diff --git a/alib2algo/src/stringology/matching/HammingMatchingAutomaton.cpp b/alib2algo/src/stringology/matching/HammingMatchingAutomaton.cpp
index 2e712d87d7328bf0b87a0c72ab1ce52cfe785403..758aaed89a7c5abcbcfe4ce28d552b846b0fae80 100644
--- a/alib2algo/src/stringology/matching/HammingMatchingAutomaton.cpp
+++ b/alib2algo/src/stringology/matching/HammingMatchingAutomaton.cpp
@@ -12,7 +12,7 @@ namespace stringology {
namespace matching {
-auto HammingMatchingAutomatonLinearString = registration::AbstractRegister <HammingMatchingAutomaton, automaton::NFA < DefaultSymbolType, unsigned >, const string::LinearString < > &, unsigned > ( HammingMatchingAutomaton::construct );
+auto HammingMatchingAutomatonLinearString = registration::AbstractRegister <HammingMatchingAutomaton, automaton::NFA < DefaultSymbolType, ext::pair<unsigned int, unsigned int> >, const string::LinearString < > &, unsigned > ( HammingMatchingAutomaton::construct );
} /* namespace matching */
diff --git a/alib2algo/src/stringology/matching/HammingMatchingAutomaton.h b/alib2algo/src/stringology/matching/HammingMatchingAutomaton.h
index 1e7fbab3ab1ef812c135290e719975e4938ee27c..495856e8e16ec7675169fb6dd68da1fd97ec063e 100644
--- a/alib2algo/src/stringology/matching/HammingMatchingAutomaton.h
+++ b/alib2algo/src/stringology/matching/HammingMatchingAutomaton.h
@@ -5,8 +5,8 @@
* Author: Tomas Capek
*/
-#ifndef _EXACT_MATCHING_AUTOMATON_H__
-#define _EXACT_MATCHING_AUTOMATON_H__
+#ifndef _HAMMING_MATCHING_AUTOMATON_H__
+#define _HAMMING_MATCHING_AUTOMATON_H__
#include <automaton/FSM/NFA.h>
#include <automaton/simplify/UnreachableStatesRemover.h>
@@ -24,53 +24,55 @@ public:
* @return automata for aproximate string matching using Hamming algorithm
*/
template < class SymbolType >
- static automaton::NFA < SymbolType, unsigned > construct(const string::LinearString < SymbolType > & pattern, unsigned int allowed_errors);
+ static automaton::NFA < SymbolType, ext::pair<unsigned int, unsigned int> > construct(const string::LinearString < SymbolType > & pattern, unsigned int allowed_errors);
};
-
template < class SymbolType >
-automaton::NFA < SymbolType, unsigned > HammingMatchingAutomaton::construct(const string::LinearString < SymbolType > & pattern, unsigned int allowed_errors) {
- automaton::NFA < SymbolType, unsigned > res( 0 );
- res.setInputAlphabet(pattern.getAlphabet());
-
+automaton::NFA < SymbolType, ext::pair<unsigned int, unsigned int> > HammingMatchingAutomaton::construct(const string::LinearString < SymbolType > & pattern, unsigned int allowed_errors) {
+ automaton::NFA < SymbolType, ext::pair<unsigned int, unsigned int > > result( ext::make_pair(0, 0) );
+ result.setInputAlphabet(pattern.getAlphabet());
+
// add k+1 paralel automatas (sfoeco type = exact matching) (where k is allowed_errors)
- unsigned current_state = 0;
- for (unsigned i = 0; i <= allowed_errors; i++) {
- if ( current_state > 0 ) {
- ++current_state;
- res.addState(current_state);
+ for (unsigned int i = 0; i<pattern.getContent().size() + 1; i++) {
+ for (unsigned int j = 0; j<allowed_errors + 1; j++) {
+ result.addState(ext::make_pair(i, j));
+ if (i == pattern.getContent().size()) {
+ result.addFinalState(ext::make_pair(i, j));
+ }
}
+ }
- for(const SymbolType& symbol : pattern.getAlphabet()) {
- res.addTransition( current_state, symbol, current_state);
+ for (unsigned int i = 0; i<allowed_errors + 1; i++) {
+ for (const SymbolType& symbol : pattern.getAlphabet()) {
+ auto initial_state = ext::make_pair(0, i);
+ result.addTransition(initial_state, symbol, initial_state);
}
+ }
- for(const SymbolType& symbol : pattern.getContent()) {
- ++current_state;
- res.addState( current_state );
- res.addTransition( current_state - 1, symbol, current_state );
+ for (unsigned int i = 0; i<pattern.getContent().size(); i++) {
+ for (unsigned int j = 0; j < allowed_errors + 1; j++) {
+ auto from = ext::make_pair(i, j);
+ auto to = ext::make_pair(i+1, j);
+ result.addTransition(from, pattern.getContent()[i], to);
}
- res.addFinalState( current_state );
}
- for (unsigned int i = 0; i < pattern.getContent().size(); i++) {
- for (unsigned int j = 0; j < allowed_errors; j++) {
- unsigned int states_count = pattern.getContent().size() + 1;
-
- unsigned int from_state = j*states_count + i;
- unsigned int to_state = (j+1)*states_count + i + 1;
+ // add diagonal addTransition
+ for (unsigned int i = 0; i<pattern.getContent().size(); i++) {
+ for (unsigned int j = 0; j<allowed_errors; j++) {
+ auto from = ext::make_pair(i, j);
+ auto to = ext::make_pair(i + 1, j + 1);
- for ( const SymbolType& symbol : pattern.getAlphabet()) {
+ for ( const SymbolType & symbol : pattern.getAlphabet()) {
if (symbol != pattern.getContent()[i]) {
- // add diagonal addTransition
- res.addTransition(from_state, symbol, to_state);
+ result.addTransition(from, symbol, to);
}
}
}
}
-
+
// remove all inaccessible states from state
- return automaton::simplify::UnreachableStatesRemover::remove(res);
+ return automaton::simplify::UnreachableStatesRemover::remove(result);
}
diff --git a/alib2algo/test-src/stringology/matching/HammingMatchingAutomatonTest.cpp b/alib2algo/test-src/stringology/matching/HammingMatchingAutomatonTest.cpp
index 0a0059cc7813f3828ba2f41b8374873a59894b39..fa005f18463caf674be65bf6be10d7ce91b260f9 100644
--- a/alib2algo/test-src/stringology/matching/HammingMatchingAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/matching/HammingMatchingAutomatonTest.cpp
@@ -1,5 +1,3 @@
-#include <iostream>
-
#include <stringology/matching/HammingMatchingAutomaton.h>
#include <automaton/FSM/NFA.h>
#include <string/LinearString.h>
@@ -7,57 +5,68 @@
#include "HammingMatchingAutomatonTest.h"
+
CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( HammingMatchingAutomatonTest, "stringology" );
CPPUNIT_TEST_SUITE_REGISTRATION ( HammingMatchingAutomatonTest );
-
void HammingMatchingAutomatonTest::testSimpleConstruction() {
- automaton::NFA < char, unsigned > test(0);
-
ext::set<char> alphabet{'a', 'b', 'c', 'd'};
string::LinearString <char> input_string(alphabet, ext::vector<char>{'a', 'b', 'c'});
auto resulting_automata = stringology::matching::HammingMatchingAutomaton::construct(input_string, 2);
- automaton::NFA < char, unsigned > res(0);
+ typedef ext::pair<unsigned int, unsigned int> State;
+
+ automaton::NFA < char, State > res(ext::make_pair(0,0));
res.setInputAlphabet(ext::set<char>{'a', 'b', 'c', 'd'});
- res.setStates(ext::set<unsigned> {0, 1, 2, 3, 5, 6, 7, 10, 11});
- res.setFinalStates(ext::set<unsigned> {3, 7, 11});
-
- res.addTransition(0, 'a', 1);
-
- res.addTransition(1, 'b', 2);
- res.addTransition(5, 'b', 6);
-
- res.addTransition(2, 'c', 3);
- res.addTransition(6, 'c', 7);
- res.addTransition(10, 'c', 11);
-
- res.addTransition(0, 'a', 0);
- res.addTransition(0, 'b', 0);
- res.addTransition(0, 'c', 0);
- res.addTransition(0, 'd', 0);
-
- res.addTransition(0, 'b', 5);
- res.addTransition(0, 'c', 5);
- res.addTransition(0, 'd', 5);
-
- res.addTransition(1, 'a', 6);
- res.addTransition(1, 'c', 6);
- res.addTransition(1, 'd', 6);
- res.addTransition(5, 'a', 10);
- res.addTransition(5, 'c', 10);
- res.addTransition(5, 'd', 10);
-
- res.addTransition(2, 'a', 7);
- res.addTransition(2, 'b', 7);
- res.addTransition(2, 'd', 7);
- res.addTransition(6, 'a', 11);
- res.addTransition(6, 'b', 11);
- res.addTransition(6, 'd', 11);
-
+
+ State q0 = ext::make_pair(0,0);
+ State q1 = ext::make_pair(1,0);
+ State q2 = ext::make_pair(2,0);
+ State q3 = ext::make_pair(3,0);
+ State q4 = ext::make_pair(1,1);
+ State q5 = ext::make_pair(2,1);
+ State q6 = ext::make_pair(3,1);
+ State q7 = ext::make_pair(2,2);
+ State q8 = ext::make_pair(3,2);
+
+ res.setStates(ext::set<State> {q0, q1, q2, q3, q4, q5, q6, q7, q8});
+ res.setFinalStates(ext::set<State> {q3, q6, q8});
+
+ res.addTransition(q0, 'a', q1); // vertical transitions (exact matching automata)
+
+ res.addTransition(q1, 'b', q2);
+ res.addTransition(q4, 'b', q5);
+
+ res.addTransition(q2, 'c', q3);
+ res.addTransition(q5, 'c', q6);
+ res.addTransition(q7, 'c', q8);
+
+ res.addTransition(q0, 'a', q0); // loops in initial state
+ res.addTransition(q0, 'b', q0);
+ res.addTransition(q0, 'c', q0);
+ res.addTransition(q0, 'd', q0);
+
+ res.addTransition(q0, 'b', q4); // diagonal transitions representing replace
+ res.addTransition(q0, 'c', q4);
+ res.addTransition(q0, 'd', q4);
+
+ res.addTransition(q1, 'a', q5);
+ res.addTransition(q1, 'c', q5);
+ res.addTransition(q1, 'd', q5);
+ res.addTransition(q4, 'a', q7);
+ res.addTransition(q4, 'c', q7);
+ res.addTransition(q4, 'd', q7);
+
+ res.addTransition(q2, 'a', q6);
+ res.addTransition(q2, 'b', q6);
+ res.addTransition(q2, 'd', q6);
+ res.addTransition(q5, 'a', q8);
+ res.addTransition(q5, 'b', q8);
+ res.addTransition(q5, 'd', q8);
+
CPPUNIT_ASSERT(resulting_automata == res);
}
void HammingMatchingAutomatonTest::setUp() { }
-void HammingMatchingAutomatonTest::tearDown() { }
\ No newline at end of file
+void HammingMatchingAutomatonTest::tearDown() { }