From 03443ad1722d0b25ed2696d8a6b10498b9482588 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1=C5=A1=20=C4=8Capek?= <tomas@capek.io>
Date: Wed, 18 Apr 2018 19:00:04 +0200
Subject: [PATCH] Implement WildcardLinearString into
LevenshteinSequenceMatchingAutomaton
---
.../LevenshteinSequenceMatchingAutomaton.h | 37 ++++++++++-
...venshteinSequenceMatchingAutomatonTest.cpp | 66 +++++++++++++++++++
...LevenshteinSequenceMatchingAutomatonTest.h | 6 +-
3 files changed, 106 insertions(+), 3 deletions(-)
diff --git a/alib2algo/src/stringology/matching/LevenshteinSequenceMatchingAutomaton.h b/alib2algo/src/stringology/matching/LevenshteinSequenceMatchingAutomaton.h
index 3c241f82cb..3ec7a7fee7 100644
--- a/alib2algo/src/stringology/matching/LevenshteinSequenceMatchingAutomaton.h
+++ b/alib2algo/src/stringology/matching/LevenshteinSequenceMatchingAutomaton.h
@@ -11,6 +11,7 @@
#include <automaton/FSM/EpsilonNFA.h>
#include <stringology/matching/LevenshteinMatchingAutomaton.h>
#include <string/LinearString.h>
+#include <string/WildcardLinearString.h>
namespace stringology {
@@ -20,12 +21,21 @@ namespace matching {
class LevenshteinSequenceMatchingAutomaton {
public:
/**
- * Creates Levenshtein matching automata for sequence matching.
+ * Creates Levenshtein matching automata for sequence matching from LinearString.
*
* @return automata for aproximate sequence matching using Levenshtein method.
*/
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 for sequence matching from WildcardLinearString.
+ *
+ * @return automata for aproximate sequence matching using Levenshtein method.
+ */
+ template < class SymbolType >
+ static automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int> > construct(const string::WildcardLinearString < SymbolType > & pattern, unsigned int allowed_errors);
+
};
template < class SymbolType >
@@ -47,6 +57,31 @@ automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int>
return result;
}
+template < class SymbolType >
+automaton::EpsilonNFA < SymbolType, void, ext::pair<unsigned int, unsigned int> > LevenshteinSequenceMatchingAutomaton::construct(const string::WildcardLinearString < SymbolType > & pattern, unsigned int allowed_errors) {
+ auto result = stringology::matching::LevenshteinMatchingAutomaton::construct(pattern, allowed_errors);
+
+ SymbolType wildcard = pattern.getWildcardSymbol();
+ ext::set<SymbolType> alphabet_without_wildcard = pattern.getAlphabet();
+ alphabet_without_wildcard.erase(wildcard);
+
+ for (unsigned int j = 0; j<allowed_errors + 1; j++) {
+ for (unsigned int i = j; i<pattern.getContent().size(); i++) {
+ auto current_state = ext::make_pair(i, j);
+
+ if (pattern.getContent()[i] != wildcard) {
+ for (const SymbolType & symbol : alphabet_without_wildcard) {
+ if (symbol != pattern.getContent()[i]) {
+ result.addTransition(current_state, symbol, current_state);
+ }
+ }
+ }
+ }
+ }
+
+ return result;
+}
+
} /* namespace matching */
diff --git a/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.cpp b/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.cpp
index 0515fb0a65..939e431e60 100644
--- a/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.cpp
@@ -106,6 +106,72 @@ void LevenshteinSequenceMatchingAutomatonTest::testSimpleConstruction() {
CPPUNIT_ASSERT(resulting_automata == test);
}
+
+void LevenshteinSequenceMatchingAutomatonTest::testSimpleWildcardConstruction() {
+ ext::set<char> alphabet{'a', 'b', '@'};
+ string::WildcardLinearString <char> input_string(alphabet, ext::vector<char>{'a', '@', 'b'}, '@');
+ auto resulting_automata = stringology::matching::LevenshteinSequenceMatchingAutomaton::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);
+
+ res.addTransition(q2, 'a', q2); // sequence matching loops
+ res.addTransition(q5, 'a', q5);
+ res.addTransition(q7, 'a', q7);
+
+ CPPUNIT_ASSERT(resulting_automata == res);
+}
+
void LevenshteinSequenceMatchingAutomatonTest::setUp() { }
void LevenshteinSequenceMatchingAutomatonTest::tearDown() { }
diff --git a/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.h b/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.h
index 7e85e84501..a08b8c64ab 100644
--- a/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.h
+++ b/alib2algo/test-src/stringology/matching/LevenshteinSequenceMatchingAutomatonTest.h
@@ -5,13 +5,15 @@
class LevenshteinSequenceMatchingAutomatonTest : public CppUnit::TestFixture {
CPPUNIT_TEST_SUITE(LevenshteinSequenceMatchingAutomatonTest);
- 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 //LEVENSHTEIN_SEQUENCE_MATCHING_AUTOMATA_TEST_H_
--
GitLab