GeneralizedLevenshteinDynamicProgramming.h

Go to the documentation of this file.

 
#pragma once
 
#include <algorithm>
#include <limits.h>
 
#include <string/LinearString.h>
 
namespace stringology {
 
namespace simulations {
 
class GeneralizedLevenshteinDynamicProgramming {
public:
    template <class SymbolType>
    static ext::vector<ext::vector<unsigned int>> compute_table(const string::LinearString<SymbolType> & text, const string::LinearString<SymbolType> & pattern);
 
    template <class SymbolType>
    static ext::set<unsigned int> search(const string::LinearString<SymbolType> & text, const string::LinearString<SymbolType> & pattern, unsigned int errors);
};
 
template <class SymbolType>
ext::vector<ext::vector<unsigned int>> GeneralizedLevenshteinDynamicProgramming::compute_table(const string::LinearString<SymbolType> & text, const string::LinearString<SymbolType> & pattern) {
  ext::vector< ext::vector <unsigned int> > table =
    ext::vector<ext::vector<unsigned int> > (
      pattern.getContent().size() + 1,
      ext::vector<unsigned int>(text.getContent().size() + 1, 0)
    );
 
  for(unsigned int j = 0; j <= pattern.getContent().size(); j++) {
    table[j][0] = j;
  }
 
  for(unsigned int i = 1; i<=text.getContent().size(); i++) {
    for(unsigned int j = 1; j<=pattern.getContent().size(); j++) {
      unsigned int value_a;
      if(pattern.getContent()[j-1] == text.getContent()[i-1]) {
        value_a = table[j-1][i-1];
      } else {
        value_a = table[j-1][i-1] + 1;
      }
 
      unsigned int value_b = UINT_MAX;
      if(j < pattern.getContent().size()) {
        value_b = table[j][i-1] + 1;
      }
 
      value_b = std::min(table[j-1][i] + 1, value_b);
 
      unsigned int value_c = UINT_MAX;
      if(j>1 && i>1 && pattern.getContent()[j-2] == text.getContent()[i-1] && pattern.getContent()[j-1] == text.getContent()[i-2]) {
        value_c = table[j-2][i-2] + 1;
      }
 
      table[j][i] = std::min({value_a, value_b, value_c});
    }
  }
 
  return table;
}
 
template <class SymbolType>
ext::set<unsigned int> GeneralizedLevenshteinDynamicProgramming::search(const string::LinearString<SymbolType> & text, const string::LinearString<SymbolType> & pattern, unsigned int errors) {
  auto table = GeneralizedLevenshteinDynamicProgramming::compute_table(text, pattern);
 
  ext::set<unsigned int> result;
 
  for(unsigned int i = 0; i<= text.getContent().size(); i++) {
    if(table[pattern.getContent().size()][i] <= errors) {
      result.insert(i);
    }
  }
 
  return result;
}
 
 
} // namespace simulations
 
} // namespace stringology