Epsilon free CFG

alib2data/src/grammar/ContextFree/EpsilonFreeCFG.cpp

@@ -64,17 +64,8 @@ bool EpsilonFreeCFG::removeNonterminalSymbol(const alphabet::Symbol& symbol) {
 bool EpsilonFreeCFG::addRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
 	int rSize = rightHandSide.size();
-	if(leftHandSide == initialSymbol && rSize == 0) {
-		for(const auto& rule : rules) {
-			for(const auto& ruleRHS : rule.second) {
-				if(any_of(ruleRHS.begin(), ruleRHS.end(), [&](const alphabet::Symbol& symbol) { return initialSymbol == symbol; })) {
-					throw GrammarException("Initial symbol " + (std::string) initialSymbol + "used on right hand side of already existing rule");
-				}
-			}
-		}
-
-		generatesEpsilon = true;
-		return rules[leftHandSide].insert(rightHandSide).second;
+	if(rSize == 0) {
+		throw GrammarException("Epsilon rule is not allowed");
 	} else {
 		if(!nonterminalAlphabet.count(leftHandSide))
 			throw GrammarException("Rule must rewrite nonterminal symbol");
@@ -82,8 +73,6 @@ bool EpsilonFreeCFG::addRule(const alphabet::Symbol& leftHandSide, const std::ve
 		for(const alphabet::Symbol& symbol : rightHandSide) {
 			if(terminalAlphabet.find(symbol) == terminalAlphabet.end() && nonterminalAlphabet.find(symbol) == nonterminalAlphabet.end())
 				throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is not neither terminal nor nonterminal symbol");
-			if(generatesEpsilon && symbol == initialSymbol)
-				throw GrammarException("Initial symbol is already allowed to be rewritten to epsilon");
 		}
 		return rules[leftHandSide].insert(rightHandSide).second;
@@ -95,20 +84,23 @@ const std::map<alphabet::Symbol, std::set<std::vector<alphabet::Symbol>>> Epsilo
 }
 bool EpsilonFreeCFG::removeRule(const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
-	int rSize = rightHandSide.size();
-
-	if(leftHandSide == initialSymbol && rSize == 0) {
-		generatesEpsilon = false;
-	}
 	return rules[leftHandSide].erase(rightHandSide);
 }
+void EpsilonFreeCFG::setGeneratesEpsilon(bool genEps) {
+	generatesEpsilon = genEps;
+}
+
+bool EpsilonFreeCFG::getGeneratesEpsilon() const {
+	return generatesEpsilon;
+}
+
 bool EpsilonFreeCFG::operator==(const GrammarBase& other) const {
 	return other == *this;
 }
 bool EpsilonFreeCFG::operator==(const EpsilonFreeCFG& other) const {
-	return this->nonterminalAlphabet == other.nonterminalAlphabet && this->terminalAlphabet == other.terminalAlphabet && this->initialSymbol == other.initialSymbol && this->rules == other.rules;
+	return this->nonterminalAlphabet == other.nonterminalAlphabet && this->terminalAlphabet == other.terminalAlphabet && this->initialSymbol == other.initialSymbol && this->rules == other.rules && this->generatesEpsilon == other.generatesEpsilon;
 }
 void EpsilonFreeCFG::operator>>(std::ostream& out) const {
@@ -117,6 +109,7 @@ void EpsilonFreeCFG::operator>>(std::ostream& out) const {
 			<< "terminalAlphabet = " << terminalAlphabet
 			<< "initialSymbol = " << initialSymbol
 			<< "rules = " << rules
+			<< "generatesEpsilon = " << generatesEpsilon
 			<< ")";
 }

alib2data/src/grammar/ContextFree/EpsilonFreeCFG.h

@@ -40,6 +40,9 @@ public:
 	bool removeNonterminalSymbol(const alphabet::Symbol& symbol);
+	void setGeneratesEpsilon(bool genEps);
+	bool getGeneratesEpsilon() const;
+
 	virtual bool operator==(const GrammarBase& other) const;
 	virtual bool operator==(const EpsilonFreeCFG& other) const;

alib2data/src/grammar/GrammarFromXMLParser.cpp

@@ -211,6 +211,9 @@ EpsilonFreeCFG GrammarFromXMLParser::parseEpsilonFreeCFG(std::list<sax::Token>&
 	parseRules(input, grammar);
+	bool generatesEpsilon = parseGeneratesEpsilon(input);
+	grammar.setGeneratesEpsilon(generatesEpsilon);
+
 	popToken(input, sax::Token::TokenType::END_ELEMENT, "EpsilonFreeCFG");
 	return grammar;
 }

alib2data/src/grammar/GrammarToXMLComposer.cpp

@@ -126,6 +126,7 @@ std::list<sax::Token> GrammarToXMLComposer::compose(const EpsilonFreeCFG& gramma
 	composeTerminalAlphabet(out, grammar.getTerminalAlphabet());
 	composeInitialSymbol(out, grammar.getInitialSymbol());
 	composeRules(out, grammar);
+	composeGeneratesEpsilon(out, grammar.getGeneratesEpsilon());
 	out.push_back(sax::Token("EpsilonFreeCFG", sax::Token::TokenType::END_ELEMENT));
 	return out;

alib2data/test-src/grammar/GrammarTest.cpp

@@ -170,6 +170,36 @@ void GrammarTest::testContextFreeParser() {
 	grammar::GrammarFromXMLParser parser;
 	grammar::LG grammar2 = parser.parseLG(tokens2);
+	CPPUNIT_ASSERT( grammar == grammar2 );
+    }
+  }
+  {
+    grammar::EpsilonFreeCFG 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.setGeneratesEpsilon(true);
+
+    grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), 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::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::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3))))});
+    grammar.addRule(alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), 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::EpsilonFreeCFG grammar2 = parser.parseEpsilonFreeCFG(tokens2);
+
 	CPPUNIT_ASSERT( grammar == grammar2 );
     }
   }