From b07e8e3ac7aaae062ce25ff7d88e2df159140381 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1=C5=A1=20Pecka?= <peckato1@fit.cvut.cz>
Date: Mon, 14 Apr 2014 13:31:16 +0200
Subject: [PATCH] aconversions: Brzozowski Derivation Method - follow math
precisely
---
aconversions/src/re2fa/Brzozowski.cpp | 45 +++++++----------
aconversions/src/re2fa/Brzozowski.h | 24 ---------
.../re2rg/re2rrg/BrzozowskiDerivationRRG.cpp | 50 +++++++------------
.../re2rg/re2rrg/BrzozowskiDerivationRRG.h | 11 ----
4 files changed, 35 insertions(+), 95 deletions(-)
diff --git a/aconversions/src/re2fa/Brzozowski.cpp b/aconversions/src/re2fa/Brzozowski.cpp
index 90e268a227..24415f7739 100644
--- a/aconversions/src/re2fa/Brzozowski.cpp
+++ b/aconversions/src/re2fa/Brzozowski.cpp
@@ -54,14 +54,9 @@ FSM Brzozowski::convert( void )
RegExp derived = deriv.derivation( a );
derived = opt.optimize( derived );
- // no "trash state" (though algorithm probably considers it)
- if( ! derived.isEmpty( ) )
- {
- if( ! isInSet( derived, Q ) ) // if this state has already been found, do not add
- Qi.at( i ).insert( derived );
-
- m_transitions.insert( Transition( regexp, Symbol( a.getSymbol( ) ), derived ) );
- }
+ // this will also add \emptyset as a regexp (and as FA state)
+ if( ! isInSet( derived, Q ) ) // if this state has already been found, do not add
+ Qi.at( i ).insert( derived );
}
}
@@ -82,8 +77,20 @@ FSM Brzozowski::convert( void )
for( const auto & a : alphabet )
m_fsm.addInputSymbol( a.getSymbol( ) );
- for( const auto & t : m_transitions )
- m_fsm.addTransition( builder.getState( t.m_from ), t.m_symbol, builder.getState( t.m_to ) );
+ for( const auto & r : Q )
+ {
+ RegExpDerivation deriv( r );
+
+ for( const auto & a: m_fsm.getInputAlphabet( ) )
+ {
+ RegExp derived = deriv.derivation( a );
+ derived = opt.optimize( derived );
+
+ TransitionFSM t( builder.getState( r ), a, builder.getState( derived ) );
+ if( ! isInSet( t, m_fsm.getTransitions( ) ) )
+ m_fsm.addTransition( t );
+ }
+ }
m_fsm.addInitialState( builder.getState( V ) );
@@ -96,24 +103,6 @@ FSM Brzozowski::convert( void )
// ----------------------------------------------------------------------------
-Brzozowski::Transition::Transition( const RegExp & from, const Symbol & symb, const RegExp & to )
- : m_from( from ), m_to( to ), m_symbol( symb )
-{
-
-}
-
-bool Brzozowski::Transition::operator<( const Transition & x ) const
-{
- if( m_from != x.m_from )
- return m_from < x.m_from;
- else if( m_symbol != x.m_symbol )
- return m_symbol < x.m_symbol;
- else
- return m_to < x.m_to;
-}
-
-// ----------------------------------------------------------------------------
-
Brzozowski::StateBuilder::StateBuilder( const set<RegExp> & Q )
{
m_stateId = 0;
diff --git a/aconversions/src/re2fa/Brzozowski.h b/aconversions/src/re2fa/Brzozowski.h
index d455ff5308..c858dfb5c7 100644
--- a/aconversions/src/re2fa/Brzozowski.h
+++ b/aconversions/src/re2fa/Brzozowski.h
@@ -77,30 +77,6 @@ private:
unsigned int m_stateId;
};
-
- /**
- * Stores original regular epxression (m_from), and result of derivation (m_to) over m_symbol.
- */
- struct Transition
- {
- const regexp::RegExp m_from, m_to;
- const alphabet::Symbol m_symbol;
-
- /**
- * @param from original regexp
- * @param to resulting regexp
- * @param symb
- */
- Transition( const regexp::RegExp & from, const alphabet::Symbol & symb, const regexp::RegExp & to );
-
- bool operator<( const Transition & other ) const;
- };
-
- /**
- * Set of transitions.
- * @see Brzozowski::Transition
- */
- std::set<Transition> m_transitions;
};
} /* namespace conversions */
diff --git a/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp b/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp
index eff5b812bf..0797f640ce 100644
--- a/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp
+++ b/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp
@@ -56,13 +56,8 @@ RightRegularGrammar BrzozowskiDerivationRRG::convert( void )
RegExp derived = deriv.derivation( a );
derived = opt.optimize( derived );
- if( ! derived.isEmpty( ) )
- {
- if( ! isInSet( derived, N ) ) // if this state has already been found, do not add
- Ni.at( i ).insert( derived );
-
- m_transitions.insert( Transition( regexp, Symbol( a.getSymbol( ) ), derived ) );
- }
+ if( ! isInSet( derived, N ) ) // if this state has already been found, do not add
+ Ni.at( i ).insert( derived );
}
}
@@ -77,18 +72,27 @@ RightRegularGrammar BrzozowskiDerivationRRG::convert( void )
NonTerminalBuilder builder( N, m_grammar );
- for( const auto & t : m_transitions )
+ for( const auto & r : N )
{
- list<Symbol> leftSide = { builder.getNonTerminal( t.m_from ) };
- list<Symbol> rightSide = { t.m_symbol, builder.getNonTerminal( t.m_to ) };
- Rule r( leftSide, rightSide );
- m_grammar.addRule( r );
+ RegExpDerivation deriv( r );
- if( t.m_to.containsEmptyString( ) )
+ for( const auto & a : alphabet )
{
- list<Symbol> rightSide = { t.m_symbol };
+ RegExp derived = deriv.derivation( a );
+ derived = opt.optimize( derived );
+
+ list<Symbol> leftSide = { builder.getNonTerminal( r ) };
+ list<Symbol> rightSide = { a, builder.getNonTerminal( derived ) };
+
Rule r( leftSide, rightSide );
m_grammar.addRule( r );
+
+ if( derived.containsEmptyString( ) )
+ {
+ list<Symbol> rightSide = { a };
+ Rule r( leftSide, rightSide );
+ m_grammar.addRule( r );
+ }
}
}
@@ -122,24 +126,6 @@ RightRegularGrammar BrzozowskiDerivationRRG::convert( void )
// ----------------------------------------------------------------------------
-BrzozowskiDerivationRRG::Transition::Transition( const RegExp & from, const Symbol & symb, const RegExp & to )
- : m_from( from ), m_to( to ), m_symbol( symb )
-{
-
-}
-
-bool BrzozowskiDerivationRRG::Transition::operator<( const Transition & x ) const
-{
- if( m_from != x.m_from )
- return m_from < x.m_from;
- else if( m_symbol != x.m_symbol )
- return m_symbol < x.m_symbol;
- else
- return m_to < x.m_to;
-}
-
-// ----------------------------------------------------------------------------
-
BrzozowskiDerivationRRG::NonTerminalBuilder::NonTerminalBuilder( const set<RegExp> & Q, RightRegularGrammar & grammar ) : m_grammar( grammar )
{
m_nonTerminalId = 0;
diff --git a/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.h b/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.h
index 47b1546dcc..90182cf0cc 100644
--- a/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.h
+++ b/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.h
@@ -45,15 +45,6 @@ public:
grammar::RightRegularGrammar convert( void );
protected:
- struct Transition
- {
- const regexp::RegExp m_from, m_to;
- const alphabet::Symbol m_symbol;
-
- Transition( const regexp::RegExp & from, const alphabet::Symbol & symb, const regexp::RegExp & to );
- bool operator<( const Transition & other ) const;
- };
-
class NonTerminalBuilder
{
public:
@@ -66,8 +57,6 @@ protected:
std::string createNewName( void );
unsigned int m_nonTerminalId;
};
-
- std::set<Transition> m_transitions;
};
} /* namespace conversions */
--
GitLab