词法分析ver1.0

int main(){
int a/*hehe*/=100;//hehe
int b=100;
int c;
/*
xixi
*/
c = a+b;
char[] str1 = "miao";
char[] str2 = "mi
ao";
printf("%d",c);
scanf("%d",&c);
return 0;
}

#include <iostream>
#include <fstream>
#include <string>
using namespace std;
#define _RESERVE_NUM 38
#define _DELIMILTER_NUM 26

string key[_RESERVE_NUM] = {//关键字表
	"void", "int", "char", "float", "double", "while",
	"do", "break", "case", "const", "continue", "default",
	"else", "enum", "extern", "for", "if", "long", "return",
	"short", "signed", "sizeof", "static", "struct", "switch",
	"typedef", "union", "unsigned", "volatile", "redister", "auto", "goto"
	"short", "long", "main", "return", "sizeof", "printf", "scanf"
};

string symbol[_DELIMILTER_NUM] = {//符号表
	"=", "+", "-", "*", "/", "++", "--", "+=", "-=",
	"*=", "/=", "==", "!=", ">", "<", ">=", "<=", "(",
	")", "[", "]", "{", "}", ",", ":", ";"
};

char ch;//存放刚刚读取的一个数据
string strToken;       //存放构成单词的字符串
string strline;        //存放读取的一行
int pText = 0;        //搜索指示器
int line = 1;

int FindSymbol(string str){//寻找符号在符号表中的种别码
	for (int i = 0; i < _DELIMILTER_NUM; i++){
		if (str == symbol[i])return i + 37;
	}
	return -1;
}

int Reserve()   //寻找字符串在关键字表中的种别码，如果没有则是标示符
{
	for (int i = 0; i < 37; i++)
	{
		if (strToken == key[i])  return i;
	}
	return -1;
}

bool IsLetter(){//检查是否是字母
	if (((ch >= 'a') && (ch <= 'z')) || ((ch >= 'A') && (ch <= 'Z')))return true;
	return false;
}

bool IsDigit(){//检查是否是数字
	if ((ch >= '0') && (ch <= '9'))return true;
	return false;
}

int FloatOrReal(string str){//检查是浮点还是实数
	if (str.find("."))return 36;
	return 35;
}



int main(){
	ifstream in;
	in.open("input.txt");
	if (!in){
		cout << "打开文件失效。" << endl;
		return 1;
	}
	
	int sta = 0;//初始状态为0
	while (!in.eof()){
		getline(in,strline);//读取一行
		cout << strline << endl;
		pText = 0;	
		while (1){//在行内分析
			sta = 0;
			strToken = "";
			ch = strline[pText];
			//cout << ch << endl;
			if ((ch == ' ') || (ch == '\t')){//跳过空格和进格
				sta = 0;
			}
			else if (IsLetter() || ch == '_'){//判断字符串
				sta = 1;
				while (IsLetter() || ch == '_' || IsDigit()){
					strToken += ch;
					pText++;
					ch = strline[pText];
				}
				sta = 2;
				pText--;
				int code = Reserve();
				if (code){
					cout << "(" << 26+code << "," << strToken << ",line=" << line << ")" << endl;
				}
			}
			else if (IsDigit()){//判断数字
				sta = 3;
				while (IsDigit()||ch == '.'){
					strToken += ch;
					pText++;
					ch = strline[pText];
				}
				sta = 4;
				pText--;
				int code = FloatOrReal(strToken);
				cout << "(" << code << "," << strToken << ",line=" << line << ")" << endl;
			}
			else if (ch == ';'){//判断界符
				sta = 5;
				cout << "(" << 34 << ",;" << ",line=" << line << ")" << endl;
				
			}
			else if (ch == '(') {
				sta = 5;
				cout << "(" << 26 << ",(" << ",line=" << line << ")" << endl;
				
			}
			else if (ch == ')'){
				sta = 5;
				cout << "(" << 27 << ",)" << ",line=" << line << ")" << endl;
				
			}
			else if (ch == '['){
				sta = 5;
				cout << "(" << 28 << ",[" << ",line=" << line << ")" << endl;
			}
			else if (ch == ']'){
				sta = 5;
				cout << "(" << 29 << ",]" << ",line=" << line << ")" << endl;
			}
			else if (ch == '{') {
				sta = 5;
				cout << "(" << 30 << ",{" << ",line=" << line << ")" << endl;
				
			}
			else if (ch == '}')  {
				sta = 5;
				cout << "(" << 31 << ",}" << ",line=" << line << ")" << endl;
				
			}
			else if (ch == ',')  {
				sta = 5;
				cout << "(" << 32 << ",," << ",line=" << line << ")" << endl;
				
			}
			else if ((ch == '+')){//判断++,+,+=
				sta = 7;
				strToken += ch;
				pText++;
				ch = strline[pText];
				if ((ch == '=')){
					sta = 8;
					strToken += ch;
				}
				else if (ch == '+'){
					sta = 9;
					strToken += ch;
				}
				else{
					sta = 7;
					pText--;
				}
				int code = FindSymbol(strToken);
				cout << "(" << code << "," << strToken << ",line=" << line << ")" << endl;
				
			}
			else if ((ch == '-')){//判断--,-,-=
				sta = 11;
				strToken += ch;
				pText++;
				ch = strline[pText];
				if ((ch == '=')){
					sta = 12;
					strToken += ch;
				}
				else if (ch == '-'){
					sta = 13;
					strToken += ch;
				}
				else{
					sta = 11;
					pText--;
				}
				int code = FindSymbol(strToken);
				cout << "(" << code << "," << strToken << ",line=" << line << ")" << endl;
				
			}
			else if ((ch == '=')){//判断=,==
				sta = 15;
				strToken += ch;
				pText++;
				ch = strline[pText];
				if (ch == '='){
					sta = 16;
					strToken += ch;
				}
				else{
					sta = 15;
					pText--;
				}
				int code = FindSymbol(strToken);
				cout << "(" << code << "," << strToken << ",line=" << line << ")" << endl;
			}
			else if (ch == '*'){//判断*，*=
				sta = 18;
				strToken += ch;
				pText++;
				ch = strline[pText];
				if (ch == '='){
					sta = 19;
					strToken += ch;
				}
				else{
					sta = 18;
					pText--;
				}
				int code = FindSymbol(strToken);
				cout << "(" << code << "," << strToken << ",line=" << line << ")" << endl;
			}
			else if (ch == '/'){//判断/,/=,//,/*
				sta = 21;
				strToken += ch;
				pText++;
				ch = strline[pText];
				if (ch == '='){
					sta = 22;
					strToken += ch;
				}
				else if (ch == '/'){//行注释
					sta = 23;
					break;
				}
				else if (ch == '*'){//块注释
					sta = 24;
					pText++;
					while (1)
					{
						if (strline[pText] == '\0'){
							getline(in, strline);
							cout << strline << endl;
							line++;
							pText = 0;
						}
						if ((strline[pText] == '*') && (strline[pText + 1] == '/')){
							break;
						}
						pText++;
					}
					pText+=2;
					if (strline[pText] == '\0'){
						break;
					}
					continue;//进行块注释后面的内容
				}
				else{
					sta = 21;
					pText--;
				}
				int code = FindSymbol(strToken);
				cout << "(" << code << "," << strToken << ",line=" << line << ")" << endl;
			}
			else if (ch == '"'){//判断字符常量
				sta = 24;
				pText++;
				while (strline[pText] != '"'){
					ch = strline[pText];
					strToken += ch;
					pText++;
					if (strline[pText] == '\0'){//字符常量换行处理
						getline(in, strline);
						cout << strline << endl;
						line++;
						pText = 0;
						while ((strline[pText] == ' ') || (strline[pText] == '\t')){
							pText++;
						}
					}
				}
				sta = 25;
				cout << "(-1" << "," << strToken << ",line=" << line << ")" << endl;
			}
			else if (ch == '&'){
				sta = 26;
				cout << "(65" << ",&,line=" << line << ")" << endl;
			}
			else if (ch == '\0'){//读完一行
				break;
			}
			pText++;
		}
		line++;	
	}
	in.close();
	return 0;
}