Skip to content
Snippets Groups Projects
Commit f223545d authored by Jan Trávníček's avatar Jan Trávníček
Browse files

template RightRG in regexp to grammar algo

parent bb4925db
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
alib2algo/src/regexp/convert/ToGrammar.cpp
+ 2
− 2
View file @ f223545d
......@@ -12,8 +12,8 @@ namespace regexp {
 
namespace convert {
 
auto ToGrammarFormalRegExp = registration::AbstractRegister < ToGrammar, grammar::RightRG < >, const regexp::FormalRegExp < > & > ( ToGrammar::convert );
auto ToGrammarUnboundedRegExp = registration::AbstractRegister < ToGrammar, grammar::RightRG < >, const regexp::UnboundedRegExp < > & > ( ToGrammar::convert );
auto ToGrammarFormalRegExp = registration::AbstractRegister < ToGrammar, grammar::RightRG < DefaultSymbolType, ext::pair < DefaultSymbolType, unsigned > >, const regexp::FormalRegExp < > & > ( ToGrammar::convert );
auto ToGrammarUnboundedRegExp = registration::AbstractRegister < ToGrammar, grammar::RightRG < DefaultSymbolType, ext::pair < DefaultSymbolType, unsigned > >, const regexp::UnboundedRegExp < > & > ( ToGrammar::convert );
 
} /* namespace convert */
 
......
alib2algo/src/regexp/convert/ToGrammar.h
+ 16
− 7
View file @ f223545d
......@@ -18,23 +18,32 @@ class ToGrammar {
public:
/**
* Performs conversion.
* @return right regular grammar equivalent to source regexp.
*
* \tparam SymbolType type of symbols in the regexp
*
* \param regexp the converted regexp
*
* \return right regular grammar equivalent to source regexp.
*/
template < class SymbolType >
static grammar::RightRG < > convert(const regexp::FormalRegExp < SymbolType > & regexp);
static grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > convert(const regexp::FormalRegExp < SymbolType > & regexp);
/**
* \overload
*/
template < class SymbolType >
static grammar::RightRG < > convert(const regexp::UnboundedRegExp < SymbolType > & regexp);
static grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > convert(const regexp::UnboundedRegExp < SymbolType > & regexp);
 
};
 
template < class SymbolType >
grammar::RightRG < > ToGrammar::convert(const regexp::FormalRegExp < SymbolType > & regexp) {
return grammar::RightRG < >(ToGrammarRightRGGlushkov::convert(regexp));
grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > ToGrammar::convert ( const regexp::FormalRegExp < SymbolType > & regexp ) {
return ToGrammarRightRGGlushkov::convert ( regexp );
}
 
template < class SymbolType >
grammar::RightRG < > ToGrammar::convert(const regexp::UnboundedRegExp < SymbolType > & regexp) {
return grammar::RightRG < >(ToGrammarRightRGGlushkov::convert(regexp));
grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > ToGrammar::convert ( const regexp::UnboundedRegExp < SymbolType > & regexp ) {
return ToGrammarRightRGGlushkov::convert ( regexp);
}
 
} /* namespace convert */
......
alib2algo/src/regexp/convert/ToGrammarRightRGGlushkov.cpp
+ 2
− 76
View file @ f223545d
......@@ -6,88 +6,14 @@
*/
 
#include "ToGrammarRightRGGlushkov.h"
#include <alib/algorithm>
#include <alphabet/InitialSymbol.h>
#include "../glushkov/GlushkovFollow.h"
#include "../glushkov/GlushkovIndexate.h"
#include "../glushkov/GlushkovFirst.h"
#include "../glushkov/GlushkovLast.h"
#include <regexp/properties/RegExpEpsilon.h>
#include <exception/CommonException.h>
#include <registration/AlgoRegistration.hpp>
 
#include <container/ObjectsPair.h>
namespace regexp {
 
namespace convert {
 
grammar::RightRG < > ToGrammarRightRGGlushkov::convert ( const regexp::UnboundedRegExp < > & regexp ) {
DefaultSymbolType S = alphabet::InitialSymbol::instance < DefaultSymbolType > ( );
grammar::RightRG < > grammar ( S );
// step 1
grammar.setTerminalAlphabet ( regexp.getAlphabet ( ) );
regexp::UnboundedRegExp < ext::pair < DefaultSymbolType, unsigned > > indexedRegExp = regexp::GlushkovIndexate::index ( regexp );
// steps 2, 3, 4
const ext::set < regexp::UnboundedRegExpSymbol < ext::pair < DefaultSymbolType, unsigned > > > first = regexp::GlushkovFirst::first ( indexedRegExp );
const ext::set < regexp::UnboundedRegExpSymbol < ext::pair < DefaultSymbolType, unsigned > > > last = regexp::GlushkovLast::last ( indexedRegExp );
// \e in q0 check is in step 7
// step 5
for ( const ext::pair < DefaultSymbolType, unsigned > & symbol : indexedRegExp.getAlphabet ( ) )
grammar.addNonterminalSymbol ( DefaultSymbolType ( container::ObjectsPair < DefaultSymbolType, unsigned > ( symbol ) ) );
// step 6
for ( const regexp::UnboundedRegExpSymbol < ext::pair < DefaultSymbolType, unsigned > > & symbol : first )
grammar.addRule ( S, ext::make_pair ( symbol.getSymbol ( ).first, DefaultSymbolType ( container::ObjectsPair < DefaultSymbolType, unsigned > ( symbol.getSymbol ( ) ) ) ) );
for ( const ext::pair < DefaultSymbolType, unsigned > & x : indexedRegExp.getAlphabet ( ) )
for ( const regexp::UnboundedRegExpSymbol < ext::pair < DefaultSymbolType, unsigned > > & f : regexp::GlushkovFollow::follow ( indexedRegExp, UnboundedRegExpSymbol < ext::pair < DefaultSymbolType, unsigned > > ( x ) ) ) {
const ext::pair < DefaultSymbolType, unsigned > & a = x;
const ext::pair < DefaultSymbolType, unsigned > & b = f.getSymbol ( );
grammar.addRule ( DefaultSymbolType ( container::ObjectsPair < DefaultSymbolType, unsigned > ( a ) ), ext::make_pair ( b.first, DefaultSymbolType ( container::ObjectsPair < DefaultSymbolType, unsigned > ( b ) ) ) );
}
// step 7
/*
* for all rules where ns.m_nonTerminal is on rightSide:
* add Rule: leftSide -> symbol.getSymbol( )
* unless it already exists
*/
for ( const auto & rule : grammar.getRules ( ) )
for ( const auto & rhs : rule.second ) {
if ( ! rhs.is < ext::pair < DefaultSymbolType, DefaultSymbolType > > ( ) )
continue;
const ext::pair < DefaultSymbolType, DefaultSymbolType > & rawRhs = rhs.get < ext::pair < DefaultSymbolType, DefaultSymbolType > > ( );
for ( const regexp::UnboundedRegExpSymbol < ext::pair < DefaultSymbolType, unsigned > > & symbol : last )
if ( rawRhs.second == DefaultSymbolType ( container::ObjectsPair < DefaultSymbolType, unsigned > ( symbol.getSymbol ( ) ) ) )
grammar.addRule ( rule.first, rawRhs.first );
}
if ( regexp::properties::RegExpEpsilon::languageContainsEpsilon ( regexp ) )
grammar.setGeneratesEpsilon ( true );
return grammar;
}
grammar::RightRG < > ToGrammarRightRGGlushkov::convert ( const regexp::FormalRegExp < > & regexp ) {
return ToGrammarRightRGGlushkov::convert ( regexp::UnboundedRegExp < > ( regexp ) );
}
auto ToGrammarRightRGGlushkovUnboundedRegExp = registration::AbstractRegister < ToGrammarRightRGGlushkov, grammar::RightRG < >, const regexp::UnboundedRegExp < > & > ( ToGrammarRightRGGlushkov::convert );
auto ToGrammarRightRGGlushkovFormalRegExp = registration::AbstractRegister < ToGrammarRightRGGlushkov, grammar::RightRG < >, const regexp::FormalRegExp < > & > ( ToGrammarRightRGGlushkov::convert );
auto ToGrammarRightRGGlushkovUnboundedRegExp = registration::AbstractRegister < ToGrammarRightRGGlushkov, grammar::RightRG < DefaultSymbolType, ext::pair < DefaultSymbolType, unsigned > >, const regexp::UnboundedRegExp < > & > ( ToGrammarRightRGGlushkov::convert );
auto ToGrammarRightRGGlushkovFormalRegExp = registration::AbstractRegister < ToGrammarRightRGGlushkov, grammar::RightRG < DefaultSymbolType, ext::pair < DefaultSymbolType, unsigned > >, const regexp::FormalRegExp < > & > ( ToGrammarRightRGGlushkov::convert );
 
} /* namespace convert */
 
......
alib2algo/src/regexp/convert/ToGrammarRightRGGlushkov.h
+ 87
− 4
View file @ f223545d
......@@ -12,6 +12,17 @@
#include <regexp/formal/FormalRegExp.h>
#include <regexp/unbounded/UnboundedRegExp.h>
 
#include <alib/algorithm>
#include <alphabet/InitialSymbol.h>
#include "../glushkov/GlushkovFollow.h"
#include "../glushkov/GlushkovIndexate.h"
#include "../glushkov/GlushkovFirst.h"
#include "../glushkov/GlushkovLast.h"
#include <regexp/properties/RegExpEpsilon.h>
namespace regexp {
 
namespace convert {
......@@ -24,13 +35,85 @@ class ToGrammarRightRGGlushkov {
public:
/**
* Performs conversion.
* @param regexp original regular expression
* @return right regular grammar equivalent to source regexp.
*
* \tparam SymbolType tye of symbols in the regular expression
*
* \param regexp original regular expression
*
* \return right regular grammar equivalent to source regexp.
*/
template < class SymbolType >
static grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > convert(const regexp::FormalRegExp < SymbolType > & regexp);
/**
* \override
*/
static grammar::RightRG < > convert(const regexp::FormalRegExp < > & regexp);
static grammar::RightRG < > convert(const regexp::UnboundedRegExp < > & regexp);
template < class SymbolType >
static grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > convert(const regexp::UnboundedRegExp < SymbolType > & regexp);
};
 
template < class SymbolType >
grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > ToGrammarRightRGGlushkov::convert ( const regexp::UnboundedRegExp < SymbolType > & regexp ) {
ext::pair < SymbolType, unsigned > S = ext::make_pair ( alphabet::InitialSymbol::instance < SymbolType > ( ), 0 );
grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > grammar ( S );
// step 1
grammar.setTerminalAlphabet ( regexp.getAlphabet ( ) );
regexp::UnboundedRegExp < ext::pair < SymbolType, unsigned > > indexedRegExp = regexp::GlushkovIndexate::index ( regexp );
// steps 2, 3, 4
const ext::set < regexp::UnboundedRegExpSymbol < ext::pair < SymbolType, unsigned > > > first = regexp::GlushkovFirst::first ( indexedRegExp );
const ext::set < regexp::UnboundedRegExpSymbol < ext::pair < SymbolType, unsigned > > > last = regexp::GlushkovLast::last ( indexedRegExp );
// \e in q0 check is in step 7
// step 5
for ( const auto & nonterminal : indexedRegExp.getAlphabet ( ) )
grammar.addNonterminalSymbol ( nonterminal );
// step 6
for ( const regexp::UnboundedRegExpSymbol < ext::pair < SymbolType, unsigned > > & symbol : first )
grammar.addRule ( S, ext::make_pair ( symbol.getSymbol ( ).first, symbol.getSymbol ( ) ) );
for ( const ext::pair < SymbolType, unsigned > & x : indexedRegExp.getAlphabet ( ) )
for ( const regexp::UnboundedRegExpSymbol < ext::pair < SymbolType, unsigned > > & f : regexp::GlushkovFollow::follow ( indexedRegExp, UnboundedRegExpSymbol < ext::pair < SymbolType, unsigned > > ( x ) ) ) {
const ext::pair < SymbolType, unsigned > & a = x;
const ext::pair < SymbolType, unsigned > & b = f.getSymbol ( );
grammar.addRule ( a, ext::make_pair ( b.first, b ) );
}
// step 7
/*
* for all rules where ns.m_nonTerminal is on rightSide:
* add Rule: leftSide -> symbol.getSymbol( )
* unless it already exists
*/
for ( const auto & rule : grammar.getRules ( ) )
for ( const auto & rhs : rule.second ) {
if ( ! rhs.template is < ext::pair < SymbolType, ext::pair < SymbolType, unsigned > > > ( ) )
continue;
const ext::pair < SymbolType, ext::pair < SymbolType, unsigned > > & rawRhs = rhs.template get < ext::pair < SymbolType, ext::pair < SymbolType, unsigned > > > ( );
for ( const regexp::UnboundedRegExpSymbol < ext::pair < SymbolType, unsigned > > & symbol : last )
if ( rawRhs.second == symbol.getSymbol ( ) )
grammar.addRule ( rule.first, rawRhs.first );
}
if ( regexp::properties::RegExpEpsilon::languageContainsEpsilon ( regexp ) )
grammar.setGeneratesEpsilon ( true );
return grammar;
}
template < class SymbolType >
grammar::RightRG < SymbolType, ext::pair < SymbolType, unsigned > > ToGrammarRightRGGlushkov::convert ( const regexp::FormalRegExp < SymbolType > & regexp ) {
return ToGrammarRightRGGlushkov::convert ( regexp::UnboundedRegExp < SymbolType > ( regexp ) );
}
} /* namespace convert */
 
} /* namespace regexp */
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment