From dc33ec0ccebeedea3efa4db4bf4c957ee45abf49 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1=C5=A1=20Pecka?= <peckato1@fit.cvut.cz>
Date: Thu, 13 Mar 2014 15:34:35 +0100
Subject: [PATCH] Glushkov RE->FA cleanup
---
aconversions/src/re2fa/Glushkov.cpp | 270 +---------------------------
aconversions/src/re2fa/Glushkov.h | 28 +--
2 files changed, 5 insertions(+), 293 deletions(-)
diff --git a/aconversions/src/re2fa/Glushkov.cpp b/aconversions/src/re2fa/Glushkov.cpp
index b61ecf12ac..ccf00c4b0d 100644
--- a/aconversions/src/re2fa/Glushkov.cpp
+++ b/aconversions/src/re2fa/Glushkov.cpp
@@ -79,271 +79,24 @@ void Glushkov::initNumberSymbols( void )
void Glushkov::constructBeginSymbolSet( void )
{
- for( const auto & s : getLeftmostSymbolsInTree( m_re.getRegExp( ) ) )
+ for( const auto & s : GlushkovTraversal::getLeftmostSymbolsInTree( m_re ) )
m_beginSymbolSet.insert( m_numberedSymbols.find( s )->second );
}
void Glushkov::constructEndSymbolSet( void )
{
- for( const auto & s : getRightmostSymbolsInTree( m_re.getRegExp( ) ) )
+ for( const auto & s : GlushkovTraversal::getRightmostSymbolsInTree( m_re ) )
m_endSymbolSet.insert( m_numberedSymbols.find( s )->second );
}
void Glushkov::constructNeighbourSymbolSet( void )
{
- for( const auto & n : getNeighbours( m_re.getRegExp( ) ) )
+ for( const auto & n : GlushkovTraversal::getNeighbours( m_re ) )
m_neighbourSymbolSet.insert( n );
}
// ----------------------------------------------------------------------------
-set<const RegExpSymbol*> Glushkov::getLeftmostSymbolsInTree( const RegExpElement * node ) const
-{
- const Alternation* alternation = dynamic_cast<const Alternation*>( node );
- const Concatenation* concatenation = dynamic_cast<const Concatenation*>( node );
- const Iteration* iteration = dynamic_cast<const Iteration*>( node );
- const RegExpSymbol* symbol = dynamic_cast<const RegExpSymbol*>( node );
- const RegExpEmpty* empty = dynamic_cast<const RegExpEmpty*>( node );
- const RegExpEpsilon* eps = dynamic_cast<const RegExpEpsilon*>( node );
-
- if( symbol )
- return getLeftmostSymbolsInTree( symbol );
- else if( alternation )
- return getLeftmostSymbolsInTree( alternation );
- else if( concatenation )
- return getLeftmostSymbolsInTree( concatenation );
- else if( iteration )
- return getLeftmostSymbolsInTree( iteration );
- else if( eps )
- return getLeftmostSymbolsInTree( eps );
- else if( empty )
- return getLeftmostSymbolsInTree( empty );
-
- throw AlibException( "Glushkov::getLeftmostSymbolsInTree - invalid RegExpElement node" );
-}
-
-set<const RegExpSymbol*> Glushkov::getLeftmostSymbolsInTree( const Alternation * node ) const
-{
- set<const RegExpSymbol*> ret;
-
- for( const auto & e : node->getElements( ) )
- {
- const set<const RegExpSymbol*> tmp = getLeftmostSymbolsInTree( e );
- ret.insert( tmp.begin( ), tmp.end( ) );
- }
-
- return ret;
-}
-
-set<const RegExpSymbol*> Glushkov::getLeftmostSymbolsInTree( const Concatenation * node ) const
-{
- set<const RegExpSymbol*> ret;
-
- for( const auto & e : node->getElements( ) )
- {
- set<const RegExpSymbol*> tmp = getLeftmostSymbolsInTree( e );
- ret.insert( tmp.begin( ), tmp.end( ) );
-
- if( ! e->containsEmptyString( ) ) // If this subtree can be epsilon, then we need to add next subtree also
- break;
- }
-
- return ret;
-}
-
-set<const RegExpSymbol*> Glushkov::getLeftmostSymbolsInTree( const Iteration * node ) const
-{
- return getLeftmostSymbolsInTree( node->getElement( ) );
-}
-
-set<const RegExpSymbol*> Glushkov::getLeftmostSymbolsInTree( const RegExpSymbol * node ) const
-{
- return set<const RegExpSymbol*> { node };
-}
-
-set<const RegExpSymbol*> Glushkov::getLeftmostSymbolsInTree( const RegExpEpsilon * node ) const
-{
- return set<const RegExpSymbol*>( );
-}
-
-set<const RegExpSymbol*> Glushkov::getLeftmostSymbolsInTree( const RegExpEmpty * node ) const
-{
- return set<const RegExpSymbol*>( );
-}
-// ----------------------------------------------------------------------------
-
-set<const RegExpSymbol*> Glushkov::getRightmostSymbolsInTree( const RegExpElement * node ) const
-{
- const Alternation* alternation = dynamic_cast<const Alternation*>( node );
- const Concatenation* concatenation = dynamic_cast<const Concatenation*>( node );
- const Iteration* iteration = dynamic_cast<const Iteration*>( node );
- const RegExpSymbol* symbol = dynamic_cast<const RegExpSymbol*>( node );
- const RegExpEmpty* empty = dynamic_cast<const RegExpEmpty*>( node );
- const RegExpEpsilon* eps = dynamic_cast<const RegExpEpsilon*>( node );
-
- if( symbol )
- return getRightmostSymbolsInTree( symbol );
- else if( alternation )
- return getRightmostSymbolsInTree( alternation );
- else if( concatenation )
- return getRightmostSymbolsInTree( concatenation );
- else if( iteration )
- return getRightmostSymbolsInTree( iteration );
- else if( eps )
- return getRightmostSymbolsInTree( eps );
- else if( empty )
- return getRightmostSymbolsInTree( empty );
-
- throw AlibException( "Glushkov::getRightmostSymbolsInTree - invalid RegExpElement node" );
-}
-
-set<const RegExpSymbol*> Glushkov::getRightmostSymbolsInTree( const Alternation * node ) const
-{
- set<const RegExpSymbol*> ret;
-
- for( const auto & e : node->getElements( ) )
- {
- set<const RegExpSymbol*> tmp = getRightmostSymbolsInTree( e );
- ret.insert( tmp.begin( ), tmp.end( ) );
- }
-
- return ret;
-}
-
-set<const RegExpSymbol*> Glushkov::getRightmostSymbolsInTree( const Concatenation * node ) const
-{
- set<const RegExpSymbol*> ret;
-
- for( auto it = node->getElements( ).rbegin( ); it != node->getElements( ).rend( ) ; it ++ )
- {
- set<const RegExpSymbol*> tmp = getRightmostSymbolsInTree( *it );
- ret.insert( tmp.begin( ), tmp.end( ) );
-
- if( ! ( * it )->containsEmptyString( ) )
- break;
- }
-
- return ret;
-}
-
-set<const RegExpSymbol*> Glushkov::getRightmostSymbolsInTree( const Iteration * node ) const
-{
- return getRightmostSymbolsInTree( node->getElement( ) );
-}
-
-set<const RegExpSymbol*> Glushkov::getRightmostSymbolsInTree( const RegExpSymbol * node ) const
-{
- return set<const RegExpSymbol*> { node };
-}
-
-set<const RegExpSymbol*> Glushkov::getRightmostSymbolsInTree( const RegExpEpsilon * node ) const
-{
- return set<const RegExpSymbol*>( );
-}
-
-set<const RegExpSymbol*> Glushkov::getRightmostSymbolsInTree( const RegExpEmpty * node ) const
-{
- return set<const RegExpSymbol*>( );
-}
-
-// ----------------------------------------------------------------------------
-
-set<Glushkov::Neighbours> Glushkov::getNeighbours( const RegExpElement * node ) const
-{
- const Alternation* alternation = dynamic_cast<const Alternation*>( node );
- const Concatenation* concatenation = dynamic_cast<const Concatenation*>( node );
- const Iteration* iteration = dynamic_cast<const Iteration*>( node );
- const RegExpSymbol* symbol = dynamic_cast<const RegExpSymbol*>( node );
- const RegExpEmpty* empty = dynamic_cast<const RegExpEmpty*>( node );
- const RegExpEpsilon* eps = dynamic_cast<const RegExpEpsilon*>( node );
-
- if( symbol )
- return getNeighbours( symbol );
- else if( alternation )
- return getNeighbours( alternation );
- else if( concatenation )
- return getNeighbours( concatenation );
- else if( iteration )
- return getNeighbours( iteration );
- else if( eps )
- return getNeighbours( eps );
- else if( empty )
- return getNeighbours( empty );
-
- throw AlibException( "Glushkov::getNeighbours - unknown RegExpElement* " );
-}
-
-set<Glushkov::Neighbours> Glushkov::getNeighbours( const Alternation * node ) const
-{
- set<Neighbours> n;
- for( const auto & e : node->getElements( ) )
- {
- set<Neighbours> tmp = getNeighbours( e );
- n.insert( tmp.begin( ), tmp.end( ) );
- }
-
- return n;
-}
-
-set<Glushkov::Neighbours> Glushkov::getNeighbours( const Concatenation * node ) const
-{
- set<Neighbours> n;
- for( const auto & e : node->getElements( ) )
- {
- set<Neighbours> tmp = getNeighbours( e );
- n.insert( tmp.begin( ), tmp.end( ) );
- }
-
- for( auto e = node->getElements( ).begin( ); e != node->getElements( ).end( ); e ++ )
- {
- auto f = e;
- if( f == node->getElements( ).end( ) )
- continue;
-
- for( f++ ; f != node->getElements( ).end( ); f ++ )
- {
- for( const auto & x : getRightmostSymbolsInTree( * e ) )
- for( const auto & y : getLeftmostSymbolsInTree( * f ) )
- n.insert( Neighbours( x, y ) );
-
- if( ! ( * f )->containsEmptyString( ) )
- break;
- }
- }
-
- return n;
-}
-
-set<Glushkov::Neighbours> Glushkov::getNeighbours( const Iteration * node ) const
-{
- set<Neighbours> n;
- set<Neighbours> tmp = getNeighbours( node->getElement( ) );
- n.insert( tmp.begin( ), tmp.end( ) );
-
- for( const auto & x : getRightmostSymbolsInTree( node->getElement( ) ) )
- for( const auto & y : getLeftmostSymbolsInTree( node->getElement( ) ) )
- n.insert( Neighbours( x, y ) );
-
- return n;
-}
-
-set<Glushkov::Neighbours> Glushkov::getNeighbours( const RegExpSymbol * node ) const
-{
- return set<Neighbours>( );
-}
-
-set<Glushkov::Neighbours> Glushkov::getNeighbours( const RegExpEpsilon * node ) const
-{
- return set<Neighbours>( );
-}
-
-set<Glushkov::Neighbours> Glushkov::getNeighbours( const RegExpEmpty * node ) const
-{
- return set<Neighbours>( );
-}
-
-// ----------------------------------------------------------------------------
-
bool Glushkov::NumberedSymbol::operator<( const NumberedSymbol & x ) const
{
return m_i < x.m_i;
@@ -357,21 +110,4 @@ Glushkov::NumberedSymbol::NumberedSymbol( const RegExpSymbol * symbol, int i ) :
}
-// ----------------------------------------------------------------------------
-
-Glushkov::Neighbours::Neighbours( const RegExpSymbol * first, const RegExpSymbol * second ) :
- m_first( first ),
- m_second( second )
-{
-
-}
-
-bool Glushkov::Neighbours::operator<( const Neighbours & x ) const
-{
- if( m_first != x.m_first )
- return m_first < x.m_first;
- else
- return m_second < x.m_second;
-}
-
} /* namespace conversions */
diff --git a/aconversions/src/re2fa/Glushkov.h b/aconversions/src/re2fa/Glushkov.h
index 452028c1f4..5a83b87c1a 100644
--- a/aconversions/src/re2fa/Glushkov.h
+++ b/aconversions/src/re2fa/Glushkov.h
@@ -15,6 +15,7 @@
#include <regexp/RegExp.h>
#include "AbstractREtoFAConverter.h"
+#include "../shared/GlushkovTraversal.h"
#include "RegExpAlphabet.h"
@@ -58,34 +59,9 @@ private:
void constructEndSymbolSet( void );
void constructNeighbourSymbolSet( void );
- // TODO: consider moving these to independent class or regexputils
- std::set<const regexp::RegExpSymbol*> getLeftmostSymbolsInTree( const regexp::RegExpElement * node ) const;
- std::set<const regexp::RegExpSymbol*> getLeftmostSymbolsInTree( const regexp::Alternation * node ) const;
- std::set<const regexp::RegExpSymbol*> getLeftmostSymbolsInTree( const regexp::Concatenation * node ) const;
- std::set<const regexp::RegExpSymbol*> getLeftmostSymbolsInTree( const regexp::Iteration * node ) const;
- std::set<const regexp::RegExpSymbol*> getLeftmostSymbolsInTree( const regexp::RegExpSymbol * node ) const;
- std::set<const regexp::RegExpSymbol*> getLeftmostSymbolsInTree( const regexp::RegExpEmpty * node ) const;
- std::set<const regexp::RegExpSymbol*> getLeftmostSymbolsInTree( const regexp::RegExpEpsilon * node ) const;
-
- std::set<const regexp::RegExpSymbol*> getRightmostSymbolsInTree( const regexp::RegExpElement * node ) const;
- std::set<const regexp::RegExpSymbol*> getRightmostSymbolsInTree( const regexp::Alternation * node ) const;
- std::set<const regexp::RegExpSymbol*> getRightmostSymbolsInTree( const regexp::Concatenation * node ) const;
- std::set<const regexp::RegExpSymbol*> getRightmostSymbolsInTree( const regexp::Iteration * node ) const;
- std::set<const regexp::RegExpSymbol*> getRightmostSymbolsInTree( const regexp::RegExpSymbol * node ) const;
- std::set<const regexp::RegExpSymbol*> getRightmostSymbolsInTree( const regexp::RegExpEmpty * node ) const;
- std::set<const regexp::RegExpSymbol*> getRightmostSymbolsInTree( const regexp::RegExpEpsilon * node ) const;
-
- std::set<Neighbours> getNeighbours( const regexp::RegExpElement * node ) const;
- std::set<Neighbours> getNeighbours( const regexp::Alternation * node ) const;
- std::set<Neighbours> getNeighbours( const regexp::Concatenation * node ) const;
- std::set<Neighbours> getNeighbours( const regexp::Iteration * node ) const;
- std::set<Neighbours> getNeighbours( const regexp::RegExpSymbol * node ) const;
- std::set<Neighbours> getNeighbours( const regexp::RegExpEmpty * node ) const;
- std::set<Neighbours> getNeighbours( const regexp::RegExpEpsilon * node ) const;
-
std::map<const regexp::RegExpElement*, NumberedSymbol> m_numberedSymbols;
std::set<NumberedSymbol> m_beginSymbolSet, m_endSymbolSet;
- std::set<Neighbours> m_neighbourSymbolSet;
+ std::set<GlushkovTraversal::Neighbours> m_neighbourSymbolSet;
};
--
GitLab