From 997692cb458db2bae48758f746f5f698bf3a15f2 Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Tue, 13 Feb 2018 12:31:14 +0100
Subject: [PATCH] update suffix automaton and factor oracle indexes
---
.../indexing/ExactFactorOracleAutomaton.h | 2 +-
.../indexing/ExactSuffixAutomatonTest.cpp | 30 +++++++++---------
.../indexing/FactorOracleAutomatonTest.cpp | 30 +++++++++---------
.../stringology/FactorOracleAutomaton.h | 31 ++++++++++++-------
.../src/indexes/stringology/SuffixAutomaton.h | 19 +++++++++---
5 files changed, 67 insertions(+), 45 deletions(-)
diff --git a/alib2algo/src/stringology/indexing/ExactFactorOracleAutomaton.h b/alib2algo/src/stringology/indexing/ExactFactorOracleAutomaton.h
index fb2e2cb695..c2092b9834 100644
--- a/alib2algo/src/stringology/indexing/ExactFactorOracleAutomaton.h
+++ b/alib2algo/src/stringology/indexing/ExactFactorOracleAutomaton.h
@@ -40,7 +40,7 @@ indexes::stringology::FactorOracleAutomaton < SymbolType > ExactFactorOracleAuto
for ( const SymbolType & symbol : pattern.getContent ( ) )
oracleAddLetter ( oracleAutomaton, symbol, supplyFunction );
- return indexes::stringology::FactorOracleAutomaton < SymbolType > ( std::move ( oracleAutomaton ), pattern.getContent ( ).size ( ) );
+ return indexes::stringology::FactorOracleAutomaton < SymbolType > ( std::move ( oracleAutomaton ) );
}
template < class SymbolType >
diff --git a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
index e18f732f13..1a76d9579e 100644
--- a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
@@ -14,11 +14,11 @@ void ExactSuffixAutomatonTest::tearDown ( ) {
void ExactSuffixAutomatonTest::testSuffixAutomatonConstruction ( ) {
- string::LinearString < > pattern ( "atatac" );
+ string::LinearString < char > pattern ( "atatac" );
- indexes::stringology::SuffixAutomaton < > suffixAutomaton = stringology::indexing::ExactSuffixAutomaton::construct ( pattern );
+ indexes::stringology::SuffixAutomaton < char > suffixAutomaton = stringology::indexing::ExactSuffixAutomaton::construct ( pattern );
- automaton::DFA < DefaultSymbolType, unsigned > tmp ( 0 );
+ automaton::DFA < char, unsigned > tmp ( 0 );
tmp.setInputAlphabet ( pattern.getAlphabet ( ) );
@@ -29,21 +29,23 @@ void ExactSuffixAutomatonTest::testSuffixAutomatonConstruction ( ) {
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 );
+ tmp.addTransition ( 0, 'a', 1 );
+ tmp.addTransition ( 0, 't', 2 );
+ tmp.addTransition ( 0, 'c', 6 );
+ tmp.addTransition ( 1, 't', 2 );
+ tmp.addTransition ( 1, 'c', 6 );
+ tmp.addTransition ( 2, 'a', 3 );
+ tmp.addTransition ( 3, 't', 4 );
+ tmp.addTransition ( 3, 'c', 6 );
+ tmp.addTransition ( 4, 'a', 5 );
+ tmp.addTransition ( 5, 'c', 6 );
- indexes::stringology::SuffixAutomaton < > refSuffixAutomaton ( std::move ( tmp ), 6 );
+ indexes::stringology::SuffixAutomaton < char > refSuffixAutomaton ( std::move ( tmp ), 6 );
std::cout << suffixAutomaton << std::endl;
std::cout << refSuffixAutomaton << std::endl;
CPPUNIT_ASSERT ( suffixAutomaton == refSuffixAutomaton );
+
+ core::normalize < indexes::stringology::SuffixAutomaton < char > >::eval ( std::move ( refSuffixAutomaton ) );
}
diff --git a/alib2algo/test-src/stringology/indexing/FactorOracleAutomatonTest.cpp b/alib2algo/test-src/stringology/indexing/FactorOracleAutomatonTest.cpp
index 495dce3da0..8a7c184ef3 100644
--- a/alib2algo/test-src/stringology/indexing/FactorOracleAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/indexing/FactorOracleAutomatonTest.cpp
@@ -13,11 +13,11 @@ void FactorOracleAutomatonTest::tearDown ( ) {
}
void FactorOracleAutomatonTest::testFactorOracleConstruction ( ) {
- string::LinearString < > pattern ( "atatac" );
+ string::LinearString < char > pattern ( "atatac" );
- indexes::stringology::FactorOracleAutomaton < > oracle = stringology::indexing::ExactFactorOracleAutomaton::construct ( pattern );
+ indexes::stringology::FactorOracleAutomaton < char > oracle = stringology::indexing::ExactFactorOracleAutomaton::construct ( pattern );
- automaton::DFA < DefaultSymbolType, unsigned > tmp ( 0 );
+ automaton::DFA < char, unsigned > tmp ( 0 );
tmp.addFinalState ( 0 );
@@ -28,18 +28,20 @@ void FactorOracleAutomatonTest::testFactorOracleConstruction ( ) {
tmp.addFinalState ( i );
}
- 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 );
+ tmp.addTransition ( 0, 'a', 1 );
+ tmp.addTransition ( 0, 't', 2 );
+ tmp.addTransition ( 0, 'c', 6 );
+ tmp.addTransition ( 1, 't', 2 );
+ tmp.addTransition ( 1, 'c', 6 );
+ tmp.addTransition ( 2, 'a', 3 );
+ tmp.addTransition ( 3, 't', 4 );
+ tmp.addTransition ( 3, 'c', 6 );
+ tmp.addTransition ( 4, 'a', 5 );
+ tmp.addTransition ( 5, 'c', 6 );
- indexes::stringology::FactorOracleAutomaton < > refOracle ( std::move ( tmp ), 6 );
+ indexes::stringology::FactorOracleAutomaton < char > refOracle ( std::move ( tmp ) );
CPPUNIT_ASSERT ( oracle == refOracle );
+
+ core::normalize < indexes::stringology::FactorOracleAutomaton < char > >::eval ( std::move ( refOracle ) );
}
diff --git a/alib2data/src/indexes/stringology/FactorOracleAutomaton.h b/alib2data/src/indexes/stringology/FactorOracleAutomaton.h
index d73668ef81..0058f23701 100644
--- a/alib2data/src/indexes/stringology/FactorOracleAutomaton.h
+++ b/alib2data/src/indexes/stringology/FactorOracleAutomaton.h
@@ -42,8 +42,6 @@ class FactorOracleAutomaton final : public object::ObjectBase {
protected:
automaton::DFA < SymbolType, unsigned > m_automaton;
- unsigned m_backboneLength;
-
public:
/**
* @copydoc OracleTrieNode::clone() const
@@ -55,7 +53,7 @@ public:
*/
virtual ObjectBase * plunder ( ) &&;
- explicit FactorOracleAutomaton ( automaton::DFA < SymbolType, unsigned > automaton, unsigned backboneLength );
+ explicit FactorOracleAutomaton ( automaton::DFA < SymbolType, unsigned > automaton );
/**
* @return Root node of the trie
@@ -81,7 +79,7 @@ public:
}
unsigned getBackboneLength ( ) const {
- return m_backboneLength;
+ return m_automaton.getStates ( ).size ( ) - 1;
}
/**
@@ -117,7 +115,7 @@ namespace indexes {
namespace stringology {
template < class SymbolType >
-FactorOracleAutomaton < SymbolType >::FactorOracleAutomaton ( automaton::DFA < SymbolType, unsigned > automaton, unsigned backboneLength ) : m_automaton ( std::move ( automaton ) ), m_backboneLength ( backboneLength ) {
+FactorOracleAutomaton < SymbolType >::FactorOracleAutomaton ( automaton::DFA < SymbolType, unsigned > automaton ) : m_automaton ( std::move ( automaton ) ) {
}
template < class SymbolType >
@@ -181,11 +179,22 @@ namespace core {
template < class SymbolType >
struct normalize < indexes::stringology::FactorOracleAutomaton < SymbolType >, typename std::enable_if < ! std::is_same < indexes::stringology::FactorOracleAutomaton < SymbolType >, indexes::stringology::FactorOracleAutomaton < > >::value >::type > {
static indexes::stringology::FactorOracleAutomaton < > eval ( indexes::stringology::FactorOracleAutomaton < SymbolType > && value ) {
- // FIXME this does not look right
- ext::set < DefaultSymbolType > alphabet = alphabet::SymbolNormalize::normalizeAlphabet ( std::move ( value ).getAlphabet ( ) );
- ext::vector < DefaultSymbolType > string = alphabet::SymbolNormalize::normalizeSymbols ( std::move ( value ).getString ( ) );
+ ext::set < DefaultSymbolType > alphabet = alphabet::SymbolNormalize::normalizeAlphabet ( std::move ( std::move ( value ).getAutomaton ( ) ).getInputAlphabet ( ) );
+ ext::set < unsigned > states = std::move ( std::move ( value ).getAutomaton ( ) ).getStates ( );
+ unsigned initialState = std::move ( std::move ( value ).getAutomaton ( ) ).getInitialState ( );
+ ext::set < unsigned > finalStates = std::move ( std::move ( value ).getAutomaton ( ) ).getFinalStates ( );
+
+ automaton::DFA < DefaultStateType, unsigned > res ( std::move ( states ), std::move ( alphabet ), std::move ( initialState ), std::move ( finalStates ) );
+
+ for ( std::pair < ext::pair < unsigned, SymbolType >, unsigned > && transition : ext::make_moveable_map ( std::move ( std::move ( value ).getAutomaton ( ) ).getTransitions ( ) ) ) {
+ unsigned from = transition.first.first;
+ DefaultSymbolType input = alphabet::SymbolNormalize::normalizeSymbol ( std::move ( transition.first.second ) );
+ unsigned to = transition.second;
+
+ res.addTransition ( std::move ( from ), std::move ( input ), std::move ( to ) );
+ }
- return indexes::stringology::FactorOracleAutomaton < > ( std::move ( alphabet ), std::move ( value ).getData ( ), std::move ( string ) );
+ return indexes::stringology::FactorOracleAutomaton < > ( std::move ( res ) );
}
};
@@ -201,8 +210,7 @@ template < class SymbolType >
indexes::stringology::FactorOracleAutomaton < SymbolType > xmlApi < indexes::stringology::FactorOracleAutomaton < 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 );
- unsigned backboneLength = core::xmlApi < unsigned >::parse ( input );
- indexes::stringology::FactorOracleAutomaton < SymbolType > res ( std::move ( automaton ), backboneLength );
+ indexes::stringology::FactorOracleAutomaton < SymbolType > res ( std::move ( automaton ) );
sax::FromXMLParserHelper::popToken ( input, sax::Token::TokenType::END_ELEMENT, xmlTagName ( ) );
return res;
@@ -224,7 +232,6 @@ template < class SymbolType >
void xmlApi < indexes::stringology::FactorOracleAutomaton < SymbolType > >::compose ( ext::deque < sax::Token > & output, const indexes::stringology::FactorOracleAutomaton < SymbolType > & index ) {
output.emplace_back ( xmlTagName ( ), sax::Token::TokenType::START_ELEMENT );
core::xmlApi < automaton::DFA < SymbolType, unsigned > >::compose ( output, index.getAutomaton ( ) );
- core::xmlApi < unsigned >::compose ( output, index.getBackboneLength ( ) );
output.emplace_back ( xmlTagName ( ), sax::Token::TokenType::END_ELEMENT );
}
diff --git a/alib2data/src/indexes/stringology/SuffixAutomaton.h b/alib2data/src/indexes/stringology/SuffixAutomaton.h
index 5a32a1369a..18b5056771 100644
--- a/alib2data/src/indexes/stringology/SuffixAutomaton.h
+++ b/alib2data/src/indexes/stringology/SuffixAutomaton.h
@@ -181,11 +181,22 @@ 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 ) {
- // FIXME this does not look right
- ext::set < DefaultSymbolType > alphabet = alphabet::SymbolNormalize::normalizeAlphabet ( std::move ( value ).getAlphabet ( ) );
- ext::vector < DefaultSymbolType > string = alphabet::SymbolNormalize::normalizeSymbols ( std::move ( value ).getString ( ) );
+ ext::set < DefaultSymbolType > alphabet = alphabet::SymbolNormalize::normalizeAlphabet ( std::move ( std::move ( value ).getAutomaton ( ) ).getInputAlphabet ( ) );
+ ext::set < unsigned > states = std::move ( std::move ( value ).getAutomaton ( ) ).getStates ( );
+ unsigned initialState = std::move ( std::move ( value ).getAutomaton ( ) ).getInitialState ( );
+ ext::set < unsigned > finalStates = std::move ( std::move ( value ).getAutomaton ( ) ).getFinalStates ( );
- return indexes::stringology::SuffixAutomaton < > ( std::move ( alphabet ), std::move ( value ).getData ( ), std::move ( string ) );
+ automaton::DFA < DefaultStateType, unsigned > res ( std::move ( states ), std::move ( alphabet ), std::move ( initialState ), std::move ( finalStates ) );
+
+ for ( std::pair < ext::pair < unsigned, SymbolType >, unsigned > && transition : ext::make_moveable_map ( std::move ( std::move ( value ).getAutomaton ( ) ).getTransitions ( ) ) ) {
+ unsigned from = transition.first.first;
+ DefaultSymbolType input = alphabet::SymbolNormalize::normalizeSymbol ( std::move ( transition.first.second ) );
+ unsigned to = transition.second;
+
+ res.addTransition ( std::move ( from ), std::move ( input ), std::move ( to ) );
+ }
+
+ return indexes::stringology::SuffixAutomaton < > ( std::move ( res ), std::move ( value ).getBackboneLength ( ) );
}
};
--
GitLab