From 8e96c8b6d8adab3846ccaf2575864a3f508b4680 Mon Sep 17 00:00:00 2001
From: Jan Travnicek <Jan.Travnicek@fit.cvut.cz>
Date: Fri, 18 Jul 2014 10:34:18 +0200
Subject: [PATCH] ContextPreservingUnrestrictedGrammar internal form
---
.../src/grammar/GrammarFromXMLParser.cpp | 6 +-
.../src/grammar/GrammarToXMLComposer.cpp | 4 +-
.../ContextPreservingUnrestrictedGrammar.cpp | 64 ++++++++++---------
.../ContextPreservingUnrestrictedGrammar.h | 11 ++--
alib2data/test-src/grammar/GrammarTest.cpp | 29 +++++++++
5 files changed, 74 insertions(+), 40 deletions(-)
diff --git a/alib2data/src/grammar/GrammarFromXMLParser.cpp b/alib2data/src/grammar/GrammarFromXMLParser.cpp
index 3bff447a56..547cd3d624 100644
--- a/alib2data/src/grammar/GrammarFromXMLParser.cpp
+++ b/alib2data/src/grammar/GrammarFromXMLParser.cpp
@@ -416,9 +416,11 @@ void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, UnrestrictedG
}
void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, ContextPreservingUnrestrictedGrammar& grammar) const {
- std::vector<alphabet::Symbol> lhs = parseRuleLHS(input);
+ std::vector<alphabet::Symbol> lContext = parseRuleLContext(input);
+ alphabet::Symbol lhs = parseRuleSingleSymbolLHS(input);
+ std::vector<alphabet::Symbol> rContext = parseRuleRContext(input);
std::vector<alphabet::Symbol> rhs = parseRuleRHS(input);
- grammar.addRule(lhs, rhs);
+ grammar.addRule(lContext, lhs, rContext, rhs);
}
void GrammarFromXMLParser::parseRule(std::list<sax::Token>& input, NonContractingGrammar& grammar) const {
diff --git a/alib2data/src/grammar/GrammarToXMLComposer.cpp b/alib2data/src/grammar/GrammarToXMLComposer.cpp
index 41a592d3d1..90a648879d 100644
--- a/alib2data/src/grammar/GrammarToXMLComposer.cpp
+++ b/alib2data/src/grammar/GrammarToXMLComposer.cpp
@@ -290,7 +290,9 @@ void GrammarToXMLComposer::composeRules(std::list<sax::Token>& out, const Contex
for(const auto& rhs : rule.second) {
out.push_back(sax::Token("rule", sax::Token::TokenType::START_ELEMENT));
- composeRuleLHS(out, rule.first);
+ composeRuleLContext(out, std::get<0>(rule.first));
+ composeRuleSingleSymbolLHS(out, std::get<1>(rule.first));
+ composeRuleRContext(out, std::get<2>(rule.first));
composeRuleRHS(out, rhs);
out.push_back(sax::Token("rule", sax::Token::TokenType::END_ELEMENT));
diff --git a/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.cpp b/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.cpp
index 3cf0060277..6938804fc1 100644
--- a/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.cpp
+++ b/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.cpp
@@ -32,9 +32,14 @@ GrammarBase* ContextPreservingUnrestrictedGrammar::plunder() && {
}
bool ContextPreservingUnrestrictedGrammar::removeTerminalSymbol(const alphabet::Symbol& symbol) {
- for(const std::pair<std::vector<alphabet::Symbol>, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
- if(std::find(rule.first.begin(), rule.first.end(), symbol) != rule.first.end())
- throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ for(const std::pair<std::tuple<std::vector<alphabet::Symbol>, alphabet::Symbol, std::vector<alphabet::Symbol>>, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
+ for(const alphabet::Symbol& lCont : std::get<0>(rule.first))
+ if(lCont == symbol)
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+
+ for(const alphabet::Symbol& rCont : std::get<2>(rule.first))
+ if(rCont == 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())
@@ -45,10 +50,18 @@ bool ContextPreservingUnrestrictedGrammar::removeTerminalSymbol(const alphabet::
}
bool ContextPreservingUnrestrictedGrammar::removeNonterminalSymbol(const alphabet::Symbol& symbol) {
- for(const std::pair<std::vector<alphabet::Symbol>, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
- if(std::find(rule.first.begin(), rule.first.end(), symbol) != rule.first.end())
+ for(const std::pair<std::tuple<std::vector<alphabet::Symbol>, alphabet::Symbol, std::vector<alphabet::Symbol>>, std::set<std::vector<alphabet::Symbol>>>& rule : rules) {
+ for(const alphabet::Symbol& lCont : std::get<0>(rule.first))
+ if(lCont == symbol)
+ throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+
+ if(std::get<1>(rule.first) == symbol)
throw GrammarException("Symbol \"" + (std::string) symbol.getSymbol() + "\" is used in rule.");
+ for(const alphabet::Symbol& rCont : std::get<2>(rule.first))
+ if(rCont == 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.");
@@ -61,45 +74,34 @@ bool ContextPreservingUnrestrictedGrammar::removeNonterminalSymbol(const alphabe
return nonterminalAlphabet.erase(symbol);
}
-bool ContextPreservingUnrestrictedGrammar::addRule(const std::vector<alphabet::Symbol>& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
- int lSize = leftHandSide.size();
- int rSize = rightHandSide.size();
-
- if(lSize > rSize + 1)
- throw GrammarException("Invalid size of right hand side of a rule");
-
- int lContext;
- int rContext;
- for(lContext = 0; lContext < lSize - 1 && leftHandSide[lContext] == rightHandSide[lContext]; lContext++);
- for(rContext = 0; rContext < lSize - 1 && leftHandSide[lSize - rContext] == rightHandSide[rSize - rContext]; rContext++);
-
- if(lContext + rContext + 1 < lSize) {
- throw GrammarException("Rule must rewrite only one symbol");
- } else
- if(lContext + rContext + 1 == lSize) {
- if(!nonterminalAlphabet.count(leftHandSide[lContext - 1]) && !nonterminalAlphabet.count(leftHandSide[lSize - rContext])) throw GrammarException("Rule must rewrite nonterminal symbol");
- } else
- if(/* lContext + rContext + 1 > lSize */ std::all_of(leftHandSide.end() - rContext, leftHandSide.begin() + lContext + 1, [&](const alphabet::Symbol symbol) {return !nonterminalAlphabet.count(symbol);}))
- throw GrammarException("Rule must rewrite nonterminal symbol");
+bool ContextPreservingUnrestrictedGrammar::addRule(const std::vector<alphabet::Symbol>& lContext, const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rContext, const std::vector<alphabet::Symbol>& rightHandSide) {
+ for(const alphabet::Symbol& symbol : lContext) {
+ 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(!nonterminalAlphabet.count(leftHandSide))
+ throw GrammarException("Rule must rewrite nonterminal symbol");
- for(const alphabet::Symbol& symbol : leftHandSide)
+ for(const alphabet::Symbol& symbol : rContext) {
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");
+ }
- for(const alphabet::Symbol& symbol : rightHandSide)
+ 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");
+ }
- return rules[leftHandSide].insert(rightHandSide).second;
+ return rules[make_tuple(lContext, leftHandSide, rContext)].insert(rightHandSide).second;
}
-const std::map<std::vector<alphabet::Symbol>, std::set<std::vector<alphabet::Symbol>>> ContextPreservingUnrestrictedGrammar::getRules() const {
+const std::map<std::tuple<std::vector<alphabet::Symbol>, alphabet::Symbol, std::vector<alphabet::Symbol>>, std::set<std::vector<alphabet::Symbol>>> ContextPreservingUnrestrictedGrammar::getRules() const {
return rules;
}
-bool ContextPreservingUnrestrictedGrammar::removeRule(const std::vector<alphabet::Symbol>& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide) {
- return rules[leftHandSide].erase(rightHandSide);
+bool ContextPreservingUnrestrictedGrammar::removeRule(const std::vector<alphabet::Symbol>& lContext, const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rContext, const std::vector<alphabet::Symbol>& rightHandSide) {
+ return rules[make_tuple(lContext, leftHandSide, rContext)].erase(rightHandSide);
}
bool ContextPreservingUnrestrictedGrammar::operator==(const GrammarBase& other) const {
diff --git a/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.h b/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.h
index 079e023c92..81784fb841 100644
--- a/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.h
+++ b/alib2data/src/grammar/Unrestricted/ContextPreservingUnrestrictedGrammar.h
@@ -16,11 +16,10 @@
namespace grammar {
/**
- * Context preserving unrestricted grammar. Type 0 in Chomsky hierarchy. Produces recursively enumerable language.
+ * Epsilon free context free grammar. Type 2 in Chomsky hierarchy. Produces context free languages.
*/
class ContextPreservingUnrestrictedGrammar : public std::element<ContextPreservingUnrestrictedGrammar, GrammarBase>, public TerminalNonterminalAlphabetInitialSymbol {
- std::map<std::vector<alphabet::Symbol>, std::set<std::vector<alphabet::Symbol>>> rules;
-
+ std::map<std::tuple<std::vector<alphabet::Symbol>, alphabet::Symbol, std::vector<alphabet::Symbol>>, std::set<std::vector<alphabet::Symbol>>> rules;
public:
ContextPreservingUnrestrictedGrammar(const alphabet::Symbol& initialSymbol);
@@ -30,11 +29,11 @@ public:
virtual GrammarBase* plunder() &&;
- bool addRule(const std::vector<alphabet::Symbol>& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ bool addRule(const std::vector<alphabet::Symbol>& lContext, const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rContext, const std::vector<alphabet::Symbol>& rightHandSide);
- const std::map<std::vector<alphabet::Symbol>, std::set<std::vector<alphabet::Symbol>>> getRules() const;
+ const std::map<std::tuple<std::vector<alphabet::Symbol>, alphabet::Symbol, std::vector<alphabet::Symbol>>, std::set<std::vector<alphabet::Symbol>>> getRules() const;
- bool removeRule(const std::vector<alphabet::Symbol>& leftHandSide, const std::vector<alphabet::Symbol>& rightHandSide);
+ bool removeRule(const std::vector<alphabet::Symbol>& lContext, const alphabet::Symbol& leftHandSide, const std::vector<alphabet::Symbol>& rContext, const std::vector<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 91397cd168..20cb3ed63b 100644
--- a/alib2data/test-src/grammar/GrammarTest.cpp
+++ b/alib2data/test-src/grammar/GrammarTest.cpp
@@ -86,6 +86,35 @@ void GrammarTest::testUnrestrictedParser() {
CPPUNIT_ASSERT( grammar == grammar2 );
}
+ {
+ grammar::ContextPreservingUnrestrictedGrammar 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(std::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2))))}, 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))))}, std::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("a")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2))))});
+ grammar.addRule(std::vector<alphabet::Symbol> {}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))), std::vector<alphabet::Symbol> {}, std::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::StringLabel("b")))), alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3))))});
+ grammar.addRule(std::vector<alphabet::Symbol> {}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), std::vector<alphabet::Symbol> {}, std::vector<alphabet::Symbol> {alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(3))))});
+ grammar.addRule(std::vector<alphabet::Symbol> {}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(1)))), std::vector<alphabet::Symbol> {}, 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))))});
+ grammar.addRule(std::vector<alphabet::Symbol> {}, alphabet::Symbol(alphabet::LabeledSymbol(label::Label(label::IntegerLabel(2)))), std::vector<alphabet::Symbol> {}, std::vector<alphabet::Symbol> {});
+ 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::ContextPreservingUnrestrictedGrammar grammar2 = parser.parseContextPreservingUnrestrictedGrammar(tokens2);
+
+ CPPUNIT_ASSERT( grammar == grammar2 );
+ }
+ }
}
void GrammarTest::testRegularParser() {
--
GitLab