From 3f21513ac5a245b284c9b7d0c5697e03949b833b Mon Sep 17 00:00:00 2001
From: Robin Jadoul <robin.jadoul@gmail.com>
Date: Sun, 24 Apr 2016 15:42:29 +0200
Subject: [PATCH] First work
---
include/Lexesis/automata.h | 46 ++++++++++++++++++
include/Lexesis/re.h | 90 +++++++++++++++++++++++++++++++++++
src/automata.cpp | 97 ++++++++++++++++++++++++++++++++++++++
3 files changed, 233 insertions(+)
create mode 100644 include/Lexesis/automata.h
create mode 100644 include/Lexesis/re.h
create mode 100644 src/automata.cpp
diff --git a/include/Lexesis/automata.h b/include/Lexesis/automata.h
new file mode 100644
index 0000000..de8bd5d
--- /dev/null
+++ b/include/Lexesis/automata.h
@@ -0,0 +1,46 @@
+#pragma once
+#ifndef AUTOMATA_H
+#define AUTOMATA_H
+
+#include <map>
+#include <set>
+#include <string>
+#include <utility>
+#include <vector>
+
+namespace lxs {
+ typedef unsigned long long State;
+ typedef unsigned long long Priority;
+
+ struct Automaton {
+ State numStates;
+ std::set<State> accepting;
+ std::map<State, Priority> priority;
+ std::map<State, std::string> acceptingToken;
+ State starting;
+ };
+
+ struct DFA : public Automaton {
+ std::map<State, std::map<char, State> > delta;
+ };
+
+ struct NFA : public Automaton {
+ std::map<State, std::map<char, std::set<State> > > delta;
+ virtual std::set<State> eClose(State);
+ };
+
+ struct ENFA : public NFA {
+ std::map<State, std::set<State> > epsilonTransitions;
+ virtual std::set<State> eClose(State);
+ };
+
+ std::string toDot(const DFA& d);
+ std::string toDot(const NFA& n);
+ std::string toDot(const ENFA& e);
+
+ ENFA merge(const std::vector<ENFA>& enfas);
+ DFA mssc(const NFA& e);
+ DFA minimize(const DFA& d);
+} //namespace lxs
+
+#endif //AUTOMATA_H
diff --git a/include/Lexesis/re.h b/include/Lexesis/re.h
new file mode 100644
index 0000000..9671a30
--- /dev/null
+++ b/include/Lexesis/re.h
@@ -0,0 +1,90 @@
+#pragma once
+#ifndef RE_H
+#define RE_H
+
+#include "Lexesis/automata.h"
+
+#include <stdexcept>
+
+namespace lxs {
+ class RE
+ {
+ public:
+ virtual ~RE() {}
+ virtual int toENFA(ENFA& enfa, int attach) = 0;
+ virtual std::string toRe() = 0;
+ };
+
+ class EmptyRE : public RE
+ {
+ public:
+ EmptyRE() {}
+ ~EmptyRE() {}
+ virtual int toENFA(ENFA& enfa, int attach);
+ virtual std::string toRe();
+ };
+
+ class EpsilonRE : public RE
+ {
+ public:
+ EpsilonRE() {}
+ ~EpsilonRE() {}
+ virtual int toENFA(ENFA& enfa, int attach);
+ virtual std::string toRe();
+ };
+
+ class SingleRE : public RE
+ {
+ public:
+ SingleRE(char c) : c(c) {}
+ ~SingleRE() {}
+ virtual int toENFA(ENFA& enfa, int attach);
+ virtual std::string toRe();
+
+ char c;
+ };
+
+ class ConcatRE : public RE
+ {
+ public:
+ ConcatRE(RE* e, RE* f) : e(e), f(f) {}
+ ~ConcatRE() {delete e; delete f;}
+ virtual int toENFA(ENFA& enfa, int attach);
+ virtual std::string toRe();
+
+ RE* e, *f;
+ };
+
+ class StarRE : public RE
+ {
+ public:
+ StarRE(RE* e) : e(e) {}
+ ~StarRE() {delete e;}
+ virtual int toENFA(ENFA& enfa, int attach);
+ virtual std::string toRe();
+
+ RE* e;
+ };
+
+ class PlusRE : public RE
+ {
+ public:
+ PlusRE(RE* e, RE* f) : e(e), f(f) {}
+ ~PlusRE() {delete e; delete f;}
+ virtual int toENFA(ENFA& enfa, int attach);
+ virtual std::string toRe();
+
+ RE* e, *f;
+ };
+
+ RE* parseRE(std::string& input);
+
+ class SyntaxError : public std::runtime_error
+ {
+ public:
+ SyntaxError(const char* w) : std::runtime_error(w) {}
+ };
+
+} //namespace lxs
+
+#endif //RE_H
diff --git a/src/automata.cpp b/src/automata.cpp
new file mode 100644
index 0000000..e79d684
--- /dev/null
+++ b/src/automata.cpp
@@ -0,0 +1,97 @@
+#include "automata.h"
+
+#include <cctype>
+#include <algorithm>
+#include <stdexcept>
+#include <string>
+
+namespace lxs {
+ std::string toDot(const DFA& d)
+ {
+ std::string s = "digraph {\nrankdir=LR\nin [shape=point style=invis]";
+
+ for (State state = 0; state < d.numStates; state++)
+ {
+ s += std::to_string(state) + " [label=\"" + std::to_string(state) + "\"";
+ if (state == d.starting)
+ s += " color=yellow";
+ if (d.accepting.count(state) > 0)
+ s += " color=green shape=doublecircle";
+ s += "]\n";
+ }
+
+ for (const auto& tmp : d.delta)
+ {
+ const auto& from = tmp.first;
+ for (const auto& trans : tmp.second)
+ {
+ s += std::to_string(from) + " -> " + std::to_string(trans.second) + " [label=\"" + trans.first + "\"]\n";
+ }
+ }
+
+ s += "in -> " + std::to_string(d.starting) + "\n}\n";
+ return s;
+ }
+
+ std::string toDot(const NFA& n)
+ {
+ std::string s = "digraph {\nrankdir=LR\nin [shape=point style=invis]\n";
+
+ for (State state = 0; state < n.numStates; state++)
+ {
+ s += std::to_string(state) + " [label=\"" + std::to_string(state) + "\"";
+ if (state == n.starting)
+ s += " color=yellow";
+ if (n.accepting.count(state) > 0)
+ s += " color=green shape=doublecircle";
+ s += "]\n";
+ }
+
+ for (const auto& tmp : n.delta)
+ {
+ const auto& from = tmp.first;
+ for (const auto& trans : tmp.second)
+ {
+ for (const auto& to : trans.second)
+ s += std::to_string(from) + " -> " + std::to_string(to) + " [label=\"" + trans.first + "\"]\n";
+ }
+ }
+
+ s += "in -> " + std::to_string(n.starting) + "\n}\n";
+ return s;
+ }
+
+ std::string toDot(const ENFA& e)
+ {
+ std::string s = "digraph {\nrankdir=LR\nin [shape=point style=invis]\n";
+
+ for (State state = 0; state < e.numStates; state++)
+ {
+ s += std::to_string(state) + " [label=\"" + std::to_string(state) + "\"";
+ if (state == e.starting)
+ s += " color=yellow";
+ if (e.accepting.count(state) > 0)
+ s += " color=green shape=doublecircle";
+ s += "]\n";
+ }
+
+ for (const auto& tmp : e.delta)
+ {
+ const auto& from = tmp.first;
+ for (const auto& trans : tmp.second)
+ {
+ for (const auto& to : trans.second)
+ s += std::to_string(from) + " -> " + std::to_string(to) + " [label=\"" + trans.first + "\"]\n";
+ }
+ }
+
+ for (const auto& etrans : e.epsilonTransitions)
+ {
+ for (const auto& dest : etrans.second)
+ s += std::to_string(etrans.first) + " -> " + std::to_string(dest) + " [label=\"ε\"]\n";
+ }
+
+ s += "in -> " + std::to_string(e.starting) + "\n}\n";
+ return s;
+ }
+} //namespace lxs