From 700f84166172412387e7273a29c924017ccfd78a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1=C5=A1=20=C4=8Capek?= <tomas@capek.io>
Date: Sun, 15 Apr 2018 10:48:20 +0200
Subject: [PATCH] Implement WildcardLinearString into
LevenshteinMatchingAutomaton
---
.../matching/LevenshteinMatchingAutomaton.h | 47 +++++++++++++-
.../LevenshteinMatchingAutomatonTest.cpp | 61 +++++++++++++++++++
.../LevenshteinMatchingAutomatonTest.h | 6 +-
3 files changed, 111 insertions(+), 3 deletions(-)
diff --git a/alib2algo/src/stringology/matching/LevenshteinMatchingAutomaton.h b/alib2algo/src/stringology/matching/LevenshteinMatchingAutomaton.h
index 38f3add4a8..add20ca732 100644
--- a/alib2algo/src/stringology/matching/LevenshteinMatchingAutomaton.h
+++ b/alib2algo/src/stringology/matching/LevenshteinMatchingAutomaton.h
@@ -9,7 +9,9 @@
#define _LEVENSHTEIN_MATCHING_AUTOMATON_H__
#include <automaton/FSM/EpsilonNFA.h>
+#include <automaton/simplify/UnreachableStatesRemover.h>
#include <string/LinearString.h>
+#include <string/WildcardLinearString.h>
#include <stringology/matching/HammingMatchingAutomaton.h>
@@ -20,12 +22,20 @@ namespace matching {
class LevenshteinMatchingAutomaton {
public:
/**
- * Creates Levenshtein matching automata.
+ * Creates Levenshtein matching automata form LinearString.
*
* @return automata for aproximate string matching using Hamming algorithm
*/
template < class SymbolType >
static automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int> > construct(const string::LinearString < SymbolType > & pattern, unsigned int allowed_errors);
+
+ /**
+ * Creates Levenshtein matching automata from WildcardLinearString
+ *
+ * @return automata for aproximate string matching using Hamming algorithm
+ */
+ template < class SymbolType >
+ static automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int> > construct(const string::WildcardLinearString < SymbolType > & pattern, unsigned int allowed_errors);
};
@@ -62,6 +72,41 @@ automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int>
}
+template < class SymbolType >
+automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int> > LevenshteinMatchingAutomaton::construct(const string::WildcardLinearString < SymbolType > & pattern, unsigned int allowed_errors) {
+ auto hamming_matching_automaton = stringology::matching::HammingMatchingAutomaton::construct_unclean(pattern, allowed_errors);
+
+ automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int> > result (hamming_matching_automaton);
+
+ ext::set<SymbolType> alphabet_without_wildcard = pattern.getAlphabet();
+ alphabet_without_wildcard.erase(pattern.getWildcardSymbol());
+
+ for (unsigned int j = 0; j<allowed_errors; j++) {
+ for (unsigned int i = j; i<pattern.getContent().size(); i++) {
+ auto from = ext::make_pair(i, j);
+ auto to = ext::make_pair(i + 1, j + 1);
+
+ // add diagonal transition representing deletion
+ result.addTransition(from, to);
+
+ if (i == j) {
+ continue;
+ }
+
+ to = ext::make_pair(i, j + 1);
+
+ for (const SymbolType& symbol : alphabet_without_wildcard) {
+ // add horizontal transition representing insertion
+ result.addTransition(from, symbol, to);
+ }
+ }
+ }
+
+ return automaton::simplify::UnreachableStatesRemover::remove(result);
+}
+
+
+
} /* namespace matching */
} /* namespace stringology */
diff --git a/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.cpp b/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.cpp
index 4f9ab10ac9..301b53d142 100644
--- a/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.cpp
@@ -87,6 +87,67 @@ void LevenshteinMatchingAutomatonTest::testSimpleConstruction() {
CPPUNIT_ASSERT(resulting_automata == res);
}
+void LevenshteinMatchingAutomatonTest::testSimpleWildcardConstruction() {
+ ext::set<char> alphabet{'a', 'b', '@'};
+ string::WildcardLinearString <char> input_string(alphabet, ext::vector<char>{'a', '@', 'b'}, '@');
+ auto resulting_automata = stringology::matching::LevenshteinMatchingAutomaton::construct(input_string, 2);
+
+ typedef ext::pair<unsigned int, unsigned int> State;
+
+ automaton::EpsilonNFA < char, void, State > res(ext::make_pair(0,0));
+ res.setInputAlphabet(alphabet);
+
+ 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', q0); // initial loops
+ res.addTransition(q0, 'b', q0);
+
+ res.addTransition(q0, 'a', q1); // 3 simple matching automatas
+
+ res.addTransition(q1, 'a', q2);
+ res.addTransition(q1, 'b', q2);
+ res.addTransition(q4, 'a', q5);
+ res.addTransition(q4, 'b', q5);
+
+ res.addTransition(q2, 'b', q3);
+ res.addTransition(q5, 'b', q6);
+ res.addTransition(q7, 'b', q8);
+
+ res.addTransition(q0, 'b', q4); // error transitions for replace
+
+ res.addTransition(q2, 'a', q6);
+ res.addTransition(q5, 'a', q8);
+
+ res.addTransition(q0, q4); // delete transition
+ res.addTransition(q1, q5);
+ res.addTransition(q2, q6);
+
+ res.addTransition(q4, q7);
+ res.addTransition(q5, q8);
+
+ res.addTransition(q1, 'a', q4);
+ res.addTransition(q1, 'b', q4);
+
+ res.addTransition(q2, 'a', q5);
+ res.addTransition(q2, 'b', q5);
+
+ res.addTransition(q5, 'a', q7);
+ res.addTransition(q5, 'b', q7);
+
+ CPPUNIT_ASSERT(resulting_automata == res);
+}
+
void LevenshteinMatchingAutomatonTest::setUp() { }
void LevenshteinMatchingAutomatonTest::tearDown() { }
diff --git a/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.h b/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.h
index ab650ce568..6ec3013fa8 100644
--- a/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.h
+++ b/alib2algo/test-src/stringology/matching/LevenshteinMatchingAutomatonTest.h
@@ -5,13 +5,15 @@
class LevenshteinMatchingAutomatonTest : public CppUnit::TestFixture {
CPPUNIT_TEST_SUITE(LevenshteinMatchingAutomatonTest);
- CPPUNIT_TEST(testSimpleConstruction);
- CPPUNIT_TEST_SUITE_END();
+ CPPUNIT_TEST(testSimpleConstruction);
+ CPPUNIT_TEST(testSimpleWildcardConstruction);
+ CPPUNIT_TEST_SUITE_END();
public:
void setUp ( );
void tearDown ( );
void testSimpleConstruction();
+ void testSimpleWildcardConstruction();
};
#endif //HAMMING_MATCHING_AUTOMATA_TEST_H_
--
GitLab