generate DFA states from NFA

README.md

@@ -15,11 +15,11 @@ Each of these directories has a `Makefile` that puts stuff in a `bin/` folder.
 - `madoka/`: one of my first passes, arguably my first success I could call a "compiler". Pre 2014-summer
 - `sayaka/` (in progress): successor to `madoka/`, had ideas on what to do differently. The ideas were pre 2014-summer, most of the work on it is post 2014-summer
 - `primed/`: hand written LL(1) regex parser and NFA state generator
-- `siyu/` (in progress): DFA state generator, lexer-generator, and parser-generator
+- `siyu/` (in progress): successor to `primed/`, DFA state generator, lexer-generator, and parser-generator
 
 ### Siyu
-- Successor to `primed/`; working based on the dragon book, trying to follow the "harder, more theoretical" stuff, needed a fresh start :P
-- generating DFA states directly from regex
+- generate DFA states from NFA states
+- `siyu-1/` is a start at generating DFA states directly from a regex based on the dragon book
 
 ### Primed
 - hand written, recursive descent basic regex parser (builds AST)

siyu-1/.gitignore

@@ -0,0 +1 @@
+bin/

siyu-1/Makefile

@@ -0,0 +1,37 @@
+CXX = g++
+LD = g++
+CXXFLAGS = -g -std=c++11 -c
+LDFLAGS = -g
+LDLIBS =
+
+SRC_DIR = src
+BIN_DIR = bin
+
+EXECUTABLE = $(BIN_DIR)/siyu
+
+SRCS = $(sort $(wildcard $(SRC_DIR)/*.cpp))
+OBJS = $(addsuffix .o,$(subst $(SRC_DIR),$(BIN_DIR),$(SRCS)))
+DEPS = $(addsuffix .d,$(subst $(SRC_DIR),$(BIN_DIR),$(SRCS)))
+
+.PHONY: all build clean
+
+all: build
+
+build: $(EXECUTABLE)
+
+clean:
+	rm -rf $(BIN_DIR)
+
+$(EXECUTABLE): $(OBJS)
+	$(LD) $(LDFLAGS) -o $@ $(OBJS) $(LDLIBS)
+
+$(BIN_DIR)/%.o: $(SRC_DIR)/%
+	$(CXX) $(CXXFLAGS) $< -o $@
+
+$(BIN_DIR):
+	mkdir -p $@
+
+$(BIN_DIR)/%.d: $(SRC_DIR)/% $(BIN_DIR)
+	$(CXX) $(CXXFLAGS) -MM $< -MT '$(@:.d=.o)' -MF $@
+
+include $(DEPS)

siyu-1/src/finite_automata.cpp

@@ -0,0 +1,23 @@
+#include "finite_automata.h"
+
+#include <cstdio>
+
+DFAState::DFAState() : terminal(false) {
+	return;
+}
+
+DFA::DFA() {
+	this->root = this->new_state();
+}
+
+DFAState* DFA::new_state() {
+	std::unique_ptr<DFAState> state(new DFAState());
+	DFAState* ptr = state.get();
+	this->id_from_state[ptr] = this->states.size();
+	this->states.push_back(std::move(state));
+	return ptr;
+}
+
+UInt DFA::id(DFAState* state) {
+	return this->id_from_state.at(state);
+}

siyu-1/src/finite_automata.h

@@ -0,0 +1,27 @@
+#ifndef SIYU_FINITE_AUTOMATA_H_INCLUDED
+#define SIYU_FINITE_AUTOMATA_H_INCLUDED
+
+#include <map>
+#include <vector>
+#include <memory>
+#include "global.h"
+
+class DFAState {
+public:
+	std::map<UInt, DFAState*> next_states;
+	bool terminal;
+	DFAState();
+};
+
+class DFA {
+	std::map<DFAState*, UInt> id_from_state;
+public:
+	DFAState* root;
+	std::vector<std::unique_ptr<DFAState>> states;
+
+	DFA();
+	DFAState* new_state();
+	UInt id(DFAState*);
+};
+
+#endif /* SIYU_FINITE_AUTOMATA_H_INCLUDED */

siyu-1/src/global.cpp

@@ -0,0 +1 @@
+#include "global.h"

siyu-1/src/global.h

@@ -0,0 +1,15 @@
+#ifndef SIYU_GLOBAL_H_INCLUDED
+#define SIYU_GLOBAL_H_INCLUDED
+
+#include <cstdint>
+
+typedef int8_t Byte;
+typedef uint8_t UByte;
+typedef int16_t Short;
+typedef uint16_t UShort;
+typedef int32_t Int;
+typedef uint32_t UInt;
+typedef int64_t Long;
+typedef uint64_t ULong;
+
+#endif /* SIYU_GLOBAL_H_INCLUDED */

siyu-1/src/main.cpp

@@ -0,0 +1,11 @@
+#include <iostream>
+#include "regex.h"
+
+int main() {
+	std::string str = "abc|def|ghi{3,5}";
+	RegexParser parser;
+	RegexAST* r = parser.parse(str);
+	RegexASTPrinter printer;
+	r->accept(&printer);
+	return 0;
+}