From a551906263e842738f0a4dd14aaa00822d70d7cd Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Tue, 24 Oct 2017 19:33:07 +0200
Subject: [PATCH] templates in suffix automaton direct construction
---
.../indexing/ExactSuffixAutomaton.cpp | 79 +----------------
.../indexing/ExactSuffixAutomaton.h | 87 +++++++++++++++++--
.../matching/DAWGMatcherConstruction.cpp | 4 +-
.../matching/DAWGMatcherConstruction.h | 4 +-
.../matching/OracleMatcherConstruction.cpp | 4 +-
.../matching/OracleMatcherConstruction.h | 2 +-
.../query/BackwardDAWGMatching.cpp | 46 ----------
.../stringology/query/BackwardDAWGMatching.h | 53 ++++++++++-
.../query/BackwardOracleMatching.cpp | 41 ---------
.../query/BackwardOracleMatching.h | 45 +++++++++-
.../indexing/ExactSuffixAutomatonTest.cpp | 34 ++++----
.../stringology/query/DAWGMatcherTest.cpp | 4 +-
.../stringology/query/OracleMatcherTest.cpp | 4 +-
13 files changed, 200 insertions(+), 207 deletions(-)
diff --git a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp
index e8a6d63657..e0081a874e 100644
--- a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp
+++ b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp
@@ -5,81 +5,8 @@
#include "ExactSuffixAutomaton.h"
#include <registration/AlgoRegistration.hpp>
-namespace stringology {
+namespace {
-namespace indexing {
+auto SuffixAutomatonLinearString = registration::AbstractRegister < stringology::indexing::ExactSuffixAutomaton, automaton::DFA < DefaultSymbolType, unsigned >, const string::LinearString < > & > ( stringology::indexing::ExactSuffixAutomaton::construct );
-automaton::DFA < > ExactSuffixAutomaton::construct ( const string::LinearString < > & pattern ) {
- automaton::DFA < > suffixAutomaton ( DefaultStateType ( 0 ) );
-
- suffixAutomaton.setInputAlphabet ( pattern.getAlphabet ( ) );
-
- ext::map < DefaultStateType, std::pair < DefaultStateType, int > > suffixLinks = { { DefaultStateType ( 0 ), { DefaultStateType ( -1 ), 0 } } };
- DefaultStateType lastState = DefaultStateType ( 0 );
-
- for ( const DefaultSymbolType & symbol : pattern.getContent ( ) )
- suffixAutomatonAddSymbol ( suffixAutomaton, symbol, suffixLinks, lastState );
-
- while ( lastState != DefaultStateType ( -1 ) ) {
- suffixAutomaton.addFinalState ( lastState );
- lastState = suffixLinks.find ( lastState )->second.first;
- }
-
- return suffixAutomaton;
-}
-
-void ExactSuffixAutomaton::suffixAutomatonAddSymbol ( automaton::DFA < > & suffixAutomaton, const DefaultSymbolType & symbol, ext::map < DefaultStateType, std::pair < DefaultStateType, int > > & suffixLinks, DefaultStateType & lastState ) {
- DefaultStateType newState = DefaultStateType ( ( int ) suffixAutomaton.getStates ( ).size ( ) );
-
- suffixAutomaton.addState ( newState );
-
- int lastSuffixLength = suffixLinks.find ( lastState )->second.second;
-
- suffixLinks.insert ( { newState, { DefaultStateType ( -1 ), lastSuffixLength + 1 } } );
-
- DefaultStateType kState = lastState;
-
- while ( kState != DefaultStateType ( -1 ) && suffixAutomaton.getTransitions ( ).find ( { kState, symbol } ) == suffixAutomaton.getTransitions ( ).end ( ) ) {
- suffixAutomaton.addTransition ( kState, symbol, newState );
- kState = suffixLinks.find ( kState )->second.first;
- }
-
- if ( kState == DefaultStateType ( -1 ) )
- suffixLinks.find ( newState )->second.first = DefaultStateType ( 0 );
- else {
- DefaultStateType qState = suffixAutomaton.getTransitions ( ).find ( { kState, symbol } )->second;
-
- int kSuffixLength = suffixLinks.find ( kState )->second.second;
- int qSuffixLength = suffixLinks.find ( qState )->second.second;
-
- if ( kSuffixLength + 1 == qSuffixLength ) {
- suffixLinks.find ( newState )->second.first = qState;
- } else {
-
- DefaultStateType cloneState = DefaultStateType ( ( int ) suffixAutomaton.getStates ( ).size ( ) );
- suffixAutomaton.addState ( cloneState );
-
- suffixLinks.insert ( { cloneState, { suffixLinks.find ( qState )->second.first, kSuffixLength + 1 } } );
-
- for ( const auto & transition : suffixAutomaton.getTransitionsFromState ( qState ) )
- suffixAutomaton.addTransition ( cloneState, transition.first.second, transition.second );
-
- while ( kState != DefaultStateType ( -1 ) && suffixAutomaton.getTransitions ( ).find ( { kState, symbol } ) != suffixAutomaton.getTransitions ( ).end ( ) && suffixAutomaton.getTransitions ( ).find ( { kState, symbol } )->second == qState ) {
- suffixAutomaton.removeTransition ( kState, symbol, qState );
- suffixAutomaton.addTransition ( kState, symbol, cloneState );
- kState = suffixLinks.find ( kState )->second.first;
- }
-
- suffixLinks.find ( qState )->second.first = cloneState;
- suffixLinks.find ( newState )->second.first = cloneState;
- }
- }
-
- lastState = newState;
-}
-
-auto SuffixAutomatonLinearString = registration::AbstractRegister < ExactSuffixAutomaton, automaton::DFA < >, const string::LinearString < > & > ( ExactSuffixAutomaton::construct );
-
-} /* namespace indexing */
-
-} /* namespace stringology */
+} /* namespace */
diff --git a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
index 7287859c5f..528d580077 100644
--- a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
+++ b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
@@ -2,8 +2,8 @@
* Author: Radovan Cerveny
*/
-#ifndef EXACT_SUFFIX_AUTOMATON_H_
-#define EXACT_SUFFIX_AUTOMATON_H_
+#ifndef _EXACT_SUFFIX_AUTOMATON_H_
+#define _EXACT_SUFFIX_AUTOMATON_H_
#include <automaton/FSM/DFA.h>
#include <string/LinearString.h>
@@ -14,19 +14,88 @@ namespace indexing {
class ExactSuffixAutomaton {
private:
- static void suffixAutomatonAddSymbol ( automaton::DFA < > & suffixAutomaton, const DefaultSymbolType & symbol, ext::map < DefaultStateType, std::pair<DefaultStateType, int > > & suffixLinks, DefaultStateType & lastState );
+ template < class SymbolType >
+ static void suffixAutomatonAddSymbol ( automaton::DFA < SymbolType, unsigned > & suffixAutomaton, const SymbolType & symbol, std::map < unsigned, std::pair < unsigned, int > > & suffixLinks, unsigned & lastState );
public:
- /**
- * Linear time on-line construction of minimal suffix automaton for given pattern.
- * @return minimal suffix automaton for given pattern.
- */
- static automaton::DFA < > construct ( const string::LinearString < > & pattern );
+ template < class SymbolType >
+ static automaton::DFA < SymbolType, unsigned > construct ( const string::LinearString < SymbolType > & pattern );
};
+template < class SymbolType >
+automaton::DFA < SymbolType, unsigned > ExactSuffixAutomaton::construct ( const string::LinearString < SymbolType > & pattern ) {
+ automaton::DFA < SymbolType, unsigned > suffixAutomaton ( 0 );
+
+ suffixAutomaton.setInputAlphabet ( pattern.getAlphabet ( ) );
+
+ std::map < unsigned, std::pair < unsigned, int > > suffixLinks = { { 0u, { ( unsigned ) -1, 0 } } };
+ unsigned lastState = 0;
+
+ for ( const DefaultSymbolType & symbol : pattern.getContent ( ) )
+ suffixAutomatonAddSymbol ( suffixAutomaton, symbol, suffixLinks, lastState );
+
+ while ( lastState != ( unsigned ) -1 ) {
+ suffixAutomaton.addFinalState ( lastState );
+ lastState = suffixLinks.find ( lastState )->second.first;
+ }
+
+ return suffixAutomaton;
+}
+
+template < class SymbolType >
+void ExactSuffixAutomaton::suffixAutomatonAddSymbol ( automaton::DFA < SymbolType, unsigned > & suffixAutomaton, const SymbolType & symbol, std::map < unsigned, std::pair < unsigned, int > > & suffixLinks, unsigned & lastState ) {
+ unsigned newState = suffixAutomaton.getStates ( ).size ( );
+
+ suffixAutomaton.addState ( newState );
+
+ int lastSuffixLength = suffixLinks.find ( lastState )->second.second;
+
+ suffixLinks.insert ( { newState, { ( unsigned ) -1, lastSuffixLength + 1 } } );
+
+ unsigned kState = lastState;
+
+ while ( kState != ( unsigned ) -1 && suffixAutomaton.getTransitions ( ).find ( { kState, symbol } ) == suffixAutomaton.getTransitions ( ).end ( ) ) {
+ suffixAutomaton.addTransition ( kState, symbol, newState );
+ kState = suffixLinks.find ( kState )->second.first;
+ }
+
+ if ( kState == ( unsigned ) -1 ) {
+ suffixLinks.find ( newState )->second.first = 0;
+ } else {
+ unsigned qState = suffixAutomaton.getTransitions ( ).find ( { kState, symbol } )->second;
+
+ int kSuffixLength = suffixLinks.find ( kState )->second.second;
+ int qSuffixLength = suffixLinks.find ( qState )->second.second;
+
+ if ( kSuffixLength + 1 == qSuffixLength ) {
+ suffixLinks.find ( newState )->second.first = qState;
+ } else {
+ unsigned cloneState = suffixAutomaton.getStates ( ).size ( );
+ suffixAutomaton.addState ( cloneState );
+
+ suffixLinks.insert ( { cloneState, { suffixLinks.find ( qState )->second.first, kSuffixLength + 1 } } );
+
+ for ( const auto & transition : suffixAutomaton.getTransitionsFromState ( qState ) )
+ suffixAutomaton.addTransition ( cloneState, transition.first.second, transition.second );
+
+ while ( kState != ( unsigned ) -1
+ && suffixAutomaton.getTransitions ( ).find ( { kState, symbol } ) != suffixAutomaton.getTransitions ( ).end ( )
+ && suffixAutomaton.getTransitions ( ).find ( { kState, symbol } )->second == qState ) {
+ suffixAutomaton.removeTransition ( kState, symbol, qState );
+ suffixAutomaton.addTransition ( kState, symbol, cloneState );
+ kState = suffixLinks.find ( kState )->second.first;
+ }
+
+ suffixLinks.find ( qState )->second.first = cloneState;
+ suffixLinks.find ( newState )->second.first = cloneState;
+ }
+ }
+ lastState = newState;
+}
+
} /* namespace indexing */
} /* namespace stringology */
-#endif /* EXACT_SUFFIX_AUTOMATON_H_ */
+#endif /* _EXACT_SUFFIX_AUTOMATON_H_ */
diff --git a/alib2algo/src/stringology/matching/DAWGMatcherConstruction.cpp b/alib2algo/src/stringology/matching/DAWGMatcherConstruction.cpp
index c962d9ee01..90231dfb04 100644
--- a/alib2algo/src/stringology/matching/DAWGMatcherConstruction.cpp
+++ b/alib2algo/src/stringology/matching/DAWGMatcherConstruction.cpp
@@ -10,7 +10,7 @@ namespace stringology {
namespace matching {
-automaton::DFA < > DAWGMatcherConstruction::construct ( const string::LinearString < > & pattern ) {
+automaton::DFA < DefaultSymbolType, unsigned > DAWGMatcherConstruction::construct ( const string::LinearString < > & pattern ) {
auto patternData = pattern.getContent ( );
reverse ( patternData.begin ( ), patternData.end ( ) );
string::LinearString < > reversedPattern ( pattern.getAlphabet ( ), std::move ( patternData ) );
@@ -18,7 +18,7 @@ automaton::DFA < > DAWGMatcherConstruction::construct ( const string::LinearStri
return stringology::indexing::ExactSuffixAutomaton::construct ( reversedPattern );
}
-auto DAWGMatcherConstructionLinearString = registration::AbstractRegister < DAWGMatcherConstruction, automaton::DFA < >, const string::LinearString < > & > ( DAWGMatcherConstruction::construct );
+auto DAWGMatcherConstructionLinearString = registration::AbstractRegister < DAWGMatcherConstruction, automaton::DFA < DefaultSymbolType, unsigned >, const string::LinearString < > & > ( DAWGMatcherConstruction::construct );
} /* namespace matching */
diff --git a/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h b/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h
index 7ffb7b274e..b3f6392ee5 100644
--- a/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h
+++ b/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h
@@ -18,9 +18,7 @@ public:
* Linear time on-line construction of minimal suffix automaton for given pattern.
* @return minimal suffix automaton for given pattern.
*/
- static automaton::Automaton construct ( const string::String & pattern );
-
- static automaton::DFA < > construct ( const string::LinearString < > & pattern );
+ static automaton::DFA < DefaultSymbolType, unsigned > construct ( const string::LinearString < > & pattern );
};
diff --git a/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp b/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp
index 943f9a8344..8f27209ffa 100644
--- a/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp
+++ b/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp
@@ -10,7 +10,7 @@ namespace stringology {
namespace matching {
-automaton::DFA < > OracleMatcherConstruction::construct ( const string::LinearString < > & pattern ) {
+automaton::DFA < DefaultSymbolType, unsigned > OracleMatcherConstruction::construct ( const string::LinearString < > & pattern ) {
auto patternData = pattern.getContent ( );
reverse ( patternData.begin ( ), patternData.end ( ) );
string::LinearString < > reversedPattern ( pattern.getAlphabet ( ), std::move ( patternData ) );
@@ -18,7 +18,7 @@ automaton::DFA < > OracleMatcherConstruction::construct ( const string::LinearSt
return stringology::indexing::ExactSuffixAutomaton::construct ( reversedPattern );
}
-auto OracleMatcherConstructionLinearString = registration::AbstractRegister < OracleMatcherConstruction, automaton::DFA < >, const string::LinearString < > & > ( OracleMatcherConstruction::construct );
+auto OracleMatcherConstructionLinearString = registration::AbstractRegister < OracleMatcherConstruction, automaton::DFA < DefaultSymbolType, unsigned >, const string::LinearString < > & > ( OracleMatcherConstruction::construct );
} /* namespace matching */
diff --git a/alib2algo/src/stringology/matching/OracleMatcherConstruction.h b/alib2algo/src/stringology/matching/OracleMatcherConstruction.h
index 4f35f4a85e..55b3bd3740 100644
--- a/alib2algo/src/stringology/matching/OracleMatcherConstruction.h
+++ b/alib2algo/src/stringology/matching/OracleMatcherConstruction.h
@@ -18,7 +18,7 @@ public:
* Linear time on-line construction of minimal suffix automaton for given pattern.
* @return minimal suffix automaton for given pattern.
*/
- static automaton::DFA < > construct ( const string::LinearString < > & pattern );
+ static automaton::DFA < DefaultSymbolType, unsigned > construct ( const string::LinearString < > & pattern );
};
diff --git a/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp b/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
index f72c0101a3..a1ea591b82 100644
--- a/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
+++ b/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
@@ -3,58 +3,12 @@
*/
#include "BackwardDAWGMatching.h"
-#include <stringology/matching/DAWGMatcherConstruction.h>
-#include <stringology/properties/BackboneLength.h>
-
-#include <string/LinearString.h>
-
#include <registration/AlgoRegistration.hpp>
namespace stringology {
namespace query {
-ext::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString < > & subject, const automaton::DFA < > & suffixAutomaton ) {
- ext::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 = registration::AbstractRegister < BackwardDAWGMatching, ext::set < unsigned >, const string::LinearString < > &, const automaton::DFA < > & > ( BackwardDAWGMatching::match );
} /* namespace query */
diff --git a/alib2algo/src/stringology/query/BackwardDAWGMatching.h b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
index 670d8fd9ac..05dfb02308 100644
--- a/alib2algo/src/stringology/query/BackwardDAWGMatching.h
+++ b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
@@ -5,11 +5,13 @@
#ifndef STRINGOLOGY_BACKWARD_DAWG_MATCHING_H__
#define STRINGOLOGY_BACKWARD_DAWG_MATCHING_H__
-#include <string/StringFeatures.h>
-#include <automaton/AutomatonFeatures.h>
-
#include <set>
+#include <automaton/FSM/DFA.h>
+#include <string/LinearString.h>
+
+#include <stringology/properties/BackboneLength.h>
+
namespace stringology {
namespace query {
@@ -23,10 +25,53 @@ public:
* Search for pattern in linear string.
* @return set set of occurences
*/
- static ext::set < unsigned > match ( const string::LinearString < > & subject, const automaton::DFA < > & suffixAutomaton );
+ template < class SymbolType, class StateType >
+ static ext::set < unsigned > match ( const string::LinearString < SymbolType > & subject, const automaton::DFA < SymbolType, StateType > & suffixAutomaton );
};
+template < class SymbolType, class StateType >
+ext::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString < SymbolType > & subject, const automaton::DFA < SymbolType, StateType > & suffixAutomaton ) {
+ ext::set < unsigned > occ;
+
+ size_t patternSize = stringology::properties::BackboneLength::length ( suffixAutomaton );
+ size_t subjectSize = subject.getContent ( ).size ( );
+
+ bool fail;
+ size_t posInSubject = 0;
+
+ while ( posInSubject <= subjectSize - patternSize ) {
+ StateType currentState = suffixAutomaton.getInitialState ( );
+
+ size_t posInPattern = patternSize;
+ size_t lastPrefixPos = posInPattern;
+
+ fail = false;
+ while ( posInPattern > 0 && ! fail ) {
+ auto transition = suffixAutomaton.getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
+
+ if ( transition == suffixAutomaton.getTransitions ( ).end ( ) )
+ fail = true;
+ 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 ( ! fail )
+ // Yay, there is match!!!
+ occ.insert ( posInSubject );
+
+ posInSubject += lastPrefixPos;
+ }
+
+ return occ;
+}
+
} /* namespace query */
} /* namespace stringology */
diff --git a/alib2algo/src/stringology/query/BackwardOracleMatching.cpp b/alib2algo/src/stringology/query/BackwardOracleMatching.cpp
index 60f9abe778..abe024f72a 100644
--- a/alib2algo/src/stringology/query/BackwardOracleMatching.cpp
+++ b/alib2algo/src/stringology/query/BackwardOracleMatching.cpp
@@ -3,53 +3,12 @@
*/
#include "BackwardOracleMatching.h"
-#include <stringology/matching/OracleMatcherConstruction.h>
-#include <stringology/properties/BackboneLength.h>
-
-#include <string/LinearString.h>
#include <registration/AlgoRegistration.hpp>
namespace stringology {
namespace query {
-ext::set < unsigned > BackwardOracleMatching::match ( const string::LinearString < > & subject, const automaton::DFA < > & factorOracle ) {
- ext::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 = registration::AbstractRegister < BackwardOracleMatching, ext::set < unsigned >, const string::LinearString < > &, const automaton::DFA < > & > ( BackwardOracleMatching::match );
} /* namespace query */
diff --git a/alib2algo/src/stringology/query/BackwardOracleMatching.h b/alib2algo/src/stringology/query/BackwardOracleMatching.h
index 911cd93ff1..8ed9a47e24 100644
--- a/alib2algo/src/stringology/query/BackwardOracleMatching.h
+++ b/alib2algo/src/stringology/query/BackwardOracleMatching.h
@@ -5,10 +5,12 @@
#ifndef STRINGOLOGY_BACKWARD_ORACLE_MATCHING_H__
#define STRINGOLOGY_BACKWARD_ORACLE_MATCHING_H__
+#include <set>
+
#include <string/LinearString.h>
#include <automaton/FSM/DFA.h>
-#include <set>
+#include <stringology/properties/BackboneLength.h>
namespace stringology {
@@ -23,10 +25,49 @@ public:
* Search for pattern in linear string.
* @return set set of occurences
*/
- static ext::set < unsigned > match ( const string::LinearString < > & subject, const automaton::DFA < > & factorOracle );
+ template < class SymbolType, class StateType >
+ static ext::set < unsigned > match ( const string::LinearString < SymbolType > & subject, const automaton::DFA < SymbolType, StateType > & factorOracle );
};
+template < class SymbolType, class StateType >
+ext::set < unsigned > BackwardOracleMatching::match ( const string::LinearString < SymbolType > & subject, const automaton::DFA < SymbolType, StateType > & factorOracle ) {
+ ext::set < unsigned > occ;
+
+ size_t patternSize = stringology::properties::BackboneLength::length ( factorOracle );
+ size_t subjectSize = subject.getContent ( ).size ( );
+
+ bool fail;
+ size_t posInSubject = 0;
+
+ while ( posInSubject <= subjectSize - patternSize ) {
+
+ StateType currentState = factorOracle.getInitialState ( );
+
+ size_t posInPattern = patternSize;
+
+ fail = false;
+ while ( posInPattern > 0 && ! fail ) {
+ auto transition = factorOracle.getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
+
+ if ( transition == factorOracle.getTransitions ( ).end ( ) )
+ fail = true;
+ else
+ currentState = transition->second;
+
+ posInPattern--;
+ }
+
+ if ( ! fail )
+ // Yay, there is match!!!
+ occ.insert ( posInSubject );
+
+ posInSubject += posInPattern + 1;
+ }
+
+ return occ;
+}
+
} /* namespace query */
} /* namespace stringology */
diff --git a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
index 3427482ec2..61587da32c 100644
--- a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
@@ -16,29 +16,29 @@ void ExactSuffixAutomatonTest::testSuffixAutomatonConstruction ( ) {
string::LinearString < > pattern ( "atatac" );
- automaton::DFA<> suffixAutomaton = stringology::indexing::ExactSuffixAutomaton::construct ( pattern );
+ automaton::DFA < DefaultSymbolType, unsigned > suffixAutomaton = stringology::indexing::ExactSuffixAutomaton::construct ( pattern );
- automaton::DFA<> refSuffixAutomaton ( DefaultStateType ( 0 ) );
+ automaton::DFA < DefaultSymbolType, unsigned > refSuffixAutomaton ( 0 );
refSuffixAutomaton.setInputAlphabet ( pattern.getAlphabet ( ) );
- for ( int i = 1; i <= 6; ++i ) {
- refSuffixAutomaton.addState ( DefaultStateType ( i ) );
+ for ( unsigned i = 1; i <= 6; ++i ) {
+ refSuffixAutomaton.addState ( i );
}
- refSuffixAutomaton.addFinalState ( DefaultStateType ( 0 ) );
- refSuffixAutomaton.addFinalState ( DefaultStateType ( 6 ) );
-
- refSuffixAutomaton.addTransition ( DefaultStateType ( 0 ), DefaultSymbolType ( 'a' ), DefaultStateType ( 1 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 0 ), DefaultSymbolType ( 't' ), DefaultStateType ( 2 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 0 ), DefaultSymbolType ( 'c' ), DefaultStateType ( 6 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 1 ), DefaultSymbolType ( 't' ), DefaultStateType ( 2 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 1 ), DefaultSymbolType ( 'c' ), DefaultStateType ( 6 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 2 ), DefaultSymbolType ( 'a' ), DefaultStateType ( 3 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 3 ), DefaultSymbolType ( 't' ), DefaultStateType ( 4 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 3 ), DefaultSymbolType ( 'c' ), DefaultStateType ( 6 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 4 ), DefaultSymbolType ( 'a' ), DefaultStateType ( 5 ) );
- refSuffixAutomaton.addTransition ( DefaultStateType ( 5 ), DefaultSymbolType ( 'c' ), DefaultStateType ( 6 ) );
+ refSuffixAutomaton.addFinalState ( 0 );
+ refSuffixAutomaton.addFinalState ( 6 );
+
+ refSuffixAutomaton.addTransition ( 0, DefaultSymbolType ( 'a' ), 1 );
+ refSuffixAutomaton.addTransition ( 0, DefaultSymbolType ( 't' ), 2 );
+ refSuffixAutomaton.addTransition ( 0, DefaultSymbolType ( 'c' ), 6 );
+ refSuffixAutomaton.addTransition ( 1, DefaultSymbolType ( 't' ), 2 );
+ refSuffixAutomaton.addTransition ( 1, DefaultSymbolType ( 'c' ), 6 );
+ refSuffixAutomaton.addTransition ( 2, DefaultSymbolType ( 'a' ), 3 );
+ refSuffixAutomaton.addTransition ( 3, DefaultSymbolType ( 't' ), 4 );
+ refSuffixAutomaton.addTransition ( 3, DefaultSymbolType ( 'c' ), 6 );
+ refSuffixAutomaton.addTransition ( 4, DefaultSymbolType ( 'a' ), 5 );
+ refSuffixAutomaton.addTransition ( 5, DefaultSymbolType ( 'c' ), 6 );
std::cout << suffixAutomaton << std::endl;
std::cout << refSuffixAutomaton << std::endl;
diff --git a/alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp b/alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp
index 6d933dc80e..254c39a8ca 100644
--- a/alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp
+++ b/alib2algo/test-src/stringology/query/DAWGMatcherTest.cpp
@@ -38,7 +38,7 @@ void DAWGMatcherTest::testBackwardDAWGMatching ( ) {
for(size_t i = 0; i < subjects.size(); ++i) {
string::LinearString < > subject ( subjects[i] );
string::LinearString < > pattern ( patterns[i] );
- automaton::DFA < > suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( pattern );
+ automaton::DFA < DefaultSymbolType, unsigned > suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( pattern );
ext::set < unsigned > res = stringology::query::BackwardDAWGMatching::match ( subject, suffixAutomaton );
std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
CPPUNIT_ASSERT ( res == expectedOccs[i] );
@@ -46,7 +46,7 @@ void DAWGMatcherTest::testBackwardDAWGMatching ( ) {
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 );
+ automaton::DFA < DefaultSymbolType, unsigned > suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( longPattern );
ext::set < unsigned > res = stringology::query::BackwardDAWGMatching::match ( longSubject, suffixAutomaton );
ext::set < unsigned > resRef = stringology::exact::ExactFactorMatch::match ( longSubject, longPattern );
std::cout << "long: " << res << std::endl;
diff --git a/alib2algo/test-src/stringology/query/OracleMatcherTest.cpp b/alib2algo/test-src/stringology/query/OracleMatcherTest.cpp
index 0a9a9d18a1..39bd8d76aa 100644
--- a/alib2algo/test-src/stringology/query/OracleMatcherTest.cpp
+++ b/alib2algo/test-src/stringology/query/OracleMatcherTest.cpp
@@ -37,7 +37,7 @@ void OracleMatcherTest::testBackwardOracleMatching ( ) {
for(size_t i = 0; i < subjects.size(); ++i) {
string::LinearString < > subject ( subjects[i] );
string::LinearString < > pattern ( patterns[i] );
- automaton::DFA < > oracleAutomaton = stringology::matching::OracleMatcherConstruction::construct ( pattern );
+ automaton::DFA < DefaultSymbolType, unsigned > oracleAutomaton = stringology::matching::OracleMatcherConstruction::construct ( pattern );
ext::set < unsigned > res = stringology::query::BackwardOracleMatching::match ( subject, oracleAutomaton );
std::cout << subjects[i] << ' ' << patterns[i] << ' ' << res << std::endl;
CPPUNIT_ASSERT ( res == expectedOccs[i] );
@@ -45,7 +45,7 @@ void OracleMatcherTest::testBackwardOracleMatching ( ) {
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 );
+ automaton::DFA < DefaultSymbolType, unsigned > oracleAutomaton = stringology::matching::OracleMatcherConstruction::construct ( longPattern );
ext::set < unsigned > res = stringology::query::BackwardOracleMatching::match ( longSubject, oracleAutomaton );
ext::set < unsigned > resRef = stringology::exact::ExactFactorMatch::match ( longSubject, longPattern );
std::cout << "long: " << res << std::endl;
--
GitLab