diff --git a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
index 64c524cfb96f8a3438c772d74ba9884b2bbd7333..3b4681c5a2e114fcfae1c11cbb81b692f9872082 100644
--- a/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
+++ b/alib2algo/src/stringology/indexing/ExactSuffixAutomaton.h
@@ -49,7 +49,7 @@ indexes::stringology::SuffixAutomaton < SymbolType > ExactSuffixAutomaton::const
lastState = suffixLinks [ lastState ].first;
}
- return indexes::stringology::SuffixAutomaton < SymbolType > ( std::move ( suffixAutomaton ) );
+ return indexes::stringology::SuffixAutomaton < SymbolType > ( std::move ( suffixAutomaton ), pattern.getContent ( ).size ( ) );
}
template < class SymbolType >
diff --git a/alib2algo/src/stringology/query/BackwardDAWGMatching.h b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
index 9094f72e2b7c18bd80c30439d77e71d675056a87..b8588a29b3f8973c397a9b9b9dc21e7f9e487072 100644
--- a/alib2algo/src/stringology/query/BackwardDAWGMatching.h
+++ b/alib2algo/src/stringology/query/BackwardDAWGMatching.h
@@ -10,8 +10,6 @@
#include <indexes/stringology/SuffixAutomaton.h>
#include <string/LinearString.h>
-#include <stringology/properties/BackboneLength.h>
-
namespace stringology {
namespace query {
@@ -34,7 +32,7 @@ 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.getAutomaton ( ) );
+ size_t patternSize = suffixAutomaton.getBackboneLength ( );
size_t subjectSize = subject.getContent ( ).size ( );
bool fail;
diff --git a/alib2algo/src/stringology/query/SuffixAutomatonFactors.h b/alib2algo/src/stringology/query/SuffixAutomatonFactors.h
index 978a86b65fc994433223c16c8db0fab391fcd895..62a8f2a271d80da35a1c95ba752acf4ec9641870 100644
--- a/alib2algo/src/stringology/query/SuffixAutomatonFactors.h
+++ b/alib2algo/src/stringology/query/SuffixAutomatonFactors.h
@@ -38,13 +38,14 @@ public:
template < class SymbolType >
ext::set < unsigned > SuffixAutomatonFactors::query ( const indexes::stringology::SuffixAutomaton < SymbolType > & suffixAutomaton, const string::LinearString < SymbolType > & string ) {
+ unsigned backboneLength = suffixAutomaton.getBackboneLength ( );
+
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 < unsigned, unsigned > > open = { { std::get < 1 > ( run ), 0u } };
ext::vector < unsigned > tmp;
- unsigned max = 0;
while ( ! open.empty ( ) ) {
std::pair < unsigned, unsigned > cur = std::move ( open.back ( ) );
open.pop_back ( );
@@ -52,18 +53,13 @@ ext::set < unsigned > SuffixAutomatonFactors::query ( const indexes::stringology
if ( suffixAutomaton.getAutomaton ( ).getFinalStates ( ).count ( cur.first ) )
tmp.push_back ( cur.second );
- if ( cur.second > max )
- max = cur.second;
-
for ( const auto & transition : suffixAutomaton.getAutomaton ( ).getTransitionsFromState ( cur.first ) )
open.emplace_back ( transition.second, cur.second + 1 );
}
- unsigned subjectSize = max + string.getContent ( ).size ( );
-
ext::set < unsigned > res;
for ( unsigned dist : tmp )
- res.insert ( subjectSize - dist );
+ res.insert ( backboneLength - dist - 1 );
return res;
}
diff --git a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
index d5f5c279f072de4843fb0ef67a8d9312fddeb682..e18f732f133de5c6306a327690c99d39343b36c5 100644
--- a/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
+++ b/alib2algo/test-src/stringology/indexing/ExactSuffixAutomatonTest.cpp
@@ -40,7 +40,7 @@ void ExactSuffixAutomatonTest::testSuffixAutomatonConstruction ( ) {
tmp.addTransition ( 4, DefaultSymbolType ( 'a' ), 5 );
tmp.addTransition ( 5, DefaultSymbolType ( 'c' ), 6 );
- indexes::stringology::SuffixAutomaton < > refSuffixAutomaton ( std::move ( tmp ) );
+ indexes::stringology::SuffixAutomaton < > refSuffixAutomaton ( std::move ( tmp ), 6 );
std::cout << suffixAutomaton << std::endl;
std::cout << refSuffixAutomaton << std::endl;
diff --git a/alib2data/src/indexes/stringology/SuffixAutomaton.cpp b/alib2data/src/indexes/stringology/SuffixAutomaton.cpp
index 691a72c7997b49b5ad06038407481da42ac71b27..3a6c9094ef0fba2658417db64d625e23d5e2c456 100644
--- a/alib2data/src/indexes/stringology/SuffixAutomaton.cpp
+++ b/alib2data/src/indexes/stringology/SuffixAutomaton.cpp
@@ -7,11 +7,14 @@
#include "SuffixAutomaton.h"
+#include <registration/CastRegistration.hpp>
#include <registration/ValuePrinterRegistration.hpp>
#include <registration/XmlRegistration.hpp>
namespace {
+static auto DFAFromSuffixAutomaton = registration::CastRegister < automaton::DFA < DefaultSymbolType, unsigned >, indexes::stringology::SuffixAutomaton < > > ( );
+
static auto valuePrinter = registration::ValuePrinterRegister < indexes::stringology::SuffixAutomaton < > > ( );
static auto xmlWrite = registration::XmlWriterRegister < indexes::stringology::SuffixAutomaton < > > ( );
diff --git a/alib2data/src/indexes/stringology/SuffixAutomaton.h b/alib2data/src/indexes/stringology/SuffixAutomaton.h
index 40a9b597aab5f7c41f71df8f8c03fb4d65b16cbb..5a32a1369abe4b0e9502e72f9e620e769634eb5e 100644
--- a/alib2data/src/indexes/stringology/SuffixAutomaton.h
+++ b/alib2data/src/indexes/stringology/SuffixAutomaton.h
@@ -42,6 +42,8 @@ class SuffixAutomaton final : public object::ObjectBase {
protected:
automaton::DFA < SymbolType, unsigned > m_automaton;
+ unsigned m_backboneLength;
+
public:
/**
* @copydoc SuffixTrieNode::clone() const
@@ -53,7 +55,7 @@ public:
*/
virtual ObjectBase * plunder ( ) &&;
- explicit SuffixAutomaton ( automaton::DFA < SymbolType, unsigned > automaton );
+ explicit SuffixAutomaton ( automaton::DFA < SymbolType, unsigned > automaton, unsigned backboneLength );
/**
* @return Root node of the trie
@@ -78,6 +80,10 @@ public:
return m_automaton.removeInputSymbol ( symbol );
}
+ unsigned getBackboneLength ( ) const {
+ return m_backboneLength;
+ }
+
/**
* Prints XML representation of the tree to the output stream.
* @param out output stream to which print the tree
@@ -95,6 +101,8 @@ public:
virtual explicit operator std::string ( ) const;
+ explicit operator automaton::DFA < SymbolType, unsigned > ( ) const;
+
virtual object::ObjectBase * inc ( ) &&;
typedef SuffixAutomaton < > normalized_type;
@@ -109,7 +117,7 @@ namespace indexes {
namespace stringology {
template < class SymbolType >
-SuffixAutomaton < SymbolType >::SuffixAutomaton ( automaton::DFA < SymbolType, unsigned > automaton ) : m_automaton ( std::move ( automaton ) ) {
+SuffixAutomaton < SymbolType >::SuffixAutomaton ( automaton::DFA < SymbolType, unsigned > automaton, unsigned backboneLength ) : m_automaton ( std::move ( automaton ) ), m_backboneLength ( backboneLength ) {
}
template < class SymbolType >
@@ -154,6 +162,11 @@ SuffixAutomaton < SymbolType >::operator std::string ( ) const {
return ss.str ( );
}
+template < class SymbolType >
+SuffixAutomaton < SymbolType >::operator automaton::DFA < SymbolType, unsigned > ( ) const {
+ return getAutomaton ( );
+}
+
template < class SymbolType >
object::ObjectBase* SuffixAutomaton < SymbolType >::inc() && {
return new object::UniqueObject(object::Object(std::move(*this)), primitive::Integer(0));
@@ -168,6 +181,7 @@ 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 ( ) );
@@ -187,7 +201,8 @@ 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 ) );
+ unsigned backboneLength = core::xmlApi < unsigned >::parse ( input );
+ indexes::stringology::SuffixAutomaton < SymbolType > res ( std::move ( automaton ), backboneLength );
sax::FromXMLParserHelper::popToken ( input, sax::Token::TokenType::END_ELEMENT, xmlTagName ( ) );
return res;
@@ -209,6 +224,7 @@ 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 ( ) );
+ core::xmlApi < unsigned >::compose ( output, index.getBackboneLength ( ) );
output.emplace_back ( xmlTagName ( ), sax::Token::TokenType::END_ELEMENT );
}