From 4b02ab228ab1682273613d684b8a60a102a816a7 Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Tue, 9 Jan 2018 11:46:09 +0100
Subject: [PATCH] introduce SuffixAutomaton as wrapper of DFA
---
.../query/FullAndLinearIndexPatterns.h | 8 +-
.../NonlinearFullAndLinearIndexPatterns.h | 8 +-
.../indexing/ExactSuffixAutomaton.cpp | 2 +-
.../indexing/ExactSuffixAutomaton.h | 8 +-
.../matching/DAWGMatcherConstruction.cpp | 4 +-
.../matching/DAWGMatcherConstruction.h | 4 +-
.../matching/OracleMatcherConstruction.cpp | 2 +-
.../query/BackwardDAWGMatching.cpp | 2 +-
.../stringology/query/BackwardDAWGMatching.h | 20 +-
.../query/SuffixAutomatonFactors.cpp | 2 +-
.../query/SuffixAutomatonFactors.h | 21 +-
.../indexing/ExactSuffixAutomatonTest.cpp | 36 +--
.../stringology/query/DAWGMatcherTest.cpp | 4 +-
.../indexes/stringology/SuffixAutomaton.cpp | 22 ++
.../src/indexes/stringology/SuffixAutomaton.h | 217 ++++++++++++++++++
15 files changed, 300 insertions(+), 60 deletions(-)
create mode 100644 alib2data/src/indexes/stringology/SuffixAutomaton.cpp
create mode 100644 alib2data/src/indexes/stringology/SuffixAutomaton.h
diff --git a/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.h b/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.h
index 3df6c8bc89..96c502815c 100644
--- a/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.h
+++ b/alib2algo/src/arbology/query/FullAndLinearIndexPatterns.h
@@ -80,14 +80,14 @@ ext::set < unsigned > FullAndLinearIndexPatterns::query ( const indexes::arbolog
}
}
- ext::vector < std::pair < unsigned, unsigned > > prevOcc = FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ) , treePatternParts [ 0 ] );
+ ext::vector < std::pair < unsigned, unsigned > > prevOcc = FullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ) , treePatternParts [ 0 ] );
for ( unsigned i = 1; i < treePatternParts.size ( ); ++ i ) {
for ( std::pair < unsigned, unsigned > & occurrence : prevOcc )
occurrence.second = fullAndLinearIndex.getJumps ( ) [ occurrence.second ];
if ( ! treePatternParts [ i ].empty ( ) )
- prevOcc = MergeOccurrences ( prevOcc, FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
+ prevOcc = MergeOccurrences ( prevOcc, FullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
}
ext::set < unsigned > res;
@@ -113,14 +113,14 @@ ext::set < unsigned > FullAndLinearIndexPatterns::query ( const indexes::arbolog
}
}
- ext::vector < std::pair < unsigned, unsigned > > prevOcc = FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ 0 ] );
+ ext::vector < std::pair < unsigned, unsigned > > prevOcc = FullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ 0 ] );
for ( unsigned i = 1; i < treePatternParts.size ( ); ++ i ) {
for ( std::pair < unsigned, unsigned > & occurrence : prevOcc )
occurrence.second = fullAndLinearIndex.getJumps ( ) [ occurrence.second ];
if ( ! treePatternParts [ i ].empty ( ) )
- prevOcc = MergeOccurrences ( prevOcc, FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
+ prevOcc = MergeOccurrences ( prevOcc, FullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
}
ext::set < unsigned > res;
diff --git a/alib2algo/src/arbology/query/NonlinearFullAndLinearIndexPatterns.h b/alib2algo/src/arbology/query/NonlinearFullAndLinearIndexPatterns.h
index e3a2cca90c..3413dd4f79 100644
--- a/alib2algo/src/arbology/query/NonlinearFullAndLinearIndexPatterns.h
+++ b/alib2algo/src/arbology/query/NonlinearFullAndLinearIndexPatterns.h
@@ -83,7 +83,7 @@ ext::set < unsigned > NonlinearFullAndLinearIndexPatterns::query ( const indexes
}
}
- ext::vector < std::pair < unsigned, unsigned > > prevOcc = FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ 0 ] );
+ ext::vector < std::pair < unsigned, unsigned > > prevOcc = NonlinearFullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ 0 ] );
for ( unsigned i = 1; i < treePatternParts.size ( ); ++ i ) {
for ( std::pair < unsigned, unsigned > & occurrence : prevOcc ) {
@@ -100,7 +100,7 @@ ext::set < unsigned > NonlinearFullAndLinearIndexPatterns::query ( const indexes
++ i;
if ( ! treePatternParts [ i ].empty ( ) )
- prevOcc = MergeOccurrences ( prevOcc, FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
+ prevOcc = MergeOccurrences ( prevOcc, NonlinearFullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
}
ext::set < unsigned > res;
@@ -132,7 +132,7 @@ ext::set < unsigned > NonlinearFullAndLinearIndexPatterns::query ( const indexes
}
}
- ext::vector < std::pair < unsigned, unsigned > > prevOcc = FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ) , treePatternParts [ 0 ] );
+ ext::vector < std::pair < unsigned, unsigned > > prevOcc = NonlinearFullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ 0 ] );
for ( unsigned i = 1; i < treePatternParts.size ( ); ++ i ) {
for ( std::pair < unsigned, unsigned > & occurrence : prevOcc ) {
@@ -149,7 +149,7 @@ ext::set < unsigned > NonlinearFullAndLinearIndexPatterns::query ( const indexes
++ i;
if ( ! treePatternParts [ i ].empty ( ) )
- prevOcc = MergeOccurrences ( prevOcc, FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
+ prevOcc = MergeOccurrences ( prevOcc, NonlinearFullAndLinearIndexPatterns::FindOccurrences < StringIndexQueryAlgo > ( fullAndLinearIndex.getStringIndex ( ), treePatternParts [ i ] ), rev );
}
ext::set < unsigned > res;
diff --git a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp
index e0081a874e..76bd9f35be 100644
--- a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp
+++ b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.cpp
@@ -7,6 +7,6 @@
namespace {
-auto SuffixAutomatonLinearString = registration::AbstractRegister < stringology::indexing::ExactSuffixAutomaton, automaton::DFA < DefaultSymbolType, unsigned >, const string::LinearString < > & > ( stringology::indexing::ExactSuffixAutomaton::construct );
+auto SuffixAutomatonLinearString = registration::AbstractRegister < stringology::indexing::ExactSuffixAutomaton, indexes::stringology::SuffixAutomaton < >, const string::LinearString < > & > ( stringology::indexing::ExactSuffixAutomaton::construct );
} /* namespace */
diff --git a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
index b057bc747c..64c524cfb9 100644
--- a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
+++ b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
@@ -5,7 +5,7 @@
#ifndef _EXACT_SUFFIX_AUTOMATON_H_
#define _EXACT_SUFFIX_AUTOMATON_H_
-#include <automaton/FSM/DFA.h>
+#include <indexes/stringology/SuffixAutomaton.h>
#include <string/LinearString.h>
#include <global/GlobalData.h>
@@ -21,12 +21,12 @@ private:
public:
template < class SymbolType >
- static automaton::DFA < SymbolType, unsigned > construct ( const string::LinearString < SymbolType > & pattern );
+ static indexes::stringology::SuffixAutomaton < SymbolType > construct ( const string::LinearString < SymbolType > & pattern );
};
template < class SymbolType >
-automaton::DFA < SymbolType, unsigned > ExactSuffixAutomaton::construct ( const string::LinearString < SymbolType > & pattern ) {
+indexes::stringology::SuffixAutomaton < SymbolType > ExactSuffixAutomaton::construct ( const string::LinearString < SymbolType > & pattern ) {
automaton::DFA < SymbolType, unsigned > suffixAutomaton ( 0 );
suffixAutomaton.setInputAlphabet ( pattern.getAlphabet ( ) );
@@ -49,7 +49,7 @@ automaton::DFA < SymbolType, unsigned > ExactSuffixAutomaton::construct ( const
lastState = suffixLinks [ lastState ].first;
}
- return suffixAutomaton;
+ return indexes::stringology::SuffixAutomaton < SymbolType > ( std::move ( suffixAutomaton ) );
}
template < class SymbolType >
diff --git a/alib2algo/src/stringology/matching/DAWGMatcherConstruction.cpp b/alib2algo/src/stringology/matching/DAWGMatcherConstruction.cpp
index 90231dfb04..7c537890ff 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 < DefaultSymbolType, unsigned > DAWGMatcherConstruction::construct ( const string::LinearString < > & pattern ) {
+indexes::stringology::SuffixAutomaton < DefaultSymbolType > 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 < DefaultSymbolType, unsigned > DAWGMatcherConstruction::construc
return stringology::indexing::ExactSuffixAutomaton::construct ( reversedPattern );
}
-auto DAWGMatcherConstructionLinearString = registration::AbstractRegister < DAWGMatcherConstruction, automaton::DFA < DefaultSymbolType, unsigned >, const string::LinearString < > & > ( DAWGMatcherConstruction::construct );
+auto DAWGMatcherConstructionLinearString = registration::AbstractRegister < DAWGMatcherConstruction, indexes::stringology::SuffixAutomaton < DefaultSymbolType >, const string::LinearString < > & > ( DAWGMatcherConstruction::construct );
} /* namespace matching */
diff --git a/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h b/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h
index b3f6392ee5..0ddb9eedc9 100644
--- a/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h
+++ b/alib2algo/src/stringology/matching/DAWGMatcherConstruction.h
@@ -5,7 +5,7 @@
#ifndef DAWG_MATCHER_CONSTRUCTION_H_
#define DAWG_MATCHER_CONSTRUCTION_H_
-#include <automaton/FSM/DFA.h>
+#include <indexes/stringology/SuffixAutomaton.h>
#include <string/LinearString.h>
namespace stringology {
@@ -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 < DefaultSymbolType, unsigned > construct ( const string::LinearString < > & pattern );
+ static indexes::stringology::SuffixAutomaton < > construct ( const string::LinearString < > & pattern );
};
diff --git a/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp b/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp
index 8f27209ffa..d56866487e 100644
--- a/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp
+++ b/alib2algo/src/stringology/matching/OracleMatcherConstruction.cpp
@@ -15,7 +15,7 @@ automaton::DFA < DefaultSymbolType, unsigned > OracleMatcherConstruction::constr
reverse ( patternData.begin ( ), patternData.end ( ) );
string::LinearString < > reversedPattern ( pattern.getAlphabet ( ), std::move ( patternData ) );
- return stringology::indexing::ExactSuffixAutomaton::construct ( reversedPattern );
+ return stringology::indexing::ExactSuffixAutomaton::construct ( reversedPattern ).getAutomaton ( ); // FIXME this is not oracle automaton
}
auto OracleMatcherConstructionLinearString = registration::AbstractRegister < OracleMatcherConstruction, automaton::DFA < DefaultSymbolType, unsigned >, const string::LinearString < > & > ( OracleMatcherConstruction::construct );
diff --git a/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp b/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
index a1ea591b82..e4ff9a4195 100644
--- a/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
+++ b/alib2algo/src/stringology/query/BackwardDAWGMatching.cpp
@@ -9,7 +9,7 @@ namespace stringology {
namespace query {
-auto BackwardDAWGMatchingLinearStringLinearString = registration::AbstractRegister < BackwardDAWGMatching, ext::set < unsigned >, const string::LinearString < > &, const automaton::DFA < > & > ( BackwardDAWGMatching::match );
+auto BackwardDAWGMatchingLinearStringLinearString = registration::AbstractRegister < BackwardDAWGMatching, ext::set < unsigned >, const string::LinearString < > &, const indexes::stringology::SuffixAutomaton < > & > ( BackwardDAWGMatching::match );
} /* namespace query */
diff --git a/alib2algo/src/stringology/query/BackwardDAWGMatching.h b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
index 57b49348de..9094f72e2b 100644
--- a/alib2algo/src/stringology/query/BackwardDAWGMatching.h
+++ b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
@@ -7,7 +7,7 @@
#include <alib/set>
-#include <automaton/FSM/DFA.h>
+#include <indexes/stringology/SuffixAutomaton.h>
#include <string/LinearString.h>
#include <stringology/properties/BackboneLength.h>
@@ -25,32 +25,32 @@ public:
* Search for pattern in linear string.
* @return set set of occurences
*/
- template < class SymbolType, class StateType >
- static ext::set < unsigned > match ( const string::LinearString < SymbolType > & subject, const automaton::DFA < SymbolType, StateType > & suffixAutomaton );
+ template < class SymbolType >
+ static ext::set < unsigned > match ( const string::LinearString < SymbolType > & subject, const indexes::stringology::SuffixAutomaton < SymbolType > & suffixAutomaton );
};
-template < class SymbolType, class StateType >
-ext::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString < SymbolType > & subject, const automaton::DFA < SymbolType, StateType > & suffixAutomaton ) {
+template < class SymbolType >
+ext::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString < SymbolType > & subject, const indexes::stringology::SuffixAutomaton < SymbolType > & suffixAutomaton ) {
ext::set < unsigned > occ;
- size_t patternSize = stringology::properties::BackboneLength::length ( suffixAutomaton );
+ size_t patternSize = stringology::properties::BackboneLength::length ( suffixAutomaton.getAutomaton ( ) );
size_t subjectSize = subject.getContent ( ).size ( );
bool fail;
size_t posInSubject = 0;
while ( posInSubject <= subjectSize - patternSize ) {
- StateType currentState = suffixAutomaton.getInitialState ( );
+ unsigned currentState = suffixAutomaton.getAutomaton ( ).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 ) } );
+ auto transition = suffixAutomaton.getAutomaton ( ).getTransitions ( ).find ( { currentState, subject.getContent ( ).at ( posInSubject + posInPattern - 1 ) } );
- if ( transition == suffixAutomaton.getTransitions ( ).end ( ) )
+ if ( transition == suffixAutomaton.getAutomaton ( ).getTransitions ( ).end ( ) )
fail = true;
else
currentState = transition->second;
@@ -58,7 +58,7 @@ ext::set < unsigned > BackwardDAWGMatching::match ( const string::LinearString <
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 ( ) ) )
+ if ( ( posInPattern != 0 ) && ( suffixAutomaton.getAutomaton ( ).getFinalStates ( ).find ( currentState ) != suffixAutomaton.getAutomaton ( ).getFinalStates ( ).end ( ) ) )
lastPrefixPos = posInPattern;
}
diff --git a/alib2algo/src/stringology/query/SuffixAutomatonFactors.cpp b/alib2algo/src/stringology/query/SuffixAutomatonFactors.cpp
index 7d457596d7..8abba44d24 100644
--- a/alib2algo/src/stringology/query/SuffixAutomatonFactors.cpp
+++ b/alib2algo/src/stringology/query/SuffixAutomatonFactors.cpp
@@ -12,7 +12,7 @@ namespace stringology {
namespace query {
-auto SuffixAutomatonFactorsLinearString = registration::AbstractRegister < SuffixAutomatonFactors, ext::set < unsigned >, const automaton::DFA < > &, const string::LinearString < > & > ( SuffixAutomatonFactors::query );
+auto SuffixAutomatonFactorsLinearString = registration::AbstractRegister < SuffixAutomatonFactors, ext::set < unsigned >, const indexes::stringology::SuffixAutomaton < > &, const string::LinearString < > & > ( SuffixAutomatonFactors::query );
} /* namespace query */
diff --git a/alib2algo/src/stringology/query/SuffixAutomatonFactors.h b/alib2algo/src/stringology/query/SuffixAutomatonFactors.h
index 42a0340b78..978a86b65f 100644
--- a/alib2algo/src/stringology/query/SuffixAutomatonFactors.h
+++ b/alib2algo/src/stringology/query/SuffixAutomatonFactors.h
@@ -8,7 +8,7 @@
#ifndef SUFFIX_AUTOMATON_FACTORS_H_
#define SUFFIX_AUTOMATON_FACTORS_H_
-#include <automaton/FSM/DFA.h>
+#include <indexes/stringology/SuffixAutomaton.h>
#include <string/LinearString.h>
#include <automaton/run/Run.h>
@@ -32,31 +32,30 @@ public:
* @param string string to query by
* @return occurences of factors
*/
- template < class SymbolType, class StateType >
- static ext::set < unsigned > query ( const automaton::DFA < SymbolType, StateType > & suffixAutomaton, const string::LinearString < SymbolType > & string );
-
+ template < class SymbolType >
+ static ext::set < unsigned > query ( const indexes::stringology::SuffixAutomaton < SymbolType > & suffixAutomaton, const string::LinearString < SymbolType > & string );
};
-template < class SymbolType, class StateType >
-ext::set < unsigned > SuffixAutomatonFactors::query ( const automaton::DFA < SymbolType, StateType > & suffixAutomaton, const string::LinearString < SymbolType > & string ) {
- ext::tuple < bool, StateType, ext::set < unsigned > > run = automaton::run::Run::calculateState ( suffixAutomaton, string );
+template < class SymbolType >
+ext::set < unsigned > SuffixAutomatonFactors::query ( const indexes::stringology::SuffixAutomaton < SymbolType > & suffixAutomaton, const string::LinearString < SymbolType > & string ) {
+ ext::tuple < bool, unsigned, ext::set < unsigned > > run = automaton::run::Run::calculateState ( suffixAutomaton.getAutomaton ( ), string );
if ( ! std::get < 0 > ( run ) )
return { };
- std::deque < std::pair < StateType, unsigned > > open = { { std::get < 1 > ( run ), 0u } };
+ std::deque < std::pair < unsigned, unsigned > > open = { { std::get < 1 > ( run ), 0u } };
ext::vector < unsigned > tmp;
unsigned max = 0;
while ( ! open.empty ( ) ) {
- std::pair < StateType, unsigned > cur = std::move ( open.back ( ) );
+ std::pair < unsigned, unsigned > cur = std::move ( open.back ( ) );
open.pop_back ( );
- if ( suffixAutomaton.getFinalStates ( ).count ( cur.first ) )
+ if ( suffixAutomaton.getAutomaton ( ).getFinalStates ( ).count ( cur.first ) )
tmp.push_back ( cur.second );
if ( cur.second > max )
max = cur.second;
- for ( const auto & transition : suffixAutomaton.getTransitionsFromState ( cur.first ) )
+ for ( const auto & transition : suffixAutomaton.getAutomaton ( ).getTransitionsFromState ( cur.first ) )
open.emplace_back ( transition.second, cur.second + 1 );
}
diff --git a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
index 61587da32c..d5f5c279f0 100644
--- a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
@@ -16,29 +16,31 @@ void ExactSuffixAutomatonTest::testSuffixAutomatonConstruction ( ) {
string::LinearString < > pattern ( "atatac" );
- automaton::DFA < DefaultSymbolType, unsigned > suffixAutomaton = stringology::indexing::ExactSuffixAutomaton::construct ( pattern );
+ indexes::stringology::SuffixAutomaton < > suffixAutomaton = stringology::indexing::ExactSuffixAutomaton::construct ( pattern );
- automaton::DFA < DefaultSymbolType, unsigned > refSuffixAutomaton ( 0 );
+ automaton::DFA < DefaultSymbolType, unsigned > tmp ( 0 );
- refSuffixAutomaton.setInputAlphabet ( pattern.getAlphabet ( ) );
+ tmp.setInputAlphabet ( pattern.getAlphabet ( ) );
for ( unsigned i = 1; i <= 6; ++i ) {
- refSuffixAutomaton.addState ( i );
+ tmp.addState ( i );
}
- 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 );
+ tmp.addFinalState ( 0 );
+ tmp.addFinalState ( 6 );
+
+ tmp.addTransition ( 0, DefaultSymbolType ( 'a' ), 1 );
+ tmp.addTransition ( 0, DefaultSymbolType ( 't' ), 2 );
+ tmp.addTransition ( 0, DefaultSymbolType ( 'c' ), 6 );
+ tmp.addTransition ( 1, DefaultSymbolType ( 't' ), 2 );
+ tmp.addTransition ( 1, DefaultSymbolType ( 'c' ), 6 );
+ tmp.addTransition ( 2, DefaultSymbolType ( 'a' ), 3 );
+ tmp.addTransition ( 3, DefaultSymbolType ( 't' ), 4 );
+ tmp.addTransition ( 3, DefaultSymbolType ( 'c' ), 6 );
+ tmp.addTransition ( 4, DefaultSymbolType ( 'a' ), 5 );
+ tmp.addTransition ( 5, DefaultSymbolType ( 'c' ), 6 );
+
+ indexes::stringology::SuffixAutomaton < > refSuffixAutomaton ( std::move ( tmp ) );
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 254c39a8ca..0aafbfbb75 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 < DefaultSymbolType, unsigned > suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( pattern );
+ indexes::stringology::SuffixAutomaton < > 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 < DefaultSymbolType, unsigned > suffixAutomaton = stringology::matching::DAWGMatcherConstruction::construct ( longPattern );
+ indexes::stringology::SuffixAutomaton < > 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/alib2data/src/indexes/stringology/SuffixAutomaton.cpp b/alib2data/src/indexes/stringology/SuffixAutomaton.cpp
new file mode 100644
index 0000000000..691a72c799
--- /dev/null
+++ b/alib2data/src/indexes/stringology/SuffixAutomaton.cpp
@@ -0,0 +1,22 @@
+/*
+ * SuffixAutomaton.cpp
+ *
+ * Created on: Jan 8, 2017
+ * Author: Jan Travnicek
+ */
+
+#include "SuffixAutomaton.h"
+
+#include <registration/ValuePrinterRegistration.hpp>
+#include <registration/XmlRegistration.hpp>
+
+namespace {
+
+static auto valuePrinter = registration::ValuePrinterRegister < indexes::stringology::SuffixAutomaton < > > ( );
+
+static auto xmlWrite = registration::XmlWriterRegister < indexes::stringology::SuffixAutomaton < > > ( );
+static auto xmlRead = registration::XmlReaderRegister < indexes::stringology::SuffixAutomaton < > > ( );
+
+static auto xmlGroup = registration::XmlRegisterTypeInGroup < object::Object, indexes::stringology::SuffixAutomaton < > > ( );
+
+} /* namespace */
diff --git a/alib2data/src/indexes/stringology/SuffixAutomaton.h b/alib2data/src/indexes/stringology/SuffixAutomaton.h
new file mode 100644
index 0000000000..40a9b597aa
--- /dev/null
+++ b/alib2data/src/indexes/stringology/SuffixAutomaton.h
@@ -0,0 +1,217 @@
+/*
+ * SuffixAutomaton.h
+ *
+ * Created on: Jan 8, 2017
+ * Author: Jan Travnicek
+ */
+
+#ifndef SUFFIX_AUTOMATON_H_
+#define SUFFIX_AUTOMATON_H_
+
+#include <alib/string>
+#include <alib/set>
+#include <alib/iostream>
+#include <sstream>
+
+#include <common/DefaultSymbolType.h>
+
+#include <object/UniqueObject.h>
+#include <object/ObjectBase.h>
+
+#include <sax/FromXMLParserHelper.h>
+#include <core/xmlApi.hpp>
+#include <primitive/Unsigned.h>
+#include <primitive/xml/Unsigned.h>
+
+#include <alphabet/common/SymbolNormalize.h>
+
+#include <automaton/FSM/DFA.h>
+
+namespace indexes {
+
+namespace stringology {
+
+class GeneralAlphabet;
+
+/**
+ * Represents regular expression parsed from the XML. Regular expression is stored
+ * as a tree of RegExpElement.
+ */
+template < class SymbolType = DefaultSymbolType >
+class SuffixAutomaton final : public object::ObjectBase {
+protected:
+ automaton::DFA < SymbolType, unsigned > m_automaton;
+
+public:
+ /**
+ * @copydoc SuffixTrieNode::clone() const
+ */
+ virtual ObjectBase * clone ( ) const;
+
+ /**
+ * @copydoc SuffixTrieNode::plunder() const
+ */
+ virtual ObjectBase * plunder ( ) &&;
+
+ explicit SuffixAutomaton ( automaton::DFA < SymbolType, unsigned > automaton );
+
+ /**
+ * @return Root node of the trie
+ */
+ const automaton::DFA < SymbolType, unsigned > & getAutomaton ( ) const &;
+
+ automaton::DFA < SymbolType, unsigned > && getAutomaton ( ) &&;
+
+ const ext::set < SymbolType > & getAlphabet ( ) const & {
+ return m_automaton.getInputAlphabet ( );
+ }
+
+ ext::set < SymbolType > && getAlphabet ( ) && {
+ return std::move ( m_automaton ).getInputAlphabet ( );
+ }
+
+ /**
+ * Removes symbol from the alphabet of symbol available in the regular expression
+ * @param symbol removed symbol from the alphabet
+ */
+ bool removeSymbolFromAlphabet ( const SymbolType & symbol ) {
+ return m_automaton.removeInputSymbol ( symbol );
+ }
+
+ /**
+ * Prints XML representation of the tree to the output stream.
+ * @param out output stream to which print the tree
+ * @param tree tree to print
+ */
+ virtual void operator >>( std::ostream & out ) const;
+
+ virtual int compare ( const ObjectBase & other ) const {
+ if ( ext::type_index ( typeid ( * this ) ) == ext::type_index ( typeid ( other ) ) ) return this->compare ( ( decltype ( * this ) )other );
+
+ return ext::type_index ( typeid ( * this ) ) - ext::type_index ( typeid ( other ) );
+ }
+
+ virtual int compare ( const SuffixAutomaton & other ) const;
+
+ virtual explicit operator std::string ( ) const;
+
+ virtual object::ObjectBase * inc ( ) &&;
+
+ typedef SuffixAutomaton < > normalized_type;
+};
+
+} /* namespace stringology */
+
+} /* namespace indexes */
+
+namespace indexes {
+
+namespace stringology {
+
+template < class SymbolType >
+SuffixAutomaton < SymbolType >::SuffixAutomaton ( automaton::DFA < SymbolType, unsigned > automaton ) : m_automaton ( std::move ( automaton ) ) {
+}
+
+template < class SymbolType >
+object::ObjectBase * SuffixAutomaton < SymbolType >::clone ( ) const {
+ return new SuffixAutomaton ( * this );
+}
+
+template < class SymbolType >
+object::ObjectBase * SuffixAutomaton < SymbolType >::plunder ( ) && {
+ return new SuffixAutomaton ( std::move ( * this ) );
+}
+
+template < class SymbolType >
+const automaton::DFA < SymbolType, unsigned > & SuffixAutomaton < SymbolType >::getAutomaton ( ) const & {
+ return m_automaton;
+}
+
+template < class SymbolType >
+automaton::DFA < SymbolType, unsigned > && SuffixAutomaton < SymbolType >::getAutomaton ( ) && {
+ return std::move ( m_automaton );
+}
+
+template < class SymbolType >
+void SuffixAutomaton < SymbolType >::operator >>( std::ostream & out ) const {
+ out << "(SuffixAutomaton " << this->m_automaton << ")";
+}
+
+template < class SymbolType >
+int SuffixAutomaton < SymbolType >::compare ( const SuffixAutomaton & other ) const {
+ auto first = ext::tie ( getAutomaton ( ) );
+ auto second = ext::tie ( other.getAutomaton ( ) );
+
+ static ext::compare < decltype ( first ) > comp;
+
+ return comp ( first, second );
+}
+
+template < class SymbolType >
+SuffixAutomaton < SymbolType >::operator std::string ( ) const {
+ std::stringstream ss;
+ ss << * this;
+ return ss.str ( );
+}
+
+template < class SymbolType >
+object::ObjectBase* SuffixAutomaton < SymbolType >::inc() && {
+ return new object::UniqueObject(object::Object(std::move(*this)), primitive::Integer(0));
+}
+
+} /* namespace stringology */
+
+} /* namespace indexes */
+
+namespace core {
+
+template < class SymbolType >
+struct normalize < indexes::stringology::SuffixAutomaton < SymbolType >, typename std::enable_if < ! std::is_same < indexes::stringology::SuffixAutomaton < SymbolType >, indexes::stringology::SuffixAutomaton < > >::value >::type > {
+ static indexes::stringology::SuffixAutomaton < > eval ( indexes::stringology::SuffixAutomaton < SymbolType > && value ) {
+ ext::set < DefaultSymbolType > alphabet = alphabet::SymbolNormalize::normalizeAlphabet ( std::move ( value ).getAlphabet ( ) );
+ ext::vector < DefaultSymbolType > string = alphabet::SymbolNormalize::normalizeSymbols ( std::move ( value ).getString ( ) );
+
+ return indexes::stringology::SuffixAutomaton < > ( std::move ( alphabet ), std::move ( value ).getData ( ), std::move ( string ) );
+ }
+};
+
+template < class SymbolType >
+struct xmlApi < indexes::stringology::SuffixAutomaton < SymbolType > > {
+ static indexes::stringology::SuffixAutomaton < SymbolType > parse ( ext::deque < sax::Token >::iterator & input );
+ static bool first ( const ext::deque < sax::Token >::const_iterator & input );
+ static const std::string & xmlTagName ( );
+ static void compose ( ext::deque < sax::Token > & output, const indexes::stringology::SuffixAutomaton < SymbolType > & data );
+};
+
+template < class SymbolType >
+indexes::stringology::SuffixAutomaton < SymbolType > xmlApi < indexes::stringology::SuffixAutomaton < SymbolType > >::parse ( ext::deque < sax::Token >::iterator & input ) {
+ sax::FromXMLParserHelper::popToken ( input, sax::Token::TokenType::START_ELEMENT, xmlTagName ( ) );
+ automaton::DFA < SymbolType, unsigned > automaton = core::xmlApi < automaton::DFA < SymbolType, unsigned > >::parse ( input );
+ indexes::stringology::SuffixAutomaton < SymbolType > res ( std::move ( automaton ) );
+
+ sax::FromXMLParserHelper::popToken ( input, sax::Token::TokenType::END_ELEMENT, xmlTagName ( ) );
+ return res;
+}
+
+template < class SymbolType >
+bool xmlApi < indexes::stringology::SuffixAutomaton < SymbolType > >::first ( const ext::deque < sax::Token >::const_iterator & input ) {
+ return sax::FromXMLParserHelper::isToken ( input, sax::Token::TokenType::START_ELEMENT, xmlTagName ( ) );
+}
+
+template < class SymbolType >
+const std::string & xmlApi < indexes::stringology::SuffixAutomaton < SymbolType > >::xmlTagName ( ) {
+ static std::string xmlTagName = "SuffixAutomaton";
+
+ return xmlTagName;
+}
+
+template < class SymbolType >
+void xmlApi < indexes::stringology::SuffixAutomaton < SymbolType > >::compose ( ext::deque < sax::Token > & output, const indexes::stringology::SuffixAutomaton < SymbolType > & index ) {
+ output.emplace_back ( xmlTagName ( ), sax::Token::TokenType::START_ELEMENT );
+ core::xmlApi < automaton::DFA < SymbolType, unsigned > >::compose ( output, index.getAutomaton ( ) );
+ output.emplace_back ( xmlTagName ( ), sax::Token::TokenType::END_ELEMENT );
+}
+
+} /* namespace core */
+
+#endif /* SUFFIX_AUTOMATON_H_ */
--
GitLab