From 6e5002422fcca8e37c432bf15e09da6eb5a3e0db Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Wed, 16 Jul 2014 21:49:14 +0200
Subject: [PATCH] New internal form for Right/Left LG
---
.../src/grammar/GrammarFromXMLParser.cpp | 16 ++--
.../src/grammar/GrammarToXMLComposer.cpp | 48 +++++++++-
alib2data/src/grammar/GrammarToXMLComposer.h | 2 +
alib2data/src/grammar/Regular/LeftLG.cpp | 89 +++++++++++++++----
alib2data/src/grammar/Regular/LeftLG.h | 11 ++-
alib2data/src/grammar/Regular/RightLG.cpp | 89 +++++++++++++++----
alib2data/src/grammar/Regular/RightLG.h | 11 ++-
alib2data/test-src/grammar/GrammarTest.cpp | 54 ++++++++---
8 files changed, 262 insertions(+), 58 deletions(-)
diff --git a/alib2data/src/grammar/GrammarFromXMLParser.cpp b/alib2data/src/grammar/GrammarFromXMLParser.cpp
index cd0e84d963..8f57b7b443 100644
--- a/alib2data/src/grammar/GrammarFromXMLParser.cpp
+++ b/alib2data/src/grammar/GrammarFromXMLParser.cpp
@@ -270,7 +270,7 @@ RightRG GrammarFromXMLParser::parseRightRG(std::list<sax::Token>& input) const {
}
RightLG GrammarFromXMLParser::parseRightLG(std::list<sax::Token>& input) const {
- popToken(input, sax::Token::TokenType::START_ELEMENT, "LeftLG");
+ popToken(input, sax::Token::TokenType::START_ELEMENT, "RightLG");
std::set<alphabet::Symbol> nonterminalAlphabet = parseNonterminalAlphabet(input);
std::set<alphabet::Symbol> terminalAlphabet = parseTerminalAlphabet(input);
@@ -457,7 +457,7 @@ void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, RightRG& gram
void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, RightLG& grammar) const {
alphabet::Symbol lhs = parseRuleSingleSymbolLHS(input);
std::vector<alphabet::Symbol> rhs = parseRuleRHS(input);
- grammar.addRule(lhs, rhs);
+ grammar.addRawRule(lhs, rhs);
}
void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, LeftRG& grammar) const {
@@ -469,14 +469,17 @@ void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, LeftRG& gramm
void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, LeftLG& grammar) const {
alphabet::Symbol lhs = parseRuleSingleSymbolLHS(input);
std::vector<alphabet::Symbol> rhs = parseRuleRHS(input);
- grammar.addRule(lhs, rhs);
+ grammar.addRawRule(lhs, rhs);
}
std::vector<alphabet::Symbol> GrammarFromXMLParser::parseRuleLHS(std::list<sax::Token>& input) const {
std::vector<alphabet::Symbol> lhs;
popToken(input, sax::Token::TokenType::START_ELEMENT, "lhs");
- while (isTokenType(input, sax::Token::TokenType::START_ELEMENT)) {
+ if(isToken(input, sax::Token::TokenType::START_ELEMENT, "epsilon")) {
+ input.pop_front();
+ popToken(input, sax::Token::TokenType::END_ELEMENT, "epsilon");
+ } else while (isTokenType(input, sax::Token::TokenType::START_ELEMENT)) {
lhs.push_back(alib::FromXMLParsers::symbolParser.parse(input));
}
popToken(input, sax::Token::TokenType::END_ELEMENT, "lhs");
@@ -493,7 +496,10 @@ alphabet::Symbol GrammarFromXMLParser::parseRuleSingleSymbolLHS(std::list<sax::T
std::vector<alphabet::Symbol> GrammarFromXMLParser::parseRuleRHS(std::list<sax::Token>& input) const {
std::vector<alphabet::Symbol> rhs;
popToken(input, sax::Token::TokenType::START_ELEMENT, "rhs");
- while (isTokenType(input, sax::Token::TokenType::START_ELEMENT)) {
+ if(isToken(input, sax::Token::TokenType::START_ELEMENT, "epsilon")) {
+ input.pop_front();
+ popToken(input, sax::Token::TokenType::END_ELEMENT, "epsilon");
+ } else while (isTokenType(input, sax::Token::TokenType::START_ELEMENT)) {
rhs.push_back(alib::FromXMLParsers::symbolParser.parse(input));
}
popToken(input, sax::Token::TokenType::END_ELEMENT, "rhs");
diff --git a/alib2data/src/grammar/GrammarToXMLComposer.cpp b/alib2data/src/grammar/GrammarToXMLComposer.cpp
index f76dcce81d..d3f2e027a5 100644
--- a/alib2data/src/grammar/GrammarToXMLComposer.cpp
+++ b/alib2data/src/grammar/GrammarToXMLComposer.cpp
@@ -422,7 +422,7 @@ void GrammarToXMLComposer::composeRules(std::list<sax::Token>& out, const RightL
out.push_back(sax::Token("rule", sax::Token::TokenType::START_ELEMENT));
composeRuleSingleSymbolLHS(out, rule.first);
- composeRuleRHS(out, rhs);
+ composeRuleRightLGRHS(out, rhs);
out.push_back(sax::Token("rule", sax::Token::TokenType::END_ELEMENT));
}
@@ -456,7 +456,7 @@ void GrammarToXMLComposer::composeRules(std::list<sax::Token>& out, const LeftLG
out.push_back(sax::Token("rule", sax::Token::TokenType::START_ELEMENT));
composeRuleSingleSymbolLHS(out, rule.first);
- composeRuleRHS(out, rhs);
+ composeRuleLeftLGRHS(out, rhs);
out.push_back(sax::Token("rule", sax::Token::TokenType::END_ELEMENT));
}
@@ -516,6 +516,50 @@ void GrammarToXMLComposer::composeRuleOneOrTwoSymbolsRHS(std::list<sax::Token>&
out.push_back(sax::Token("rhs", sax::Token::TokenType::END_ELEMENT));
}
+void GrammarToXMLComposer::composeRuleLeftLGRHS(std::list<sax::Token>& out, const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>>& symbols) const {
+ out.push_back(sax::Token("rhs", sax::Token::TokenType::START_ELEMENT));
+ if(symbols.is<std::vector<alphabet::Symbol>>()) {
+ const std::vector<alphabet::Symbol>& rhs = symbols.get<std::vector<alphabet::Symbol>>();
+ if(rhs.size() == 0) {
+ out.push_back(sax::Token("epsilon", sax::Token::TokenType::START_ELEMENT));
+ out.push_back(sax::Token("epsilon", sax::Token::TokenType::END_ELEMENT));
+ } else {
+ for (const auto& symbol : symbols.get<std::vector<alphabet::Symbol>>()) {
+ out.splice(out.end(), alib::ToXMLComposers::symbolComposer.compose(symbol));
+ }
+ }
+ } else {
+ const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rhs = symbols.get<std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>>();
+ out.splice(out.end(), alib::ToXMLComposers::symbolComposer.compose(rhs.first));
+ for (const auto& symbol : rhs.second) {
+ out.splice(out.end(), alib::ToXMLComposers::symbolComposer.compose(symbol));
+ }
+ }
+ out.push_back(sax::Token("rhs", sax::Token::TokenType::END_ELEMENT));
+}
+
+void GrammarToXMLComposer::composeRuleRightLGRHS(std::list<sax::Token>& out, const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>>& symbols) const {
+ out.push_back(sax::Token("rhs", sax::Token::TokenType::START_ELEMENT));
+ if(symbols.is<std::vector<alphabet::Symbol>>()) {
+ const std::vector<alphabet::Symbol>& rhs = symbols.get<std::vector<alphabet::Symbol>>();
+ if(rhs.size() == 0) {
+ out.push_back(sax::Token("epsilon", sax::Token::TokenType::START_ELEMENT));
+ out.push_back(sax::Token("epsilon", sax::Token::TokenType::END_ELEMENT));
+ } else {
+ for (const auto& symbol : symbols.get<std::vector<alphabet::Symbol>>()) {
+ out.splice(out.end(), alib::ToXMLComposers::symbolComposer.compose(symbol));
+ }
+ }
+ } else {
+ const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rhs = symbols.get<std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>>();
+ for (const auto& symbol : rhs.first) {
+ out.splice(out.end(), alib::ToXMLComposers::symbolComposer.compose(symbol));
+ }
+ out.splice(out.end(), alib::ToXMLComposers::symbolComposer.compose(rhs.second));
+ }
+ out.push_back(sax::Token("rhs", sax::Token::TokenType::END_ELEMENT));
+}
+
std::list<sax::Token> GrammarToXMLComposer::compose(const Grammar& grammar) const {
std::list<sax::Token> out;
grammar.getGrammar().Accept((void*) &out, *this);
diff --git a/alib2data/src/grammar/GrammarToXMLComposer.h b/alib2data/src/grammar/GrammarToXMLComposer.h
index 1d501d8b15..32268126b1 100644
--- a/alib2data/src/grammar/GrammarToXMLComposer.h
+++ b/alib2data/src/grammar/GrammarToXMLComposer.h
@@ -73,6 +73,8 @@ class GrammarToXMLComposer : public GrammarBase::const_visitor_type {
void composeRuleSingleSymbolLHS(std::list<sax::Token>& out, const alphabet::Symbol& symbol) const;
void composeRuleRHS(std::list<sax::Token>& out, const std::vector<alphabet::Symbol>& symbols) const;
void composeRuleOneOrTwoSymbolsRHS(std::list<sax::Token>& out, const std::variant<alphabet::Symbol, std::pair<alphabet::Symbol, alphabet::Symbol>>& symbols) const;
+ void composeRuleLeftLGRHS(std::list<sax::Token>& out, const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>>& symbols) const;
+ void composeRuleRightLGRHS(std::list<sax::Token>& out, const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>>& symbols) const;
public:
/**
* Prints XML representation of UnknownGrammar to the output stream.
diff --git a/alib2data/src/grammar/Regular/LeftLG.cpp b/alib2data/src/grammar/Regular/LeftLG.cpp
index 7c35514090..29db7f37bc 100644
--- a/alib2data/src/grammar/Regular/LeftLG.cpp
+++ b/alib2data/src/grammar/Regular/LeftLG.cpp
@@ -32,26 +32,39 @@ GrammarBase* LeftLG::plunder() && {
}
bool LeftLG::removeTerminalSymbol(const alphabet::Symbol& symbol) {
- for(const std::pair<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
+ for(const std::pair<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >> >& rule : rules) {
if(rule.first == symbol)
throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
- for(const std::vector<alphabet::Symbol>& rhs : rule.second)
- if(std::find(rhs.begin(), rhs.end(), symbol) != rhs.end())
- throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ for(const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >& rhsTmp : rule.second)
+ if(rhsTmp.is<std::vector<alphabet::Symbol>>()) {
+ const std::vector<alphabet::Symbol>& rhs = rhsTmp.get<std::vector<alphabet::Symbol>>();
+
+ if(std::find(rhs.begin(), rhs.end(), symbol) != rhs.end())
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ } else {
+ const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rhs = rhsTmp.get<std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>>();
+
+ if(std::find(rhs.second.begin(), rhs.second.end(), symbol) != rhs.second.end())
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ }
}
return terminalAlphabet.erase(symbol);
}
bool LeftLG::removeNonterminalSymbol(const alphabet::Symbol& symbol) {
- for(const std::pair<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
+ for(const std::pair<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >> >& rule : rules) {
if(rule.first == symbol)
throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
- for(const std::vector<alphabet::Symbol>& rhs : rule.second)
- if(std::find(rhs.begin(), rhs.end(), symbol) != rhs.end())
- throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ for(const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >& rhsTmp : rule.second)
+ if(rhsTmp.is<std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>>()) {
+ const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rhs = rhsTmp.get<std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>>();
+
+ if(rhs.first == symbol)
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ }
}
if(initialSymbol == symbol)
@@ -61,31 +74,71 @@ bool LeftLG::removeNonterminalSymbol(const alphabet::Symbol& symbol) {
return nonterminalAlphabet.erase(symbol);
}
-bool LeftLG::addRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+bool LeftLG::addRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >& rightHandSide) {
if(!nonterminalAlphabet.count(leftHandSide))
throw GrammarException("Rule must rewrite nonterminal symbol");
- std::vector<alphabet::Symbol>::const_iterator iter = rightHandSide.begin();
- if(iter == rightHandSide.end()) return rules[leftHandSide].insert(rightHandSide).second;
+ if(rightHandSide.is<std::vector<alphabet::Symbol>>()) {
+ const std::vector<alphabet::Symbol>& rhs = rightHandSide.get<std::vector<alphabet::Symbol>>();
+ for(const auto & symbol : rhs ) {
+ if(terminalAlphabet.find(symbol) == terminalAlphabet.end())
+ throw GrammarException("Symbol " + (std::string) symbol.getSymbol() + " is not a terminal symbol");
+ }
+
+ return rules[leftHandSide].insert(rightHandSide).second;
+ } else {
+ const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rhs = rightHandSide.get<std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>>();
- if(nonterminalAlphabet.find(*iter) != nonterminalAlphabet.end() && terminalAlphabet.find(*iter) != terminalAlphabet.end()) throw GrammarException("Symbol " + (std::string) iter->getSymbol() + " is not a terminal nor nonterminal symbol");
- iter++;
+ if(nonterminalAlphabet.find(rhs.first) == nonterminalAlphabet.end())
+ throw GrammarException("Symbol " + (std::string) rhs.first.getSymbol() + " is not a nonterminal symbol");
- for(; iter != rightHandSide.end() && terminalAlphabet.find(*iter) != terminalAlphabet.end(); iter++);
+ for(const auto & symbol : rhs.second ) {
+ if(terminalAlphabet.find(symbol) == terminalAlphabet.end())
+ throw GrammarException("Symbol " + (std::string) symbol.getSymbol() + " is not a terminal symbol");
+ }
- if(iter == rightHandSide.end()) return rules[leftHandSide].insert(rightHandSide).second;
+ return rules[leftHandSide].insert(rightHandSide).second;
+ }
+}
+
+bool LeftLG::addRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> > rhs(rightHandSide);
+ return addRule(leftHandSide, rhs);
+}
- throw GrammarException("Symbol " + (std::string) iter->getSymbol() + " is not a terminal symbol");
+bool LeftLG::addRule(const alphabet::Symbol& leftHandSide, const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> > rhs(rightHandSide);
+ return addRule(leftHandSide, rhs);
}
-const std::map<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>> LeftLG::getRules() const {
+bool LeftLG::addRawRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+ if(rightHandSide.size() == 0) {
+ return addRule(leftHandSide, rightHandSide);
+ } else if(nonterminalAlphabet.count(rightHandSide[0])) {
+ return addRule(leftHandSide, std::make_pair(rightHandSide[0], std::vector<alphabet::Symbol>(rightHandSide.begin() + 1, rightHandSide.end())));
+ } else {
+ return addRule(leftHandSide, rightHandSide);
+ }
+}
+
+const std::map<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >> > LeftLG::getRules() const {
return rules;
}
-bool LeftLG::removeRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+bool LeftLG::removeRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >& rightHandSide) {
return rules[leftHandSide].erase(rightHandSide);
}
+bool LeftLG::removeRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> > rhs(rightHandSide);
+ return removeRule(leftHandSide, rhs);
+}
+
+bool LeftLG::removeRule(const alphabet::Symbol& leftHandSide, const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> > rhs(rightHandSide);
+ return removeRule(leftHandSide, rhs);
+}
+
bool LeftLG::operator==(const GrammarBase& other) const {
return other == *this;
}
diff --git a/alib2data/src/grammar/Regular/LeftLG.h b/alib2data/src/grammar/Regular/LeftLG.h
index 8127000f7f..b251bb09c2 100644
--- a/alib2data/src/grammar/Regular/LeftLG.h
+++ b/alib2data/src/grammar/Regular/LeftLG.h
@@ -11,6 +11,7 @@
#include "../GrammarBase.h"
#include <map>
#include <vector>
+#include "../../std/variant.hpp"
#include "../common/TerminalNonterminalAlphabetInitialSymbol.h"
namespace grammar {
@@ -19,7 +20,7 @@ namespace grammar {
* Left linear grammar in chomsky normal form. Type 3 in Chomsky hierarchy. Produces regular languages.
*/
class LeftLG : public std::element<LeftLG, GrammarBase>, public TerminalNonterminalAlphabetInitialSymbol {
- std::map<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>> rules;
+ std::map<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >> > rules;
public:
LeftLG(const alphabet::Symbol& initialSymbol);
@@ -29,11 +30,17 @@ public:
virtual GrammarBase* plunder() &&;
+ bool addRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >& rightHandSide);
bool addRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ bool addRule(const alphabet::Symbol& leftHandSide, const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rightHandSide);
- const std::map<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>> getRules() const;
+ bool addRawRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ const std::map<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >> > getRules() const;
+
+ bool removeRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>> >& rightHandSide);
bool removeRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ bool removeRule(const alphabet::Symbol& leftHandSide, const std::pair<alphabet::Symbol, std::vector<alphabet::Symbol>>& rightHandSide);
bool removeTerminalSymbol(const alphabet::Symbol& symbol);
diff --git a/alib2data/src/grammar/Regular/RightLG.cpp b/alib2data/src/grammar/Regular/RightLG.cpp
index 39a4da54ff..a124a0151a 100644
--- a/alib2data/src/grammar/Regular/RightLG.cpp
+++ b/alib2data/src/grammar/Regular/RightLG.cpp
@@ -32,26 +32,39 @@ GrammarBase* RightLG::plunder() && {
}
bool RightLG::removeTerminalSymbol(const alphabet::Symbol& symbol) {
- for(const std::pair<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
+ for(const std::pair<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >> >& rule : rules) {
if(rule.first == symbol)
throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
- for(const std::vector<alphabet::Symbol>& rhs : rule.second)
- if(std::find(rhs.begin(), rhs.end(), symbol) != rhs.end())
- throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ for(const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >& rhsTmp : rule.second)
+ if(rhsTmp.is<std::vector<alphabet::Symbol>>()) {
+ const std::vector<alphabet::Symbol>& rhs = rhsTmp.get<std::vector<alphabet::Symbol>>();
+
+ if(std::find(rhs.begin(), rhs.end(), symbol) != rhs.end())
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ } else {
+ const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rhs = rhsTmp.get<std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>>();
+
+ if(std::find(rhs.first.begin(), rhs.first.end(), symbol) != rhs.first.end())
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ }
}
return terminalAlphabet.erase(symbol);
}
bool RightLG::removeNonterminalSymbol(const alphabet::Symbol& symbol) {
- for(const std::pair<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
+ for(const std::pair<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >> >& rule : rules) {
if(rule.first == symbol)
throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
- for(const std::vector<alphabet::Symbol>& rhs : rule.second)
- if(std::find(rhs.begin(), rhs.end(), symbol) != rhs.end())
- throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ for(const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >& rhsTmp : rule.second)
+ if(rhsTmp.is<std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>>()) {
+ const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rhs = rhsTmp.get<std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>>();
+
+ if(rhs.second == symbol)
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ }
}
if(initialSymbol == symbol)
@@ -61,31 +74,71 @@ bool RightLG::removeNonterminalSymbol(const alphabet::Symbol& symbol) {
return nonterminalAlphabet.erase(symbol);
}
-bool RightLG::addRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+bool RightLG::addRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >& rightHandSide) {
if(!nonterminalAlphabet.count(leftHandSide))
throw GrammarException("Rule must rewrite nonterminal symbol");
- std::vector<alphabet::Symbol>::const_iterator iter = rightHandSide.begin();
- for(; iter != rightHandSide.end() && terminalAlphabet.find(*iter) != terminalAlphabet.end(); iter++);
+ if(rightHandSide.is<std::vector<alphabet::Symbol>>()) {
+ const std::vector<alphabet::Symbol>& rhs = rightHandSide.get<std::vector<alphabet::Symbol>>();
+ for(const auto & symbol : rhs ) {
+ if(terminalAlphabet.find(symbol) == terminalAlphabet.end())
+ throw GrammarException("Symbol " + (std::string) symbol.getSymbol() + " is not a terminal symbol");
+ }
+
+ return rules[leftHandSide].insert(rightHandSide).second;
+ } else {
+ const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rhs = rightHandSide.get<std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>>();
- if(iter == rightHandSide.end()) return rules[leftHandSide].insert(rightHandSide).second;
+ if(nonterminalAlphabet.find(rhs.second) == nonterminalAlphabet.end())
+ throw GrammarException("Symbol " + (std::string) rhs.second.getSymbol() + " is not a nonterminal symbol");
- if(nonterminalAlphabet.find(*iter) == nonterminalAlphabet.end()) throw GrammarException("Symbol " + (std::string) iter->getSymbol() + " is not a terminal nor nonterminal symbol");
- iter++;
+ for(const auto & symbol : rhs.first ) {
+ if(terminalAlphabet.find(symbol) == terminalAlphabet.end())
+ throw GrammarException("Symbol " + (std::string) symbol.getSymbol() + " is not a terminal symbol");
+ }
- if(iter == rightHandSide.end()) return rules[leftHandSide].insert(rightHandSide).second;
+ return rules[leftHandSide].insert(rightHandSide).second;
+ }
+}
+
+bool RightLG::addRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> > rhs(rightHandSide);
+ return addRule(leftHandSide, rhs);
+}
- throw GrammarException("Symbol " + (std::string) iter->getSymbol() + " is not a terminal symbol");
+bool RightLG::addRule(const alphabet::Symbol& leftHandSide, const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> > rhs(rightHandSide);
+ return addRule(leftHandSide, rhs);
}
-const std::map<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>> RightLG::getRules() const {
+bool RightLG::addRawRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+ if(rightHandSide.size() == 0) {
+ return addRule(leftHandSide, rightHandSide);
+ } else if(nonterminalAlphabet.count(rightHandSide[rightHandSide.size() - 1])) {
+ return addRule(leftHandSide, std::make_pair(std::vector<alphabet::Symbol>(rightHandSide.begin(), rightHandSide.end() - 1), rightHandSide[rightHandSide.size() - 1]));
+ } else {
+ return addRule(leftHandSide, rightHandSide);
+ }
+}
+
+const std::map<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >> > RightLG::getRules() const {
return rules;
}
-bool RightLG::removeRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+bool RightLG::removeRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >& rightHandSide) {
return rules[leftHandSide].erase(rightHandSide);
}
+bool RightLG::removeRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> > rhs(rightHandSide);
+ return removeRule(leftHandSide, rhs);
+}
+
+bool RightLG::removeRule(const alphabet::Symbol& leftHandSide, const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rightHandSide) {
+ std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> > rhs(rightHandSide);
+ return removeRule(leftHandSide, rhs);
+}
+
bool RightLG::operator==(const GrammarBase& other) const {
return other == *this;
}
diff --git a/alib2data/src/grammar/Regular/RightLG.h b/alib2data/src/grammar/Regular/RightLG.h
index 9cba0827f4..3446bf5774 100644
--- a/alib2data/src/grammar/Regular/RightLG.h
+++ b/alib2data/src/grammar/Regular/RightLG.h
@@ -11,6 +11,7 @@
#include "../GrammarBase.h"
#include <map>
#include <vector>
+#include "../../std/variant.hpp"
#include "../common/TerminalNonterminalAlphabetInitialSymbol.h"
namespace grammar {
@@ -19,7 +20,7 @@ namespace grammar {
* Right linear grammar in chomsky normal form. Type 3 in Chomsky hierarchy. Produces regular languages.
*/
class RightLG : public std::element<RightLG, GrammarBase>, public TerminalNonterminalAlphabetInitialSymbol {
- std::map<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>> rules;
+ std::map<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >> > rules;
public:
RightLG(const alphabet::Symbol& initialSymbol);
@@ -29,11 +30,17 @@ public:
virtual GrammarBase* plunder() &&;
+ bool addRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >& rightHandSide);
bool addRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ bool addRule(const alphabet::Symbol& leftHandSide, const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rightHandSide);
- const std::map<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>> getRules() const;
+ bool addRawRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ const std::map<alphabet::Symbol, std::set<std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >> > getRules() const;
+
+ bool removeRule(const alphabet::Symbol& leftHandSide, const std::variant<std::vector<alphabet::Symbol>, std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol> >& rightHandSide);
bool removeRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ bool removeRule(const alphabet::Symbol& leftHandSide, const std::pair<std::vector<alphabet::Symbol>, alphabet::Symbol>& rightHandSide);
bool removeTerminalSymbol(const alphabet::Symbol& symbol);
diff --git a/alib2data/test-src/grammar/GrammarTest.cpp b/alib2data/test-src/grammar/GrammarTest.cpp
index b150fe8d52..f85be63828 100644
--- a/alib2data/test-src/grammar/GrammarTest.cpp
+++ b/alib2data/test-src/grammar/GrammarTest.cpp
@@ -84,19 +84,20 @@ void GrammarTest::testUnrestrictedParser() {
}
void GrammarTest::testRegularParser() {
- grammar::RightRG grammar(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))));
+ {
+ grammar::RightRG grammar(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))));
- grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))));
- grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))));
- grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3)))));
- grammar.addTerminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))));
- grammar.addTerminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("b")))));
-
- grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), std::make_pair(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2))))));
- grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))), std::make_pair(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("b")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3))))));
+ grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))));
+ grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))));
+ grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3)))));
+ grammar.addTerminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))));
+ grammar.addTerminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("b")))));
- CPPUNIT_ASSERT( grammar == grammar );
- {
+ grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), std::make_pair(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2))))));
+ grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))), std::make_pair(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("b")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3))))));
+
+ CPPUNIT_ASSERT( grammar == grammar );
+ {
grammar::GrammarToXMLComposer composer;
std::list<sax::Token> tokens = composer.compose(grammar);
std::string tmp;
@@ -108,5 +109,36 @@ void GrammarTest::testRegularParser() {
grammar::RightRG grammar2 = parser.parseRightRG(tokens2);
CPPUNIT_ASSERT( grammar == grammar2 );
+ }
+ }
+
+ {
+ grammar::RightLG grammar(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))));
+
+ grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))));
+ grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))));
+ grammar.addNonterminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3)))));
+ grammar.addTerminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))));
+ grammar.addTerminalSymbol(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("b")))));
+
+ grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), std::make_pair(std::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a"))))}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2))))));
+ grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))), std::make_pair(std::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("b"))))}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3))))));
+ grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), std::make_pair(std::vector<alphabet::Symbol> {}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3))))));
+ grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), std::make_pair(std::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a"))))}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2))))));
+
+ CPPUNIT_ASSERT( grammar == grammar );
+ {
+ grammar::GrammarToXMLComposer composer;
+ std::list<sax::Token> tokens = composer.compose(grammar);
+ std::string tmp;
+ sax::SaxComposeInterface::printMemory(tmp, tokens);
+
+ std::list<sax::Token> tokens2;
+ sax::SaxParseInterface::parseMemory(tmp, tokens2);
+ grammar::GrammarFromXMLParser parser;
+ grammar::RightLG grammar2 = parser.parseRightLG(tokens2);
+
+ CPPUNIT_ASSERT( grammar == grammar2 );
+ }
}
}
--
GitLab