From a2c1733702666053a8f61635fda121a113f41b36 Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Wed, 17 May 2017 15:41:02 +0200
Subject: [PATCH] move dawg and oracle matching to algo
---
.../query/BackwardDAWGMatching.cpp | 69 +++++++++++++++++
.../stringology/query}/BackwardDAWGMatching.h | 20 ++---
.../query/BackwardOracleMatching.cpp | 65 ++++++++++++++++
.../query}/BackwardOracleMatching.h | 20 ++---
.../stringology/query/DAWGMatcherTest.cpp | 55 ++++++++++++++
.../stringology/query}/DAWGMatcherTest.h | 0
.../stringology/query/OracleMatcherTest.cpp | 53 +++++++++++++
.../stringology/query/OracleMatcherTest.h | 19 +++++
.../exact/BackwardDAWGMatching.cpp | 76 -------------------
.../exact/BackwardOracleMatching.cpp | 72 ------------------
.../stringology/exact/DAWGMatcherTest.cpp | 52 -------------
.../stringology/exact/OracleMatcherTest.cpp | 50 ------------
.../stringology/exact/OracleMatcherTest.h | 19 -----
astringology2/src/astringology.cpp | 12 +--
14 files changed, 289 insertions(+), 293 deletions(-)
create mode 100644 alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
rename {alib2algo_experimental/src/stringology/exact => alib2algo/src/stringology/query}/BackwardDAWGMatching.h (53%)
create mode 100644 alib2algo/src/stringology/query/BackwardOracleMatching.cpp
rename {alib2algo_experimental/src/stringology/exact => alib2algo/src/stringology/query}/BackwardOracleMatching.h (53%)
create mode 100644 alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp
rename {alib2algo_experimental/test-src/stringology/exact => alib2algo/test-src/stringology/query}/DAWGMatcherTest.h (100%)
create mode 100644 alib2algo/test-src/stringology/query/OracleMatcherTest.cpp
create mode 100644 alib2algo/test-src/stringology/query/OracleMatcherTest.h
delete mode 100644 alib2algo_experimental/src/stringology/exact/BackwardDAWGMatching.cpp
delete mode 100644 alib2algo_experimental/src/stringology/exact/BackwardOracleMatching.cpp
delete mode 100644 alib2algo_experimental/test-src/stringology/exact/DAWGMatcherTest.cpp
delete mode 100644 alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.cpp
delete mode 100644 alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.h
diff --git a/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp b/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
new file mode 100644
index 0000000000..e93db0dd56
--- /dev/null
+++ b/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
@@ -0,0 +1,69 @@
+/*
+ * Author: Radovan Cerveny
+ */
+
+#include "BackwardDAWGMatching.h"
+#include "stringology/matching/DAWGMatcherConstruction.h"
+#include "stringology/properties/BackboneLength.h"
+
+#include <string/LinearString.h>
+#include <alphabet/Symbol.h>
+
+#include <algorithm>
+#include <map>
+#include <measure>
+
+namespace stringology {
+
+namespace query {
+
+std::set < unsigned > BackwardDAWGMatching::match ( const string::String & subject, const automaton::Automaton & suffixAutomaton ) {
+ return dispatch ( subject.getData ( ), suffixAutomaton.getData ( ) );
+}
+
+std::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString < > & subject, const automaton::DFA < > & suffixAutomaton ) {
+ std::set < unsigned > occ;
+
+ size_t patternSize = stringology::properties::BackboneLength::length ( suffixAutomaton );
+ size_t subjectSize = subject.getContent ( ).size ( );
+
+ const DefaultStateType failState = DefaultStateType ( -1 );
+
+ size_t posInSubject = 0;
+
+ while ( posInSubject <= subjectSize - patternSize ) {
+ DefaultStateType currentState = suffixAutomaton.getInitialState ( );
+
+ size_t posInPattern = patternSize;
+ size_t lastPrefixPos = posInPattern;
+
+ while ( posInPattern > 0 && currentState != failState ) {
+ auto transition = suffixAutomaton.getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
+
+ if ( transition == suffixAutomaton.getTransitions ( ).end ( ) )
+ currentState = failState;
+ else
+ currentState = transition->second;
+
+ posInPattern--;
+
+ // found a prefix of nonreversed pattern that does not correspond to the entire pattern
+ if ( ( posInPattern != 0 ) && ( suffixAutomaton.getFinalStates ( ).find ( currentState ) != suffixAutomaton.getFinalStates ( ).end ( ) ) )
+ lastPrefixPos = posInPattern;
+ }
+
+ if ( currentState != failState )
+ // Yay, there is match!!!
+ occ.insert ( posInSubject );
+
+ posInSubject += lastPrefixPos;
+ }
+
+ return occ;
+}
+
+auto BackwardDAWGMatchingLinearStringLinearString = BackwardDAWGMatching::RegistratorWrapper < std::set < unsigned >, string::LinearString < >, automaton::DFA < > > ( BackwardDAWGMatching::match );
+
+} /* namespace query */
+
+} /* namespace stringology */
diff --git a/alib2algo_experimental/src/stringology/exact/BackwardDAWGMatching.h b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
similarity index 53%
rename from alib2algo_experimental/src/stringology/exact/BackwardDAWGMatching.h
rename to alib2algo/src/stringology/query/BackwardDAWGMatching.h
index a1aac21748..9b2bc62846 100644
--- a/alib2algo_experimental/src/stringology/exact/BackwardDAWGMatching.h
+++ b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
@@ -7,29 +7,31 @@
#include <string/String.h>
#include <string/StringFeatures.h>
+#include <automaton/Automaton.h>
+#include <automaton/AutomatonFeatures.h>
#include <core/multipleDispatch.hpp>
#include <set>
namespace stringology {
-namespace exact {
+namespace query {
/**
* Implementation of Backward DAWG Matching.
*/
-class BackwardDAWGMatching : public std::DoubleDispatch < BackwardDAWGMatching, std::set < unsigned >, const string::StringBase &, const string::StringBase & > {
+class BackwardDAWGMatching : public std::DoubleDispatch < BackwardDAWGMatching, std::set < unsigned >, const string::StringBase &, const automaton::AutomatonBase & > {
public:
- /**
- * Search for pattern in linear string.
- * @return set set of occurences
- */
- static std::set < unsigned > match ( const string::String & subject, const string::String & pattern );
- static std::set < unsigned > match ( const string::LinearString < > & subject, const string::LinearString < > & pattern );
+ /**
+ * Search for pattern in linear string.
+ * @return set set of occurences
+ */
+ static std::set < unsigned > match ( const string::String & subject, const automaton::Automaton & suffixAutomaton );
+ static std::set < unsigned > match ( const string::LinearString < > & subject, const automaton::DFA < > & suffixAutomaton );
};
-} /* namespace exact */
+} /* namespace query */
} /* namespace stringology */
diff --git a/alib2algo/src/stringology/query/BackwardOracleMatching.cpp b/alib2algo/src/stringology/query/BackwardOracleMatching.cpp
new file mode 100644
index 0000000000..57da6ea3b5
--- /dev/null
+++ b/alib2algo/src/stringology/query/BackwardOracleMatching.cpp
@@ -0,0 +1,65 @@
+/*
+ * Author: Radovan Cerveny
+ */
+
+#include "BackwardOracleMatching.h"
+#include "stringology/matching/OracleMatcherConstruction.h"
+#include "stringology/properties/BackboneLength.h"
+
+#include <string/LinearString.h>
+#include <alphabet/Symbol.h>
+
+#include <algorithm>
+#include <map>
+#include <measure>
+
+namespace stringology {
+
+namespace query {
+
+std::set < unsigned > BackwardOracleMatching::match ( const string::String & subject, const automaton::Automaton & factorOracle ) {
+ return dispatch ( subject.getData ( ), factorOracle.getData ( ) );
+}
+
+std::set < unsigned > BackwardOracleMatching::match ( const string::LinearString < > & subject, const automaton::DFA < > & factorOracle ) {
+ std::set < unsigned > occ;
+
+ size_t patternSize = stringology::properties::BackboneLength::length ( factorOracle );
+ size_t subjectSize = subject.getContent ( ).size ( );
+
+ const DefaultStateType failState = DefaultStateType ( -1 );
+
+ size_t posInSubject = 0;
+
+ while ( posInSubject <= subjectSize - patternSize ) {
+
+ DefaultStateType currentState = factorOracle.getInitialState ( );
+
+ size_t posInPattern = patternSize;
+
+ while ( posInPattern > 0 && currentState != failState ) {
+ auto transition = factorOracle.getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
+
+ if ( transition == factorOracle.getTransitions ( ).end ( ) )
+ currentState = failState;
+ else
+ currentState = transition->second;
+
+ posInPattern--;
+ }
+
+ if ( currentState != failState )
+ // Yay, there is match!!!
+ occ.insert ( posInSubject );
+
+ posInSubject += posInPattern + 1;
+ }
+
+ return occ;
+}
+
+auto BackwardOracleMatchingLinearStringLinearString = BackwardOracleMatching::RegistratorWrapper < std::set < unsigned >, string::LinearString < >, automaton::DFA < > > ( BackwardOracleMatching::match );
+
+} /* namespace query */
+
+} /* namespace stringology */
diff --git a/alib2algo_experimental/src/stringology/exact/BackwardOracleMatching.h b/alib2algo/src/stringology/query/BackwardOracleMatching.h
similarity index 53%
rename from alib2algo_experimental/src/stringology/exact/BackwardOracleMatching.h
rename to alib2algo/src/stringology/query/BackwardOracleMatching.h
index 6f2320fc12..c6df8549c6 100644
--- a/alib2algo_experimental/src/stringology/exact/BackwardOracleMatching.h
+++ b/alib2algo/src/stringology/query/BackwardOracleMatching.h
@@ -7,29 +7,31 @@
#include <string/String.h>
#include <string/StringFeatures.h>
+#include <automaton/Automaton.h>
+#include <automaton/AutomatonFeatures.h>
#include <core/multipleDispatch.hpp>
#include <set>
namespace stringology {
-namespace exact {
+namespace query {
/**
* Implementation of Backward Oracle Matching.
*/
-class BackwardOracleMatching : public std::DoubleDispatch <BackwardOracleMatching, std::set < unsigned >, const string::StringBase &, const string::StringBase & > {
+class BackwardOracleMatching : public std::DoubleDispatch <BackwardOracleMatching, std::set < unsigned >, const string::StringBase &, const automaton::AutomatonBase & > {
public:
- /**
- * Search for pattern in linear string.
- * @return set set of occurences
- */
- static std::set < unsigned > match ( const string::String & subject, const string::String & pattern );
- static std::set < unsigned > match ( const string::LinearString < > & subject, const string::LinearString < > & pattern );
+ /**
+ * Search for pattern in linear string.
+ * @return set set of occurences
+ */
+ static std::set < unsigned > match ( const string::String & subject, const automaton::Automaton & factorOracle );
+ static std::set < unsigned > match ( const string::LinearString < > & subject, const automaton::DFA < > & factorOracle );
};
-} /* namespace exact */
+} /* namespace query */
} /* namespace stringology */
diff --git a/alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp b/alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp
new file mode 100644
index 0000000000..0a724ab8b0
--- /dev/null
+++ b/alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp
@@ -0,0 +1,55 @@
+#include "DAWGMatcherTest.h"
+
+#include "string/LinearString.h"
+#include "stringology/matching/DAWGMatcherConstruction.h"
+#include "stringology/query/BackwardDAWGMatching.h"
+#include "stringology/exact/ExactFactorMatch.h"
+
+#include "string/generate/RandomStringFactory.h"
+#include "string/generate/RandomSubstringFactory.h"
+
+CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( DAWGMatcherTest, "stringology" );
+CPPUNIT_TEST_SUITE_REGISTRATION ( DAWGMatcherTest );
+
+void DAWGMatcherTest::setUp ( ) {
+}
+
+void DAWGMatcherTest::tearDown ( ) {
+}
+
+void DAWGMatcherTest::testBackwardDAWGMatching ( ) {
+ std::vector<std::string> subjects;
+ std::vector<std::string> patterns;
+ std::vector<std::set<unsigned>> expectedOccs;
+
+ subjects.push_back("a"); patterns.push_back("a"); expectedOccs.push_back({0});
+ subjects.push_back("a"); patterns.push_back("b"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfa"); expectedOccs.push_back({0});
+ subjects.push_back("alfalfalfa"); patterns.push_back("blfalfalfa"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfb"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({0});
+ subjects.push_back("alfalfalfaalfalfalfaabfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); expectedOccs.push_back({0});
+ subjects.push_back("atggccttgcc"); patterns.push_back("gcc"); expectedOccs.push_back({3,8});
+ subjects.push_back("aaaaaaaaaa"); patterns.push_back("a"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8,9});
+ subjects.push_back("aaaaaaaaaa"); patterns.push_back("aa"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8});
+
+
+ for(size_t i = 0; i < subjects.size(); ++i) {
+ string::String subject = string::stringFrom ( subjects[i] );
+ string::String pattern = string::stringFrom ( patterns[i] );
+ automaton::Automaton suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( pattern );
+ std::set < unsigned > res = stringology::query::BackwardDAWGMatching::match ( subject, suffixAutomaton );
+ std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
+ CPPUNIT_ASSERT ( res == expectedOccs[i] );
+ }
+
+ auto longSubject = string::generate::RandomStringFactory::generateLinearString (64 * 64, 512, false, true);
+ auto longPattern = string::generate::RandomSubstringFactory::generateSubstring(64 * 5, longSubject );
+ automaton::DFA < > suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( longPattern );
+ std::set < unsigned > res = stringology::query::BackwardDAWGMatching::match ( longSubject, suffixAutomaton );
+ std::set < unsigned > resRef = stringology::exact::ExactFactorMatch::match ( longSubject, longPattern );
+ std::cout << "long: " << res << std::endl;
+ CPPUNIT_ASSERT ( res == resRef);
+}
+
diff --git a/alib2algo_experimental/test-src/stringology/exact/DAWGMatcherTest.h b/alib2algo/test-src/stringology/query/DAWGMatcherTest.h
similarity index 100%
rename from alib2algo_experimental/test-src/stringology/exact/DAWGMatcherTest.h
rename to alib2algo/test-src/stringology/query/DAWGMatcherTest.h
diff --git a/alib2algo/test-src/stringology/query/OracleMatcherTest.cpp b/alib2algo/test-src/stringology/query/OracleMatcherTest.cpp
new file mode 100644
index 0000000000..e537c3a2fd
--- /dev/null
+++ b/alib2algo/test-src/stringology/query/OracleMatcherTest.cpp
@@ -0,0 +1,53 @@
+#include "OracleMatcherTest.h"
+
+#include "string/LinearString.h"
+#include "stringology/matching/OracleMatcherConstruction.h"
+#include "stringology/query/BackwardOracleMatching.h"
+#include "stringology/exact/ExactFactorMatch.h"
+
+#include "string/generate/RandomStringFactory.h"
+#include "string/generate/RandomSubstringFactory.h"
+
+CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( OracleMatcherTest, "stringology" );
+CPPUNIT_TEST_SUITE_REGISTRATION ( OracleMatcherTest );
+
+void OracleMatcherTest::setUp ( ) {
+}
+
+void OracleMatcherTest::tearDown ( ) {
+}
+
+void OracleMatcherTest::testBackwardOracleMatching ( ) {
+ std::vector<std::string> subjects;
+ std::vector<std::string> patterns;
+ std::vector<std::set<unsigned>> expectedOccs;
+
+ subjects.push_back("a"); patterns.push_back("a"); expectedOccs.push_back({0});
+ subjects.push_back("a"); patterns.push_back("b"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfa"); expectedOccs.push_back({0});
+ subjects.push_back("alfalfalfa"); patterns.push_back("blfalfalfa"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfb"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({0});
+ subjects.push_back("alfalfalfaalfalfalfaabfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({});
+ subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); expectedOccs.push_back({0});
+ subjects.push_back("atggccttgcc"); patterns.push_back("gcc"); expectedOccs.push_back({3,8});
+ subjects.push_back("aaaaaaaaaa"); patterns.push_back("a"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8,9});
+
+
+ for(size_t i = 0; i < subjects.size(); ++i) {
+ string::String subject = string::stringFrom ( subjects[i] );
+ string::String pattern = string::stringFrom ( patterns[i] );
+ automaton::Automaton oracleAutomaton = stringology::matching::OracleMatcherConstruction::construct ( pattern );
+ std::set < unsigned > res = stringology::query::BackwardOracleMatching::match ( subject, oracleAutomaton );
+ std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
+ CPPUNIT_ASSERT ( res == expectedOccs[i] );
+ }
+
+ auto longSubject = string::generate::RandomStringFactory::generateLinearString (64 * 64, 512, false, true);
+ auto longPattern = string::generate::RandomSubstringFactory::generateSubstring(64 * 5, longSubject );
+ automaton::DFA < > oracleAutomaton = stringology::matching::OracleMatcherConstruction::construct ( longPattern );
+ std::set < unsigned > res = stringology::query::BackwardOracleMatching::match ( longSubject, oracleAutomaton );
+ std::set < unsigned > resRef = stringology::exact::ExactFactorMatch::match ( longSubject, longPattern );
+ std::cout << "long: " << res << std::endl;
+ CPPUNIT_ASSERT ( res == resRef);
+}
diff --git a/alib2algo/test-src/stringology/query/OracleMatcherTest.h b/alib2algo/test-src/stringology/query/OracleMatcherTest.h
new file mode 100644
index 0000000000..886fc99358
--- /dev/null
+++ b/alib2algo/test-src/stringology/query/OracleMatcherTest.h
@@ -0,0 +1,19 @@
+#ifndef ORACLE_MATCHER_TEST_H_
+#define ORACLE_MATCHER_TEST_H_
+
+#include <cppunit/extensions/HelperMacros.h>
+
+class OracleMatcherTest : public CppUnit::TestFixture {
+ CPPUNIT_TEST_SUITE ( OracleMatcherTest );
+ CPPUNIT_TEST ( testBackwardOracleMatching );
+ CPPUNIT_TEST_SUITE_END ( );
+
+public:
+ void setUp ( );
+ void tearDown ( );
+
+ void testFactorOracleConstruction ( );
+ void testBackwardOracleMatching ( );
+};
+
+#endif // ORACLE_MATCHER_TEST_H_
diff --git a/alib2algo_experimental/src/stringology/exact/BackwardDAWGMatching.cpp b/alib2algo_experimental/src/stringology/exact/BackwardDAWGMatching.cpp
deleted file mode 100644
index e9423e82a7..0000000000
--- a/alib2algo_experimental/src/stringology/exact/BackwardDAWGMatching.cpp
+++ /dev/null
@@ -1,76 +0,0 @@
-/*
- * Author: Radovan Cerveny
- */
-
-#include "BackwardDAWGMatching.h"
-#include "stringology/matching/DAWGMatcherConstruction.h"
-
-#include <string/LinearString.h>
-#include <alphabet/Symbol.h>
-
-#include <algorithm>
-#include <map>
-#include <bitset>
-#include <measure>
-
-namespace stringology {
-
-namespace exact {
-
-std::set < unsigned > BackwardDAWGMatching::match ( const string::String & subject, const string::String & pattern ) {
- return dispatch ( subject.getData ( ), pattern.getData ( ) );
-}
-
-std::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString < > & subject, const string::LinearString < > & pattern ) {
- std::set < unsigned > occ;
-
- measurements::start ( "Preprocess", measurements::Type::PREPROCESS );
-
- automaton::DFA<> suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( pattern );
-
- measurements::end ( );
-
- measurements::start ( "Algorithm", measurements::Type::ALGORITHM );
-
- const DefaultStateType failState = DefaultStateType ( -1 );
-
- size_t posInSubject = 0;
-
- while ( posInSubject <= subject.getContent ( ).size ( ) - pattern.getContent ( ).size ( ) ) {
- DefaultStateType currentState = suffixAutomaton.getInitialState ( );
-
- size_t posInPattern = pattern.getContent ( ).size ( );
- size_t lastPrefixPos = posInPattern;
-
- while ( posInPattern > 0 && currentState != failState ) {
- auto transition = suffixAutomaton.getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
-
- if ( transition == suffixAutomaton.getTransitions ( ).end ( ) )
- currentState = failState;
- else
- currentState = transition->second;
-
- posInPattern--;
-
- // found a prefix of nonreversed pattern that does not correspond to the entire pattern
- if ( ( posInPattern != 0 ) && ( suffixAutomaton.getFinalStates ( ).find ( currentState ) != suffixAutomaton.getFinalStates ( ).end ( ) ) )
- lastPrefixPos = posInPattern;
- }
-
- if ( currentState != failState )
- // Yay, there is match!!!
- occ.insert ( posInSubject );
-
- posInSubject += lastPrefixPos;
- }
-
- measurements::end ( );
-
- return occ;
-}
-
-auto BackwardDAWGMatchingLinearStringLinearString = BackwardDAWGMatching::RegistratorWrapper < std::set < unsigned >, string::LinearString < >, string::LinearString < > > ( BackwardDAWGMatching::match );
-
-} /* namespace exact */
-
-} /* namespace stringology */
diff --git a/alib2algo_experimental/src/stringology/exact/BackwardOracleMatching.cpp b/alib2algo_experimental/src/stringology/exact/BackwardOracleMatching.cpp
deleted file mode 100644
index 7fdff52298..0000000000
--- a/alib2algo_experimental/src/stringology/exact/BackwardOracleMatching.cpp
+++ /dev/null
@@ -1,72 +0,0 @@
-/*
- * Author: Radovan Cerveny
- */
-
-#include "BackwardOracleMatching.h"
-#include "stringology/matching/OracleMatcherConstruction.h"
-
-#include <string/LinearString.h>
-#include <alphabet/Symbol.h>
-
-#include <algorithm>
-#include <map>
-#include <bitset>
-#include <measure>
-
-namespace stringology {
-
-namespace exact {
-
-std::set < unsigned > BackwardOracleMatching::match ( const string::String & subject, const string::String & pattern ) {
- return dispatch ( subject.getData ( ), pattern.getData ( ) );
-}
-
-std::set < unsigned > BackwardOracleMatching::match ( const string::LinearString < > & subject, const string::LinearString < > & pattern ) {
- std::set < unsigned > occ;
-
- measurements::start ( "Preprocess", measurements::Type::PREPROCESS );
-
- automaton::DFA<> factorOracle = stringology::matching::OracleMatcherConstruction::construct ( pattern );
-
- measurements::end ( );
-
- measurements::start ( "Algorithm", measurements::Type::ALGORITHM );
-
- const DefaultStateType failState = DefaultStateType ( -1 );
-
- size_t posInSubject = 0;
-
- while ( posInSubject <= subject.getContent ( ).size ( ) - pattern.getContent ( ).size ( ) ) {
-
- DefaultStateType currentState = factorOracle.getInitialState ( );
-
- size_t posInPattern = pattern.getContent ( ).size ( );
-
- while ( posInPattern > 0 && currentState != failState ) {
- auto transition = factorOracle.getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
-
- if ( transition == factorOracle.getTransitions ( ).end ( ) )
- currentState = failState;
- else
- currentState = transition->second;
-
- posInPattern--;
- }
-
- if ( currentState != failState )
- // Yay, there is match!!!
- occ.insert ( posInSubject );
-
- posInSubject += posInPattern + 1;
- }
-
- measurements::end ( );
-
- return occ;
-}
-
-auto BackwardOracleMatchingLinearStringLinearString = BackwardOracleMatching::RegistratorWrapper < std::set < unsigned >, string::LinearString < >, string::LinearString < > > ( BackwardOracleMatching::match );
-
-} /* namespace exact */
-
-} /* namespace stringology */
diff --git a/alib2algo_experimental/test-src/stringology/exact/DAWGMatcherTest.cpp b/alib2algo_experimental/test-src/stringology/exact/DAWGMatcherTest.cpp
deleted file mode 100644
index 94063aa444..0000000000
--- a/alib2algo_experimental/test-src/stringology/exact/DAWGMatcherTest.cpp
+++ /dev/null
@@ -1,52 +0,0 @@
-#include "DAWGMatcherTest.h"
-
-#include "string/LinearString.h"
-#include "stringology/exact/BackwardDAWGMatching.h"
-#include "stringology/exact/ExactFactorMatch.h"
-
-#include "string/generate/RandomStringFactory.h"
-#include "string/generate/RandomSubstringFactory.h"
-
-CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( DAWGMatcherTest, "stringology" );
-CPPUNIT_TEST_SUITE_REGISTRATION ( DAWGMatcherTest );
-
-void DAWGMatcherTest::setUp ( ) {
-}
-
-void DAWGMatcherTest::tearDown ( ) {
-}
-
-void DAWGMatcherTest::testBackwardDAWGMatching ( ) {
- std::vector<std::string> subjects;
- std::vector<std::string> patterns;
- std::vector<std::set<unsigned>> expectedOccs;
-
- subjects.push_back("a"); patterns.push_back("a"); expectedOccs.push_back({0});
- subjects.push_back("a"); patterns.push_back("b"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfa"); expectedOccs.push_back({0});
- subjects.push_back("alfalfalfa"); patterns.push_back("blfalfalfa"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfb"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({0});
- subjects.push_back("alfalfalfaalfalfalfaabfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); expectedOccs.push_back({0});
- subjects.push_back("atggccttgcc"); patterns.push_back("gcc"); expectedOccs.push_back({3,8});
- subjects.push_back("aaaaaaaaaa"); patterns.push_back("a"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8,9});
- subjects.push_back("aaaaaaaaaa"); patterns.push_back("aa"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8});
-
-
- for(size_t i = 0; i < subjects.size(); ++i) {
- string::String subject = string::stringFrom ( subjects[i] );
- string::String pattern = string::stringFrom ( patterns[i] );
- std::set < unsigned > res = stringology::exact::BackwardDAWGMatching::match ( subject, pattern );
- std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
- CPPUNIT_ASSERT ( res == expectedOccs[i] );
- }
-
- auto longSubject = string::generate::RandomStringFactory::generateLinearString (64 * 64, 512, false, true);
- auto longPattern = string::generate::RandomSubstringFactory::generateSubstring(64 * 32, longSubject );
- std::set < unsigned > res = stringology::exact::BackwardDAWGMatching::match ( longSubject, longPattern );
- std::set < unsigned > resRef = stringology::exact::ExactFactorMatch::match ( longSubject, longPattern );
- std::cout << "long: " << res << std::endl;
- CPPUNIT_ASSERT ( res == resRef);
-}
-
diff --git a/alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.cpp b/alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.cpp
deleted file mode 100644
index b0c3a7d4e6..0000000000
--- a/alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.cpp
+++ /dev/null
@@ -1,50 +0,0 @@
-#include "OracleMatcherTest.h"
-
-#include "string/LinearString.h"
-#include "stringology/exact/BackwardOracleMatching.h"
-#include "stringology/exact/ExactFactorMatch.h"
-
-#include "string/generate/RandomStringFactory.h"
-#include "string/generate/RandomSubstringFactory.h"
-
-CPPUNIT_TEST_SUITE_NAMED_REGISTRATION ( OracleMatcherTest, "stringology" );
-CPPUNIT_TEST_SUITE_REGISTRATION ( OracleMatcherTest );
-
-void OracleMatcherTest::setUp ( ) {
-}
-
-void OracleMatcherTest::tearDown ( ) {
-}
-
-void OracleMatcherTest::testBackwardOracleMatching ( ) {
- std::vector<std::string> subjects;
- std::vector<std::string> patterns;
- std::vector<std::set<unsigned>> expectedOccs;
-
- subjects.push_back("a"); patterns.push_back("a"); expectedOccs.push_back({0});
- subjects.push_back("a"); patterns.push_back("b"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfa"); expectedOccs.push_back({0});
- subjects.push_back("alfalfalfa"); patterns.push_back("blfalfalfa"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfa"); patterns.push_back("alfalfalfb"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({0});
- subjects.push_back("alfalfalfaalfalfalfaabfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfa"); expectedOccs.push_back({});
- subjects.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); patterns.push_back("alfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfalfalfaalfaa"); expectedOccs.push_back({0});
- subjects.push_back("atggccttgcc"); patterns.push_back("gcc"); expectedOccs.push_back({3,8});
- subjects.push_back("aaaaaaaaaa"); patterns.push_back("a"); expectedOccs.push_back({0,1,2,3,4,5,6,7,8,9});
-
-
- for(size_t i = 0; i < subjects.size(); ++i) {
- string::String subject = string::stringFrom ( subjects[i] );
- string::String pattern = string::stringFrom ( patterns[i] );
- std::set < unsigned > res = stringology::exact::BackwardOracleMatching::match ( subject, pattern );
- std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
- CPPUNIT_ASSERT ( res == expectedOccs[i] );
- }
-
- auto longSubject = string::generate::RandomStringFactory::generateLinearString (64 * 64, 512, false, true);
- auto longPattern = string::generate::RandomSubstringFactory::generateSubstring(64 * 32, longSubject );
- std::set < unsigned > res = stringology::exact::BackwardOracleMatching::match ( longSubject, longPattern );
- std::set < unsigned > resRef = stringology::exact::ExactFactorMatch::match ( longSubject, longPattern );
- std::cout << "long: " << res << std::endl;
- CPPUNIT_ASSERT ( res == resRef);
-}
diff --git a/alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.h b/alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.h
deleted file mode 100644
index f1d4980908..0000000000
--- a/alib2algo_experimental/test-src/stringology/exact/OracleMatcherTest.h
+++ /dev/null
@@ -1,19 +0,0 @@
-#ifndef ORACLE_MATCHER_TEST_H_
-#define ORACLE_MATCHER_TEST_H_
-
-#include <cppunit/extensions/HelperMacros.h>
-
-class OracleMatcherTest : public CppUnit::TestFixture {
- CPPUNIT_TEST_SUITE ( OracleMatcherTest );
- CPPUNIT_TEST ( testBackwardOracleMatching );
- CPPUNIT_TEST_SUITE_END ( );
-
-public:
- void setUp ( );
- void tearDown ( );
-
- void testFactorOracleConstruction ( );
- void testBackwardOracleMatching ( );
-};
-
-#endif // ORACLE_MATCHER_TEST_H_
diff --git a/astringology2/src/astringology.cpp b/astringology2/src/astringology.cpp
index 3f3159f4fe..eb0bfe8ea4 100644
--- a/astringology2/src/astringology.cpp
+++ b/astringology2/src/astringology.cpp
@@ -25,8 +25,8 @@
#include <stringology/exact/ReversedBoyerMooreHorspool.h>
#include <stringology/exact/DeadZoneUsingBadCharacterShift.h>
#include <stringology/query/BNDMOccurrences.h>
-#include <stringology/exact/BackwardOracleMatching.h>
-#include <stringology/exact/BackwardDAWGMatching.h>
+#include <stringology/query/BackwardOracleMatching.h>
+#include <stringology/query/BackwardDAWGMatching.h>
#include <stringology/matching/ExactMatchingAutomaton.h>
#include <stringology/indexing/NondeterministicExactFactorAutomaton.h>
#include <stringology/indexing/ExactSubsequenceAutomaton.h>
@@ -182,12 +182,12 @@ int main ( int argc, char * argv[] ) {
alib::XmlDataFactory::toStdout ( res );
} else if ( algorithm.getValue ( ) == "backwardOracleMatching" ) {
string::String subject = alib::XmlDataFactory::fromTokens ( std::move ( sax::FromXMLParserHelper::parseInput(true, subjectInput).front ( ) ) );
- string::String pattern = alib::XmlDataFactory::fromTokens ( std::move ( sax::FromXMLParserHelper::parseInput(true, patternInput).front ( ) ) );
+ automaton::Automaton pattern = alib::XmlDataFactory::fromTokens ( std::move ( sax::FromXMLParserHelper::parseInput(true, patternInput).front ( ) ) );
measurements::end ( );
measurements::start ( "Algorithm", measurements::Type::MAIN );
- std::set < unsigned > res = stringology::exact::BackwardOracleMatching::match ( subject, pattern );
+ std::set < unsigned > res = stringology::query::BackwardOracleMatching::match ( subject, pattern );
measurements::end ( );
measurements::start ( "Output write", measurements::Type::AUXILIARY );
@@ -195,12 +195,12 @@ int main ( int argc, char * argv[] ) {
alib::XmlDataFactory::toStdout ( res );
} else if ( algorithm.getValue ( ) == "backwardDAWGMatching" ) {
string::String subject = alib::XmlDataFactory::fromTokens ( std::move ( sax::FromXMLParserHelper::parseInput(true, subjectInput).front ( ) ) );
- string::String pattern = alib::XmlDataFactory::fromTokens ( std::move ( sax::FromXMLParserHelper::parseInput(true, patternInput).front ( ) ) );
+ automaton::Automaton pattern = alib::XmlDataFactory::fromTokens ( std::move ( sax::FromXMLParserHelper::parseInput(true, patternInput).front ( ) ) );
measurements::end ( );
measurements::start ( "Algorithm", measurements::Type::MAIN );
- std::set < unsigned > res = stringology::exact::BackwardDAWGMatching::match ( subject, pattern );
+ std::set < unsigned > res = stringology::query::BackwardDAWGMatching::match ( subject, pattern );
measurements::end ( );
measurements::start ( "Output write", measurements::Type::AUXILIARY );
--
GitLab