编译原理-词法分析器

#include <iostream>
#include <fstream>
#include <string> 
using namespace std;
string KEYWORD[13] ={"const","int","char","void","main","if","else","do","while","for","scanf","printf","return"};//string数组存储关键字
string keyword[13] = {"CONSTTK","INTTK","CHARTK","VOIDTK","MAINTK","IFTK","ELSETK","DOTK","WHILETK","FORTK","SCANFTK","PRINTFTK","RETURNTK"}; //string数组存储关键字内码值
string keyword2[13] = {"<常量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>","<变量定义>"};
char SEPARATER[8]={';','/,','(',')','[',']','{','}'};//分隔符
string separater[8]={"SEMICN","COMMA","LPARENT","RPARENT","LBRACK","RBRACK","LBRACE","RBRACE"};//分隔符内码值
char OPERATOR[7]={'+','-','*','/','>','<','='};//单目运算符
string OPERATOR2[4]={"<=",">=","==","!="};//双目运算符
string operato2[4]={"LEQ","GEQ","EQL","NEQ"};//双目运算符内码值
int index=0;//用于存储关键字的下标
string operato[7] = {"PLUS","MINU","MULT","DIV","GRE","LSS","ASSIGN"};//单目运算符内码值
char FILTER[4]={' ','\t','\r','\n'}; //过滤符
//const int IDENTIFIER=100;//标识符值
const int CONSTANT=101;//常数值
//const int FILTER_VALUE=102;//过滤字符值
/**判断是否为关键字并返回关键字的下标**/
int IsKeyword(string word){
    for(int i=0;i<13;i++){
        if(KEYWORD[i]==word){
            return i+1;//防止第一个0下标被返回这里对i进行了+1
        }
    }
    return 0;
}
/**判断是否为分隔符**/
bool IsSeparater(char ch){
    for(int i=0;i<10;i++){
        if(SEPARATER[i]==ch){
            return true;
        }
    }
    return false;
}
 
/**判断是否为运算符**/
bool IsOperator(char ch){
    for(int i=0;i<7;i++){
        if(OPERATOR[i]==ch){
            return true;
        }
    }
    return false;
}

/**判断是否为过滤符**/
bool IsFilter(char ch){
    for(int i=0;i<4;i++){
        if(FILTER[i]==ch){
            return true;
        }
    }
    return false;
}
/**判断是否为大写字母**/
bool IsUpLetter(char ch){
    if(ch>='A' && ch<='Z') return true;
    return false;
}
/**判断是否为小写字母**/
bool IsLowLetter(char ch){
    if(ch>='a' && ch<='z') return true;
    return false;
}
/**判断是否为数字**/
bool IsDigit(char ch){
    if(ch>='0' && ch<='9') return true;
    return false;
}
/**返回每个字的值**/
template <class T>//使用模板函数匹配string或char数组并返回匹配成功的值得下标
T value(T *a,int n,T str){
	for(int i=0;i<n;i++){
		if(a[i]==str) return i;
	}
	return -1;
}
/**词法分析**/
void analyse(FILE * fpin){
	ofstream out;
	out.open("output.txt");
	//out.open("C:\\Users\\10512\\Desktop\\此法分析器实验\\2.txt");
    char ch=' ';
    string arr="";
	string arr2="";
    while((ch=fgetc(fpin))!=EOF){//每次读取一个字符
        arr="";
        if(IsFilter(ch)){  //判断是否为过滤符



		}            
        else if(IsLowLetter(ch) | ch=='_'){       //判断是否为关键字
            while(IsLowLetter(ch)||IsDigit(ch)||ch=='_'||IsUpLetter(ch)){		//读取所有的字符
				arr += ch;
				ch=fgetc(fpin);
            }
			if(index=IsKeyword(arr)){		//判断是否为13关键字中的一个
				cout<<keyword[index-1]+" "<<arr<<endl;
				if(arr2==""){
					arr2=keyword[index-1];
				}
				out<<keyword[index-1]+" "<<arr<<endl;
				 fseek(fpin,-1L,SEEK_CUR);//指针往回退一个
			}
			else
            { 
				cout<<"IDENFR "<<arr<<endl;
				//out<<"IDENFR "<<arr<<endl;
				fseek(fpin,-1L,SEEK_CUR);
            }
        }
        else if(IsDigit(ch)){           //判断是否为数字
            while(IsDigit(ch)||(ch=='.'&&IsDigit(fgetc(fpin)))){
                arr += ch;
                ch=fgetc(fpin);
            }
            fseek(fpin,-1L,SEEK_CUR);
			cout<<"INTCON  "<<arr<<endl;
				cout<<"<无符号整数>\n<整数>"<<endl;//#语法分析器新增代码
			out<<"INTCON "<<arr<<endl;
        }
        else if(IsUpLetter(ch)||IsLowLetter(ch)){  //判断是否为字母
            while(IsUpLetter(ch)||IsLowLetter(ch)||IsDigit(ch)){
                arr += ch;
                ch=fgetc(fpin);
            }
            fseek(fpin,-1L,SEEK_CUR);
			cout<<CONSTANT+" "<<arr<<endl;
			out<<"arr"<<endl;
        }
        else switch(ch){
        case '+':
        case '-':
        case '*':
        case '/':
        case '>':
        case '<':
        case '=':
        case '!':
            {arr += ch;
				if((ch=fgetc(fpin))!=EOF&IsOperator(ch)){//匹配到单目运算符后，再次匹配后面的字符是否还有运算符
					arr +=ch;
						for(int i=0;i<4;i++){
							if(OPERATOR2[i]==arr){
								cout<<operato2[i]+" "<<arr<<endl;
								out<<operato2[i]+" "<<arr<<endl;
								break;
							}
						}
					break;
				}
				else{//匹配到单目运算符后直接输出
					fseek(fpin,-1L,SEEK_CUR);
					cout<<operato[value(OPERATOR,8,*arr.data())]+" "<<arr<<endl;
					out<<operato[value(OPERATOR,8,*arr.data())]+" "<<arr<<endl;
					break;
				}
            }
        case ';':{
				 int i = IsKeyword(arr2);
			
				 cout<<arr2<<"  "<<i<<"   "<<keyword2[i]<<endl;
				 	 arr2="";
				 }
        case ',':
        case '(':
			{//匹配（后面的字母判断是否为标识符或者关键字 else 则只输出输出（
              arr += ch;
				if((ch=fgetc(fpin))!=EOF&(IsUpLetter(ch)|(IsLowLetter(ch)))){
						//首先输出（ 及其内码值
						cout<<separater[value(SEPARATER,8,*arr.data())]+" "<<arr<<endl;
						out<<separater[value(SEPARATER,8,*arr.data())]+" "<<arr<<endl;
						arr="";//去掉（
				   while(IsUpLetter(ch)||IsLowLetter(ch)||IsDigit(ch)){		//读取所有的字符
					   arr += ch;
						ch=fgetc(fpin);
					}
				   if(index=IsKeyword(arr)){		//判断是否为13关键字中的一个
						cout<<keyword[index-1]+" "<<arr<<endl;
						out<<keyword[index-1]+" "<<arr<<endl;
						 fseek(fpin,-1L,SEEK_CUR);
							 break;
					}
					else
					{
						 cout<<"IDENFR "<<arr<<endl;
					   	 out<<"IDENFR "<<arr<<endl;
						 fseek(fpin,-1L,SEEK_CUR);
						 break;
					}
				
				}else{//匹配不到其他字符则直接输出 ; , (本身及其内码值
					 fseek(fpin,-1L,SEEK_CUR);
				  cout<<separater[value(SEPARATER,8,*arr.data())]+" "<<arr<<endl;
				  out<<separater[value(SEPARATER,8,*arr.data())]+" "<<arr<<endl;
				break;
				}
			}
        case ')':
        case '[':
        case ']':
        case '{':
        case '}':
            {//直接输出以上匹配到的字符
              arr += ch;  
			  cout<<separater[value(SEPARATER,8,*arr.data())]+" "<<arr<<endl;
			  out<<separater[value(SEPARATER,8,*arr.data())]+" "<<arr<<endl;
              break;
            }
		case '_':
			//匹配_后面的是否有字母有则输出为标识符 else 则只输出输出 _
			{	arr += ch;
				if((ch=fgetc(fpin))!=EOF&(IsUpLetter(ch)|(IsLowLetter(ch)))){
				   while(IsLowLetter(ch)||IsDigit(ch)){		//读取所有的字符
					   arr += ch;
						ch=fgetc(fpin);
					}
					cout<<"IDENFR "<<arr<<endl;
					out<<"IDENFR "<<arr<<endl;
					 fseek(fpin,-1L,SEEK_CUR);

				}else{
					cout<<"CHARCON _"<<endl;
					out<<"CHARCON _";
				}
			 break;
			}
		case '\"':
			{//用来匹配字符常量以匹配到下一个"为结束符
				ch=fgetc(fpin);
				while(ch!='\"'){
					arr += ch;
					ch=fgetc(fpin);
				}
				cout<<"STRCON "<<arr<<endl;
				out<<"STRCON "<<arr<<endl;
				  break;
			}
		case '\'':
			{	//匹配 ' 后面的是否有字符 有则输出 CHARCON  并跳过下一个 '  else则不进行输出并跳过下一个 '
				arr+=ch;
				if((ch=fgetc(fpin)) !=EOF &(IsOperator(ch)|IsUpLetter(ch)|IsLowLetter(ch)|IsDigit(ch))|ch=='\_')
				{
					arr="";
					arr=ch;
					if(ch=='_'){
						cout<<"CHARCON "<<ch<<endl;
						fseek(fpin,1L,SEEK_CUR);
						break;
					}
					cout<<"CHARCON "<<arr<<endl;
					out<<"CHARCON "<<arr<<endl;
					fseek(fpin,1L,SEEK_CUR);
					break;
				}
				fseek(fpin,1L,SEEK_CUR);
				break;
			}
        default :
			//cout<<"\""<<ch<<"\":无法识别的字符！"<<endl
			;
        }

    }
	out.close();//释放流资源
 
}
int main()
{
    FILE *fpin;//创建文件指针
	//fpin=fopen("C:\\Users\\10512\\Desktop\\此法分析器实验\\testfile.txt","r");
	fpin=fopen("testfile.txt","r");//以读的方式打开文件
    cout<<"------词法分析如下------"<<endl;
    analyse(fpin);
	fclose(fpin);
    return 0;
}