAlgorithmAllocatingOnHeap(
77  DistanceArray &distanceArray, const char *const str1, const size_t size1, const char *const str2, const size_t size2)
78 {
79  std::vector<size_t> buffer(distanceArray.totalSize());
80  distanceArray.buffer() = buffer.data();
81  initDistanceArray(distanceArray, size1, size2);
82  return performDamerauLevenshteinAlgorithm(distanceArray, str1, size1, str2, size2);
83 }
84 
87 template <typename DistanceArray>
88 size_t performDamerauLevenshteinAlgorithmAllocatingOnStack(
89  DistanceArray &distanceArray, const char *const str1, const size_t size1, const char *const str2, const size_t size2)
90 {
91  size_t buffer[128] = { 0 };
92  distanceArray.buffer() = buffer;
93  initDistanceArray(distanceArray, size1, size2);
94  return performDamerauLevenshteinAlgorithm(distanceArray, str1, size1, str2, size2);
95 }
96 
98 
117 std::size_t computeDamerauLevenshteinDistance(const char *const str1, const size_t size1, const char *const str2, const size_t size2)
118 {
119  // allocate distance array
120  auto distanceArray(makeNoneOwningMultiArray<std::size_t>(size1 + 2, size2 + 2));
121  if (distanceArray.totalSize() <= 128) {
122  return performDamerauLevenshteinAlgorithmAllocatingOnStack(distanceArray, str1, size1, str2, size2);
123  } else {
124  return performDamerauLevenshteinAlgorithmAllocatingOnHeap(distanceArray, str1, size1, str2, size2);
125  }
126 }
127 
128 } // namespace CppUtilities
Contains all utilities provides by the c++utilities library.
std::size_t computeDamerauLevenshteinDistance(const char *const str1, const size_t size1, const char *const str2, const size_t size2)
Computes Damerau–Levenshtein distance with adjacent transpositions.
constexpr T max(T first, T second)
Returns the greatest of the given items.
Definition: math.h:100
constexpr T min(T first, T second)
Returns the smallest of the given items.
Definition: math.h:88
constexpr int i