lex.l

基于C++的编译器词法分析模块生成器[Lex]

%{

#include "stdafx.h"
#include "lex.h"

ifstream fin;
ofstream fout;
string curstr;
dfa_state_map dfa;
state_action_map action;

int lineno = 1;
int state = 1;

void WriteToken(int token);
void ProcessNum();
void ProcessId();
void ReportParseError();

%}

alpha			[A-Za-z]
digit			[0-9]
alphanum		[A-Za-z0-9]
delim           [\t-\t\s-\s]
rline           [\r-\r]
nline           [\n-\n]

%%

{rline} {lineno++;}
{nline} {WriteToken(NEWLINE);}
{delim}{delim}* {;}
{alpha}{alphanum}* {WriteToken(IDENTIFIER);ProcessId();}
{digit}{digit}* {WriteToken(NUMBER);ProcessNum();}
"int" {WriteToken(INT);}
"if" {WriteToken(IF);}
"else" {WriteToken(ELSE);}
"return" {WriteToken(RETURN);}
"float" {WriteToken(FLOAT);}
"struct" {WriteToken(STRUCT);}
"(" {WriteToken(LPAR);}
")" {WriteToken(RPAR);}
"{" {WriteToken(LBRACE);}
"}" {WriteToken(RBRACE);}
"[" {WriteToken(LBRACK);}
"]" {WriteToken(RBRACK);}
"==" {WriteToken(EQUAL);}
"=" {WriteToken(ASSIGN);}
";" {WriteToken(SEMICOLON);}
"," {WriteToken(COMMA);}
"." {WriteToken(DOT);}
"+" {WriteToken(PLUS);}
"-" {WriteToken(MINUS);}
"*" {WriteToken(TIMES);}
"/" {WriteToken(DIVIDE);}

%%

void WriteToken(int token){
	fout<<unsigned char(token);
}

void ProcessNum(){
	fout<<unsigned char(0x20)<<curstr<<unsigned char(0x20);
}

void ProcessId(){
	fout<<unsigned char(0x20)<<curstr<<unsigned char(0x20);
}

void ReportParseError(){
	cerr<<"\nParse Error!\n";
	cerr<<"Line "<<lineno<<": "<<curstr<<"_<--";
	cerr<<"\n\n";
	fout<<"\nParse Error!\n";
	fout<<"Line "<<lineno<<": "<<curstr<<"_<--";
	fout<<"\n\n";

	fin.close();
	fout.close();

	exit(0);
}

void InitDFA(){
	ifstream dfain;
	dfain.open("dfa.bin", ios::binary);
	if (!dfain){
		cerr<<"\nError: DFA code file not found.\n";
		exit(0);
	}

	cout<<"\nloading DFA...\n";
	unsigned long start = GetTickCount();
	while(int(dfain.get())!=0x00){
		dfain.unget();
		int from = int(dfain.get());
		char transition = char(dfain.get());
		int to = int(dfain.get());
		dfa[from][transition] = to;
	}
	int state_count = int(dfain.get());
	for (int i = 0; i < state_count; i++){
		if (int(dfain.get())==0x00){
			action[i+1] = "";
			dfain.get();
		}
		else{
			getline(dfain, action[i+1]);
		}
	}
	cout<<"Loading complete, "<<GetTickCount()-start<<" ms.\n";

	dfain.close();
}

int _tmain(int argc, _TCHAR* argv[])
{
	InitDFA();

	switch(argc){
		case 1:
			{
				fin.open("test.txt", ios::binary);
				fout.open("tokenlist.bin", ios::binary);
			}
			break;
		case 2:
			{
				;
			}
			break;
		case 3:
			{
				fin.open(argv[1], ios::binary);
				fout.open(argv[2], ios::out);
			}
			break;
		default:
			cerr<<"\Error: illegal parameters.\n";
			break;
	}
	if (!fin){
		cerr<<"\nError: can't open source file.\n";
		exit(0);
	}
	if (!fout){
		cerr<<"\nError: can't open output file.\n";
		exit(0);
	}

	char c;
	cout<<"\n------------------------------------------\n";
	cout<<lineno<<' ';
	while(!fin.eof()){
		c = fin.get();
		if (dfa[state][c]!=0){
			state = dfa[state][c];
			curstr += c;
		}
		else{
			ProcessEndState(state);
			if (dfa[1][c]==0){
				if (fin.eof()) break;
				ReportParseError();
			}
			else{
				state = dfa[1][c];
				curstr.clear();
				curstr += c;
			}
		}
		cout<<c;
		if (c==0x0a){
			cout<<lineno<<' ';
		}
	}

	cout<<"\n------------------------------------------\n";
	cout<<"\nParse complete.\n";

	return 0;
}