From b43525d4ad9855e8d4c7561648b73ea71ecb94b4 Mon Sep 17 00:00:00 2001
From: Martin Zak <zakmart1@fit.cvut.cz>
Date: Sat, 23 Nov 2013 19:48:12 +0100
Subject: [PATCH] RegExp implementation
---
acat/src/acat.cpp | 6 ++
alib/src/regexp/Alternation.cpp | 24 +++++++
alib/src/regexp/Alternation.h | 29 ++++++++
alib/src/regexp/Iteration.cpp | 22 +++++++
alib/src/regexp/Iteration.h | 27 ++++++++
alib/src/regexp/RegExp.cpp | 34 ++++++++++
alib/src/regexp/RegExp.h | 34 ++++++++++
alib/src/regexp/RegExpElement.cpp | 15 +++++
alib/src/regexp/RegExpElement.h | 22 +++++++
alib/src/regexp/RegExpParser.cpp | 106 ++++++++++++++++++++++++++++++
alib/src/regexp/RegExpParser.h | 34 ++++++++++
alib/src/regexp/RegExpPrinter.cpp | 71 ++++++++++++++++++++
alib/src/regexp/RegExpPrinter.h | 34 ++++++++++
alib/src/regexp/RegExpSymbol.cpp | 20 ++++++
alib/src/regexp/RegExpSymbol.h | 28 ++++++++
examples/regexp/regexp.xml | 17 +++++
examples/regexp/regexp2.xml | 27 ++++++++
examples/regexp/regexp3.xml | 27 ++++++++
18 files changed, 577 insertions(+)
create mode 100644 alib/src/regexp/Alternation.cpp
create mode 100644 alib/src/regexp/Alternation.h
create mode 100644 alib/src/regexp/Iteration.cpp
create mode 100644 alib/src/regexp/Iteration.h
create mode 100644 alib/src/regexp/RegExp.cpp
create mode 100644 alib/src/regexp/RegExp.h
create mode 100644 alib/src/regexp/RegExpElement.cpp
create mode 100644 alib/src/regexp/RegExpElement.h
create mode 100644 alib/src/regexp/RegExpParser.cpp
create mode 100644 alib/src/regexp/RegExpParser.h
create mode 100644 alib/src/regexp/RegExpPrinter.cpp
create mode 100644 alib/src/regexp/RegExpPrinter.h
create mode 100644 alib/src/regexp/RegExpSymbol.cpp
create mode 100644 alib/src/regexp/RegExpSymbol.h
create mode 100644 examples/regexp/regexp.xml
create mode 100644 examples/regexp/regexp2.xml
create mode 100644 examples/regexp/regexp3.xml
diff --git a/acat/src/acat.cpp b/acat/src/acat.cpp
index 5f7ab254b8..cafee32369 100644
--- a/acat/src/acat.cpp
+++ b/acat/src/acat.cpp
@@ -11,6 +11,8 @@
#include "GrammarFactory.h"
#include "AlibException.h"
+#include "regexp/RegExpParser.h"
+
#include "automaton/AutomatonParser.h"
#include "grammar/GrammarParser.h"
#include "grammar/RightRegularGrammar.h"
@@ -23,6 +25,7 @@
using namespace std;
using namespace automaton;
+using namespace regexp;
using namespace alib;
using namespace sax;
using namespace grammar;
@@ -49,6 +52,9 @@ int main(int argc, char** argv) {
} else if (tokens.front().getData() == "grammar") {
UnknownGrammar grammar = GrammarParser::parse(tokens);
grammar.toXML(cout);
+ } else if (tokens.front().getData() == "regexp") {
+ RegExp regexp = RegExpParser::parse(tokens);
+ regexp.toXML(cout);
} else {
throw AlibException(
"Expected root tag automaton or grammar. Read: "
diff --git a/alib/src/regexp/Alternation.cpp b/alib/src/regexp/Alternation.cpp
new file mode 100644
index 0000000000..23e98e68b9
--- /dev/null
+++ b/alib/src/regexp/Alternation.cpp
@@ -0,0 +1,24 @@
+/*
+ * Alternation.cpp
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#include "Alternation.h"
+
+namespace regexp {
+
+Alternation::~Alternation() {
+}
+
+list<RegExpElement*>& Alternation::getFirst() {
+ return first;
+}
+
+list<RegExpElement*>& Alternation::getSecond() {
+ return second;
+}
+
+} /* namespace regexp */
+
diff --git a/alib/src/regexp/Alternation.h b/alib/src/regexp/Alternation.h
new file mode 100644
index 0000000000..6e1cdf5ed4
--- /dev/null
+++ b/alib/src/regexp/Alternation.h
@@ -0,0 +1,29 @@
+/*
+ * Alternation.h
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#ifndef ALTERNATION_H_
+#define ALTERNATION_H_
+
+#include <list>
+#include "RegExpElement.h"
+
+namespace regexp {
+
+using namespace std;
+
+class Alternation: public RegExpElement {
+private:
+ list<RegExpElement*> first;
+ list<RegExpElement*> second;
+public:
+ ~Alternation();
+ list<RegExpElement*>& getFirst();
+ list<RegExpElement*>& getSecond();
+};
+
+} /* namespace regexp */
+#endif /* ALTERNATION_H_ */
diff --git a/alib/src/regexp/Iteration.cpp b/alib/src/regexp/Iteration.cpp
new file mode 100644
index 0000000000..6e95c17edf
--- /dev/null
+++ b/alib/src/regexp/Iteration.cpp
@@ -0,0 +1,22 @@
+/*
+ * Iteration.cpp
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#include "Iteration.h"
+
+namespace regexp {
+regexp::Iteration::~Iteration() {
+ for (auto element : elements) {
+ delete element;
+ }
+}
+
+list<RegExpElement*>& regexp::Iteration::getElements() {
+ return elements;
+}
+
+} /* namespace regexp */
+
diff --git a/alib/src/regexp/Iteration.h b/alib/src/regexp/Iteration.h
new file mode 100644
index 0000000000..0bec4d49f5
--- /dev/null
+++ b/alib/src/regexp/Iteration.h
@@ -0,0 +1,27 @@
+/*
+ * Iteration.h
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#ifndef ITERATION_H_
+#define ITERATION_H_
+
+#include <list>
+#include "RegExpElement.h"
+
+namespace regexp {
+
+using namespace std;
+
+class Iteration : public RegExpElement {
+private:
+ list<RegExpElement*> elements;
+public:
+ ~Iteration();
+ list<RegExpElement*>& getElements();
+};
+
+} /* namespace regexp */
+#endif /* ITERATION_H_ */
diff --git a/alib/src/regexp/RegExp.cpp b/alib/src/regexp/RegExp.cpp
new file mode 100644
index 0000000000..0151ffd4f3
--- /dev/null
+++ b/alib/src/regexp/RegExp.cpp
@@ -0,0 +1,34 @@
+/*
+ * RegExp.cpp
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#include "RegExp.h"
+#include "RegExpPrinter.h"
+
+namespace regexp {
+
+RegExp::RegExp() {
+}
+
+RegExp::~RegExp() {
+ for(auto element : regexp) {
+ delete element;
+ }
+}
+
+RegExp::RegExp(const list<RegExpElement*>& regexp) :
+ regexp(regexp) {
+}
+
+list<RegExpElement*>& RegExp::getRegExp() {
+ return regexp;
+}
+
+void RegExp::toXML(ostream& out) {
+ RegExpPrinter::toXML(*this, out);
+}
+
+} /* namespace regexp */
diff --git a/alib/src/regexp/RegExp.h b/alib/src/regexp/RegExp.h
new file mode 100644
index 0000000000..35a2f77235
--- /dev/null
+++ b/alib/src/regexp/RegExp.h
@@ -0,0 +1,34 @@
+/*
+ * RegExp.h
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#ifndef REGEXP_H_
+#define REGEXP_H_
+
+#include <vector>
+#include <list>
+#include <string>
+#include "RegExpElement.h"
+
+namespace regexp {
+
+using namespace std;
+
+class RegExp {
+private:
+ list<RegExpElement*> regexp;
+
+public:
+ RegExp();
+ ~RegExp();
+ RegExp(const list<RegExpElement*>& regexp) ;
+ list<RegExpElement*>& getRegExp();
+
+ void toXML(ostream& out);
+};
+
+} /* namespace regexp */
+#endif /* REGEXP_H_ */
diff --git a/alib/src/regexp/RegExpElement.cpp b/alib/src/regexp/RegExpElement.cpp
new file mode 100644
index 0000000000..eacf72d353
--- /dev/null
+++ b/alib/src/regexp/RegExpElement.cpp
@@ -0,0 +1,15 @@
+/*
+ * RegExpElement.cpp
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#include "RegExpElement.h"
+
+namespace regexp {
+
+} /* namespace regexp */
+
+regexp::RegExpElement::~RegExpElement() {
+}
diff --git a/alib/src/regexp/RegExpElement.h b/alib/src/regexp/RegExpElement.h
new file mode 100644
index 0000000000..b2ed1df406
--- /dev/null
+++ b/alib/src/regexp/RegExpElement.h
@@ -0,0 +1,22 @@
+/*
+ * RegExpElement.h
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#ifndef REGEXPELEMENT_H_
+#define REGEXPELEMENT_H_
+
+namespace regexp {
+
+using namespace std;
+
+class RegExpElement {
+public:
+ virtual ~RegExpElement();
+
+};
+
+} /* namespace regexp */
+#endif /* REGEXPELEMENT_H_ */
diff --git a/alib/src/regexp/RegExpParser.cpp b/alib/src/regexp/RegExpParser.cpp
new file mode 100644
index 0000000000..01e380cff2
--- /dev/null
+++ b/alib/src/regexp/RegExpParser.cpp
@@ -0,0 +1,106 @@
+/*
+ * RegExpParser.cpp
+ *
+ * Created on: Nov 23, 2013
+ * Author: Martin Zak
+ */
+
+#include "RegExpParser.h"
+#include "../sax/ParserException.h"
+
+namespace regexp {
+
+RegExp RegExpParser::parse(list<Token>& input) {
+ list<RegExpElement*> elements;
+
+ try {
+ popToken(input, Token::START_ELEMENT, "regexp");
+
+ parseContent(input, elements);
+
+ popToken(input, Token::END_ELEMENT, "regexp");
+ return RegExp(elements);
+ } catch (ParserException& e) {
+ for (auto element : elements) {
+ delete element;
+ }
+ throw e;
+ }
+}
+
+Iteration* RegExpParser::parseIteration(list<Token>& input) {
+ popToken(input, Token::START_ELEMENT, "iteration");
+
+ Iteration* iteration = new Iteration();
+ parseContent(input,iteration->getElements());
+
+ popToken(input, Token::END_ELEMENT, "iteration");
+ return iteration;
+}
+
+Alternation* RegExpParser::parseAlternation(list<Token>& input) {
+ popToken(input, Token::START_ELEMENT, "alternation");
+
+ Alternation* alternation = new Alternation;
+ popToken(input, Token::START_ELEMENT, "first");
+ parseContent(input, alternation->getFirst());
+ popToken(input, Token::END_ELEMENT, "first");
+
+ popToken(input, Token::START_ELEMENT, "second");
+ parseContent(input, alternation->getSecond());
+ popToken(input, Token::END_ELEMENT, "second");
+
+ popToken(input, Token::END_ELEMENT, "alternation");
+ return alternation;
+}
+
+void RegExpParser::parseContent(list<Token>& input, list<RegExpElement*>& elements) {
+ while (true) {
+ if (isToken(input, Token::START_ELEMENT, "symbol")) {
+ elements.push_back(parseSymbol(input));
+ } else if (isToken(input, Token::START_ELEMENT, "iteration")) {
+ elements.push_back(parseIteration(input));
+ } else if (isToken(input, Token::START_ELEMENT, "alternation")) {
+ elements.push_back(parseAlternation(input));
+ } else {
+ return;
+ }
+ }
+
+}
+
+RegExpSymbol* RegExpParser::parseSymbol(list<Token>& input, string tagName) {
+ popToken(input, Token::START_ELEMENT, tagName);
+
+ if (input.front().getType() == Token::CHARACTER) {
+ RegExpSymbol* symbol = new RegExpSymbol(input.front().getData());
+ input.pop_front();
+ popToken(input, Token::END_ELEMENT, tagName);
+ return symbol;
+ } else if (isToken(input, Token::END_ELEMENT, tagName)) {
+ input.pop_front();
+ return new RegExpSymbol("");
+ } else if (isToken(input, Token::START_ELEMENT, "eps")) {
+ input.pop_front();
+ popToken(input, Token::END_ELEMENT, "eps");
+ popToken(input, Token::END_ELEMENT, tagName);
+ return new RegExpSymbol("");
+ } else {
+ throw ParserException(Token("", Token::CHARACTER), input.front());
+ }
+}
+
+bool RegExpParser::isToken(list<Token>& input, Token::TokenType type, string data) {
+ return input.front().getType() == type && input.front().getData() == data;
+}
+
+void RegExpParser::popToken(list<Token>& input, Token::TokenType type, string data) {
+ if (isToken(input, type, data)) {
+ input.pop_front();
+ } else {
+ throw ParserException(Token(data, type), input.front());
+ }
+}
+
+} /* namespace regexp */
+
diff --git a/alib/src/regexp/RegExpParser.h b/alib/src/regexp/RegExpParser.h
new file mode 100644
index 0000000000..3181ca47e9
--- /dev/null
+++ b/alib/src/regexp/RegExpParser.h
@@ -0,0 +1,34 @@
+/*
+ * RegExpParser.h
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#ifndef REGEXPPARSER_H_
+#define REGEXPPARSER_H_
+
+#include "RegExp.h"
+#include "../sax/Token.h"
+#include "RegExpSymbol.h"
+#include "Iteration.h"
+#include "Alternation.h"
+
+namespace regexp {
+
+using namespace sax;
+
+class RegExpParser {
+public:
+ static RegExp parse(list<Token>& input);
+protected:
+ static void parseContent(list<Token>& input, list<RegExpElement*>& elements);
+ static RegExpSymbol* parseSymbol(list<Token> &input, string tagName="symbol");
+ static Iteration* parseIteration(list<Token> &input);
+ static Alternation* parseAlternation(list<Token> &input);
+ static bool isToken(list<Token> &input, Token::TokenType type, string data);
+ static void popToken(list<Token> &input, Token::TokenType type, string data);
+};
+
+} /* namespace regexp */
+#endif /* REGEXPPARSER_H_ */
diff --git a/alib/src/regexp/RegExpPrinter.cpp b/alib/src/regexp/RegExpPrinter.cpp
new file mode 100644
index 0000000000..962df86e67
--- /dev/null
+++ b/alib/src/regexp/RegExpPrinter.cpp
@@ -0,0 +1,71 @@
+/*
+ * RegExpPrinter.cpp
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#include "RegExpPrinter.h"
+
+namespace regexp {
+
+const string RegExpPrinter::INDENTATION = "\t";
+
+void RegExpPrinter::toXML(RegExp& regexp, ostream& out) {
+ out << "<regexp>\n";
+ printContent(regexp.getRegExp(), out, INDENTATION);
+ out << "</regexp>\n";
+}
+
+void RegExpPrinter::printContent(list<RegExpElement*>& content, ostream& out, string prefix) {
+ for (auto element : content) {
+ Alternation* alternation = dynamic_cast<Alternation*>(element);
+ if (alternation) {
+ printAlternation(alternation, out, prefix);
+ continue;
+ }
+
+ Iteration* iteration = dynamic_cast<Iteration*>(element);
+ if (iteration) {
+ printIteration(iteration, out, prefix);
+ continue;
+ }
+
+ RegExpSymbol* symbol = dynamic_cast<RegExpSymbol*>(element);
+ if (symbol) {
+ printSymbol(symbol, out, prefix);
+ continue;
+ }
+
+ }
+}
+
+void RegExpPrinter::printAlternation(Alternation* alternation, ostream& out, string prefix) {
+ string doubleIndentation = prefix + INDENTATION + INDENTATION;
+ out << prefix << "<alternation>\n";
+
+ out << prefix << INDENTATION << "<first>\n";
+ printContent(alternation->getFirst(), out, doubleIndentation);
+ out << prefix << INDENTATION << "</first>\n";
+
+ out << prefix << INDENTATION << "<second>\n";
+ printContent(alternation->getSecond(), out, doubleIndentation);
+ out << prefix << INDENTATION << "</second>\n";
+
+ out << prefix << "</alternation>\n";
+}
+
+void RegExpPrinter::printIteration(Iteration* iteration, ostream& out, string prefix) {
+ out << prefix << "<iteration>\n";
+ printContent(iteration->getElements(), out, prefix + INDENTATION);
+ out << prefix << "</iteration>\n";
+}
+
+void RegExpPrinter::printSymbol(RegExpSymbol* symbol, ostream& out, string prefix) {
+ out << prefix << "<symbol>";
+ out << symbol->getSymbol();
+ out << "</symbol>\n";
+}
+
+} /* namespace regexp */
+
diff --git a/alib/src/regexp/RegExpPrinter.h b/alib/src/regexp/RegExpPrinter.h
new file mode 100644
index 0000000000..e1e3998175
--- /dev/null
+++ b/alib/src/regexp/RegExpPrinter.h
@@ -0,0 +1,34 @@
+/*
+ * RegExpPrinter.h
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#ifndef REGEXPPRINTER_H_
+#define REGEXPPRINTER_H_
+
+#include <ostream>
+#include "RegExp.h"
+#include "Alternation.h"
+#include "Iteration.h"
+#include "RegExpSymbol.h"
+
+namespace regexp {
+
+using namespace std;
+
+class RegExpPrinter {
+protected:
+ static const string INDENTATION;
+ static void printContent(list<RegExpElement*>& content, ostream& out, string prefix);
+ static void printAlternation(Alternation* alternation, ostream& out, string prefix);
+ static void printIteration(Iteration* iteration, ostream& out, string prefix);
+ static void printSymbol(RegExpSymbol* symbol, ostream& out, string prefix);
+
+public:
+ static void toXML(RegExp& regexp, ostream& out);
+};
+
+} /* namespace regexp */
+#endif /* REGEXPPRINTER_H_ */
diff --git a/alib/src/regexp/RegExpSymbol.cpp b/alib/src/regexp/RegExpSymbol.cpp
new file mode 100644
index 0000000000..85c6f159dd
--- /dev/null
+++ b/alib/src/regexp/RegExpSymbol.cpp
@@ -0,0 +1,20 @@
+/*
+ * RegExpSymbol.cpp
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#include "RegExpSymbol.h"
+
+namespace regexp {
+
+} /* namespace regexp */
+
+regexp::RegExpSymbol::RegExpSymbol() :
+ Symbol("") {
+}
+
+regexp::RegExpSymbol::RegExpSymbol(const string& symbol) :
+ Symbol(symbol) {
+}
diff --git a/alib/src/regexp/RegExpSymbol.h b/alib/src/regexp/RegExpSymbol.h
new file mode 100644
index 0000000000..c2bac3c251
--- /dev/null
+++ b/alib/src/regexp/RegExpSymbol.h
@@ -0,0 +1,28 @@
+/*
+ * RegExpSymbol.h
+ *
+ * Created on: Nov 23, 2013
+ * Author: martin
+ */
+
+#ifndef REGEXPSYMBOL_H_
+#define REGEXPSYMBOL_H_
+
+#include <string>
+#include "RegExpElement.h"
+#include "../alphabet/Symbol.h"
+
+namespace regexp {
+
+using namespace std;
+using namespace alphabet;
+
+class RegExpSymbol: public RegExpElement, public Symbol {
+public:
+ RegExpSymbol();
+ RegExpSymbol(const string& symbol);
+
+};
+
+} /* namespace regexp */
+#endif /* REGEXPSYMBOL_H_ */
diff --git a/examples/regexp/regexp.xml b/examples/regexp/regexp.xml
new file mode 100644
index 0000000000..f1d847322a
--- /dev/null
+++ b/examples/regexp/regexp.xml
@@ -0,0 +1,17 @@
+<regexp>
+ <symbol>0</symbol>
+ <symbol>1</symbol>
+ <iteration>
+ <symbol>0</symbol>
+ <symbol>0</symbol>
+ </iteration>
+
+ <alternation>
+ <first>
+ <symbol>11</symbol>
+ </first>
+ <second>
+ <symbol>10</symbol>
+ </second>
+ </alternation>
+</regexp>
diff --git a/examples/regexp/regexp2.xml b/examples/regexp/regexp2.xml
new file mode 100644
index 0000000000..4da5d19420
--- /dev/null
+++ b/examples/regexp/regexp2.xml
@@ -0,0 +1,27 @@
+<regexp>
+ <symbol>0</symbol>
+ <symbol>1</symbol>
+ <iteration>
+ <alternation>
+ <first>
+ <symbol>0</symbol>
+ </first>
+ <second>
+ <symbol>1</symbol>
+ </second>
+ </alternation>
+ </iteration>
+
+ <alternation>
+ <first>
+ <iteration>
+ <symbol>11</symbol>
+ </iteration>
+ </first>
+ <second>
+ <iteration>
+ <symbol>10</symbol>
+ </iteration>
+ </second>
+ </alternation>
+</regexp>
diff --git a/examples/regexp/regexp3.xml b/examples/regexp/regexp3.xml
new file mode 100644
index 0000000000..7359cb4989
--- /dev/null
+++ b/examples/regexp/regexp3.xml
@@ -0,0 +1,27 @@
+<regexp>
+ <iteration>
+ <iteration>
+ <iteration>
+ <iteration>
+ <iteration>
+ <iteration>
+ <iteration>
+ <iteration>
+ <iteration>
+ <iteration>
+ <symbol>We</symbol>
+ <symbol>have</symbol>
+ <symbol>to</symbol>
+ <symbol>go</symbol>
+ <symbol>deeper</symbol>
+ </iteration>
+ </iteration>
+ </iteration>
+ </iteration>
+ </iteration>
+ </iteration>
+ </iteration>
+ </iteration>
+ </iteration>
+ </iteration>
+</regexp>
--
GitLab