From 15d89fdc328e3b96d3ec22008a34e1ecfbbf63de 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 13:19:13 +0200
Subject: [PATCH] libaregexptree: New api for derivations and integrals
---
aconversions/src/re2fa/Brzozowski.cpp | 4 +-
.../re2rg/re2rrg/BrzozowskiDerivationRRG.cpp | 5 +--
aderivation/src/aderivation.cpp | 10 +----
aintegral/src/aintegral.cpp | 10 +----
libaregexptree/src/RegExpDerivation.cpp | 29 ++++++++-------
libaregexptree/src/RegExpDerivation.h | 24 +++++++-----
libaregexptree/src/RegExpIntegral.cpp | 37 ++++++++++---------
libaregexptree/src/RegExpIntegral.h | 23 ++++++++----
8 files changed, 70 insertions(+), 72 deletions(-)
diff --git a/aconversions/src/re2fa/Brzozowski.cpp b/aconversions/src/re2fa/Brzozowski.cpp
index 3c2048ace6..90e268a227 100644
--- a/aconversions/src/re2fa/Brzozowski.cpp
+++ b/aconversions/src/re2fa/Brzozowski.cpp
@@ -51,8 +51,7 @@ FSM Brzozowski::convert( void )
for( const auto & a : alphabet )
{
- RegExpElement* dSymbol = new RegExpSymbol( a.getSymbol( ) );
- RegExp derived = deriv.derivation( list<RegExpElement*>( 1, dSymbol ) );
+ RegExp derived = deriv.derivation( a );
derived = opt.optimize( derived );
// no "trash state" (though algorithm probably considers it)
@@ -64,7 +63,6 @@ FSM Brzozowski::convert( void )
m_transitions.insert( Transition( regexp, Symbol( a.getSymbol( ) ), derived ) );
}
- delete dSymbol;
}
}
diff --git a/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp b/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp
index a0798d9c29..eff5b812bf 100644
--- a/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp
+++ b/aconversions/src/re2rg/re2rrg/BrzozowskiDerivationRRG.cpp
@@ -53,8 +53,7 @@ RightRegularGrammar BrzozowskiDerivationRRG::convert( void )
for( const auto & a : alphabet )
{
- RegExpElement* dSymbol = new RegExpSymbol( a.getSymbol( ) );
- RegExp derived = deriv.derivation( list<RegExpElement*>( 1, dSymbol ) );
+ RegExp derived = deriv.derivation( a );
derived = opt.optimize( derived );
if( ! derived.isEmpty( ) )
@@ -64,8 +63,6 @@ RightRegularGrammar BrzozowskiDerivationRRG::convert( void )
m_transitions.insert( Transition( regexp, Symbol( a.getSymbol( ) ), derived ) );
}
-
- delete dSymbol;
}
}
diff --git a/aderivation/src/aderivation.cpp b/aderivation/src/aderivation.cpp
index 018c8da654..aafb5aeee0 100644
--- a/aderivation/src/aderivation.cpp
+++ b/aderivation/src/aderivation.cpp
@@ -27,21 +27,15 @@ int main(int argc, char** argv)
SaxInterface::parseMemory(input, tokens);
RegExp re = RegExpParser::parse(tokens);
- list<RegExpElement*> dString;
+ list<Symbol> dString;
for( int i = 1; i < argc ; i++ )
{
string symbol( argv[ i ] );
// cout << "'" << symbol << "'" << endl;
- if( symbol == "" )
- dString.push_back( new RegExpEpsilon( ) );
- else
- dString.push_back( new RegExpSymbol( symbol ) );
+ dString.push_back( Symbol( symbol ) );
}
RegExpDerivation d( re );
d.derivation( dString ).toXML( cout );
-
- for( const auto & symbol : dString )
- delete symbol;
}
catch( AlibException & e )
{
diff --git a/aintegral/src/aintegral.cpp b/aintegral/src/aintegral.cpp
index c5599a0a94..cb6b62fd6c 100644
--- a/aintegral/src/aintegral.cpp
+++ b/aintegral/src/aintegral.cpp
@@ -27,21 +27,15 @@ int main(int argc, char** argv)
SaxInterface::parseMemory(input, tokens);
RegExp re = RegExpParser::parse(tokens);
- list<RegExpElement*> dString;
+ list<Symbol> dString;
for( int i = 1; i < argc ; i++ )
{
string symbol( argv[ i ] );
// cout << "'" << symbol << "'" << endl;
- if( symbol == "" )
- dString.push_back( new RegExpEpsilon( ) );
- else
- dString.push_back( new RegExpSymbol( symbol ) );
+ dString.push_back( Symbol( symbol ) );
}
RegExpIntegral i( re );
i.integral( dString ).toXML( cout );
-
- for( const auto & symbol : dString )
- delete symbol;
}
catch( AlibException & e )
{
diff --git a/libaregexptree/src/RegExpDerivation.cpp b/libaregexptree/src/RegExpDerivation.cpp
index f52f5cb62c..444e864cef 100644
--- a/libaregexptree/src/RegExpDerivation.cpp
+++ b/libaregexptree/src/RegExpDerivation.cpp
@@ -16,7 +16,7 @@ RegExpDerivation::RegExpDerivation( const RegExp & re ) : m_re( re )
}
-RegExp RegExpDerivation::derivation ( const list<RegExpElement*> & dString ) const
+RegExp RegExpDerivation::derivation ( const list<Symbol> & dString ) const
{
// TODO: change dString to WORD object when implemented.
@@ -24,12 +24,7 @@ RegExp RegExpDerivation::derivation ( const list<RegExpElement*> & dString ) con
for( const auto & dSymbol : dString )
{
- if( dynamic_cast<RegExpEpsilon*>( dSymbol ) )
- continue;
- else if( dynamic_cast<RegExpSymbol*>( dSymbol ) )
- derivedRegExp = derivation( oldRegExp, * dynamic_cast<RegExpSymbol*>( dSymbol ) );
- else
- throw AlibException( "RegExpDerivation::derivation - invalid/unknown RegExpElement passed in dString." );
+ derivedRegExp = derivation( oldRegExp, dSymbol );
delete oldRegExp;
oldRegExp = derivedRegExp;
@@ -40,7 +35,13 @@ RegExp RegExpDerivation::derivation ( const list<RegExpElement*> & dString ) con
return ret;
}
-RegExpElement * RegExpDerivation::derivation( const RegExpElement * node, const RegExpSymbol & dSymbol ) const
+RegExp RegExpDerivation::derivation ( const Symbol & symbol ) const
+{
+ RegExp ret = derivation( m_re.getRegExp( ), symbol );
+ return ret;
+}
+
+RegExpElement * RegExpDerivation::derivation( const RegExpElement * node, const Symbol & dSymbol ) const
{
const Alternation * alternation = dynamic_cast<const Alternation*>( node );
const Concatenation * concatenation = dynamic_cast<const Concatenation*>( node );
@@ -65,7 +66,7 @@ RegExpElement * RegExpDerivation::derivation( const RegExpElement * node, const
throw AlibException( "RegExpDerivation::derivation() - unknown RegExpElement node" );
}
-RegExpElement * RegExpDerivation::derivation( const Alternation * element, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpDerivation::derivation( const Alternation * element, const Symbol & dSymbol ) const
{
Alternation* ret = new Alternation( );
@@ -75,7 +76,7 @@ RegExpElement * RegExpDerivation::derivation( const Alternation * element, const
return ret;
}
-RegExpElement * RegExpDerivation::derivation( const Concatenation * element, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpDerivation::derivation( const Concatenation * element, const Symbol & dSymbol ) const
{
Alternation* alt = new Alternation( );
@@ -99,7 +100,7 @@ RegExpElement * RegExpDerivation::derivation( const Concatenation * element, con
return alt;
}
-RegExpElement * RegExpDerivation::derivation( const Iteration * element, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpDerivation::derivation( const Iteration * element, const Symbol & dSymbol ) const
{
Concatenation* ret = new Concatenation( );
@@ -109,7 +110,7 @@ RegExpElement * RegExpDerivation::derivation( const Iteration * element, const R
return ret;
}
-RegExpElement * RegExpDerivation::derivation( const RegExpSymbol * element, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpDerivation::derivation( const RegExpSymbol * element, const Symbol & dSymbol ) const
{
if( dSymbol == element->getSymbol( ) )
return new RegExpEpsilon( );
@@ -117,12 +118,12 @@ RegExpElement * RegExpDerivation::derivation( const RegExpSymbol * element, cons
return new RegExpEmpty( );
}
-RegExpElement * RegExpDerivation::derivation( const RegExpEpsilon * element, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpDerivation::derivation( const RegExpEpsilon * element, const Symbol & dSymbol ) const
{
return new RegExpEmpty( );
}
-RegExpElement * RegExpDerivation::derivation( const RegExpEmpty * element, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpDerivation::derivation( const RegExpEmpty * element, const Symbol & dSymbol ) const
{
return new RegExpEmpty( );
}
diff --git a/libaregexptree/src/RegExpDerivation.h b/libaregexptree/src/RegExpDerivation.h
index c9e0cf9504..a4fcafe252 100644
--- a/libaregexptree/src/RegExpDerivation.h
+++ b/libaregexptree/src/RegExpDerivation.h
@@ -34,19 +34,25 @@ public:
/**
* returns derivation of regexp over word dString
* @param dString list of RegExpElements representing word. Derivation will be over this word.
- * @see RegExpElement
* @return Derivation of regexp over dString
*/
- regexp::RegExp derivation( const std::list<regexp::RegExpElement*> & dString ) const;
+ regexp::RegExp derivation( const std::list<alphabet::Symbol> & dString ) const;
+
+ /**
+ * returns derivation of regexp over one-symbol word
+ * @param symbol one-symbol word
+ * @return Derivation of regexp over word
+ */
+ regexp::RegExp derivation( const alphabet::Symbol & symbol ) const;
private:
- regexp::RegExpElement * derivation( const regexp::RegExpElement * element, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * derivation( const regexp::Alternation * element, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * derivation( const regexp::Concatenation * element, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * derivation( const regexp::Iteration * element, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * derivation( const regexp::RegExpSymbol * element, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * derivation( const regexp::RegExpEpsilon * element, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * derivation( const regexp::RegExpEmpty * element, const regexp::RegExpSymbol & dSymbol ) const;
+ regexp::RegExpElement * derivation( const regexp::RegExpElement * element, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * derivation( const regexp::Alternation * element, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * derivation( const regexp::Concatenation * element, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * derivation( const regexp::Iteration * element, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * derivation( const regexp::RegExpSymbol * element, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * derivation( const regexp::RegExpEpsilon * element, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * derivation( const regexp::RegExpEmpty * element, const alphabet::Symbol & dSymbol ) const;
/**
* stores original regexp
diff --git a/libaregexptree/src/RegExpIntegral.cpp b/libaregexptree/src/RegExpIntegral.cpp
index ee5a569682..e8365e661c 100644
--- a/libaregexptree/src/RegExpIntegral.cpp
+++ b/libaregexptree/src/RegExpIntegral.cpp
@@ -15,7 +15,7 @@ RegExpIntegral::RegExpIntegral( const RegExp & re ) : m_re( re )
}
-RegExp RegExpIntegral::integral( const list<RegExpElement*> & dString ) const
+RegExp RegExpIntegral::integral( const list<Symbol> & dString ) const
{
// TODO: change dString to WORD object when implemented.
@@ -25,12 +25,7 @@ RegExp RegExpIntegral::integral( const list<RegExpElement*> & dString ) const
{
const auto & dSymbol = * it;
- if( dynamic_cast<RegExpEpsilon*>( dSymbol ) ) // integral V d\eps = V
- continue;
- else if( dynamic_cast<RegExpSymbol*>( dSymbol ) )
- integralRegExp = integral( oldRegExp, * dynamic_cast<RegExpSymbol*>( dSymbol ) );
- else
- throw AlibException( "RegExpIntegral::integral - invalid/unknown RegExpElement passed in dString." );
+ integralRegExp = integral( oldRegExp, dSymbol );
delete oldRegExp;
oldRegExp = integralRegExp;
@@ -41,7 +36,13 @@ RegExp RegExpIntegral::integral( const list<RegExpElement*> & dString ) const
return ret;
}
-RegExpElement * RegExpIntegral::integral( const RegExpElement * node, const RegExpSymbol & dSymbol ) const
+RegExp RegExpIntegral::integral ( const Symbol & symbol ) const
+{
+ RegExp ret = integral( m_re.getRegExp( ), symbol );
+ return ret;
+}
+
+RegExpElement * RegExpIntegral::integral( const RegExpElement * node, const Symbol & dSymbol ) const
{
const Alternation * alternation = dynamic_cast<const Alternation*>( node );
const Concatenation * concatenation = dynamic_cast<const Concatenation*>( node );
@@ -66,7 +67,7 @@ RegExpElement * RegExpIntegral::integral( const RegExpElement * node, const RegE
throw AlibException( "RegExpDerivation::derivation() - unknown RegExpElement node" );
}
-RegExpElement * RegExpIntegral::integral( const Alternation * node, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpIntegral::integral( const Alternation * node, const Symbol & dSymbol ) const
{
Alternation * alt = new Alternation( );
@@ -76,10 +77,10 @@ RegExpElement * RegExpIntegral::integral( const Alternation * node, const RegExp
return alt;
}
-RegExpElement * RegExpIntegral::integral( const Concatenation * node, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpIntegral::integral( const Concatenation * node, const Symbol & dSymbol ) const
{
Concatenation * concat = new Concatenation( );
- concat->getElements( ).push_back( dSymbol.clone( ) );
+ concat->getElements( ).push_back( new RegExpSymbol( dSymbol.getSymbol( ) ) );
for( const auto & child : node->getElements( ) )
concat->getElements( ).push_back( child->clone( ) );
@@ -87,28 +88,28 @@ RegExpElement * RegExpIntegral::integral( const Concatenation * node, const RegE
return concat;
}
-RegExpElement * RegExpIntegral::integral( const Iteration * node, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpIntegral::integral( const Iteration * node, const Symbol & dSymbol ) const
{
Concatenation * concat = new Concatenation( );
- concat->getElements( ).push_back( dSymbol.clone( ) );
+ concat->getElements( ).push_back( new RegExpSymbol( dSymbol.getSymbol( ) ) );
concat->getElements( ).push_back( node->getElement( )->clone( ) );
return concat;
}
-RegExpElement * RegExpIntegral::integral( const RegExpSymbol * node, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpIntegral::integral( const RegExpSymbol * node, const Symbol & dSymbol ) const
{
Concatenation * concat = new Concatenation( );
- concat->getElements( ).push_back( dSymbol.clone( ) );
+ concat->getElements( ).push_back( new RegExpSymbol( dSymbol.getSymbol( ) ) );
concat->getElements( ).push_back( node->clone( ) );
return concat;
}
-RegExpElement * RegExpIntegral::integral( const RegExpEpsilon * node, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpIntegral::integral( const RegExpEpsilon * node, const Symbol & dSymbol ) const
{
- return dSymbol.clone( );
+ return new RegExpSymbol( dSymbol.getSymbol( ) );
}
-RegExpElement * RegExpIntegral::integral( const RegExpEmpty * node, const RegExpSymbol & dSymbol ) const
+RegExpElement * RegExpIntegral::integral( const RegExpEmpty * node, const Symbol & dSymbol ) const
{
return new RegExpEmpty( );
}
diff --git a/libaregexptree/src/RegExpIntegral.h b/libaregexptree/src/RegExpIntegral.h
index 15b27a4c80..36c55a50ec 100644
--- a/libaregexptree/src/RegExpIntegral.h
+++ b/libaregexptree/src/RegExpIntegral.h
@@ -30,16 +30,23 @@ public:
* @see RegExpElement
* @return Integral of regexp over dString
*/
- regexp::RegExp integral( const std::list<regexp::RegExpElement*> & dString ) const;
+ regexp::RegExp integral( const std::list<alphabet::Symbol> & dString ) const;
+
+ /**
+ * returns integral of regexp over one-symbol word
+ * @param symbol one-symbol word
+ * @return integral of regexp over word
+ */
+ regexp::RegExp integral( const alphabet::Symbol & symbol ) const;
private:
- regexp::RegExpElement * integral( const regexp::RegExpElement * node, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * integral( const regexp::Alternation * node, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * integral( const regexp::Concatenation * node, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * integral( const regexp::Iteration * node, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * integral( const regexp::RegExpSymbol * node, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * integral( const regexp::RegExpEpsilon * node, const regexp::RegExpSymbol & dSymbol ) const;
- regexp::RegExpElement * integral( const regexp::RegExpEmpty * node, const regexp::RegExpSymbol & dSymbol ) const;
+ regexp::RegExpElement * integral( const regexp::RegExpElement * node, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * integral( const regexp::Alternation * node, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * integral( const regexp::Concatenation * node, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * integral( const regexp::Iteration * node, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * integral( const regexp::RegExpSymbol * node, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * integral( const regexp::RegExpEpsilon * node, const alphabet::Symbol & dSymbol ) const;
+ regexp::RegExpElement * integral( const regexp::RegExpEmpty * node, const alphabet::Symbol & dSymbol ) const;
/**
* stores original regular expression
--
GitLab