c++词法分析.cpp

对*.cpp文件各个标识符的简易分析

#include<iostream.h>
#include<stdio.h>
#include<ctype.h>
#include<string.h>
int isID();
int isNbr();
int match();
void other();
void report_error();
char token[20];
char file[20];
FILE *fp;
char ch;
int i;
int main(){
	cout<<"Please input the filename you want to open:";
	cin>>file;
	cout<<endl;
	fp=fopen(file,"r");
	if(fp){
		ch=fgetc(fp);
		while(ch!=EOF){
			if(isalpha(ch)){
				if(isID()){
					if(match())	
						cout<<"("<<"KEYWORD,"<<token<<")";
					else
						cout<<"("<<"IDENTIFIER,"<<token<<")";
						}
			}
	        else{
		        if(isdigit(ch))
			        isNbr();
		        else
			        other();
			}
			ch=fgetc(fp);
			cout<<"  ";
		}
		cout<<endl<<endl;
	}
	else 
		cout<<"The file you want to open doesn't exist in the current path!"<<endl<<endl;
	return 0;
}
int isID(){
	token[0]=ch;
	ch=fgetc(fp);
	i=1;
	while(isalnum(ch)||ch=='_'){
		token[i]=ch;
		i++;
		ch=fgetc(fp);
	}
	token[i]='\0';
	fseek(fp,-1,1);
	return 1;
}
int isNbr(){
	token[0]=ch;
	ch=fgetc(fp);
	i=1;
	while(isdigit(ch)||ch=='.'){
		token[i]=ch;
		i++;
		ch=fgetc(fp);
	}
	token[i]='\0';
	fseek(fp,-1,1);
	cout<<"(NUMBER,"<<token<<")";
	return 1;
}
int match(){
	char *keyword[]={
	"if","else","operator","throw","auto","enum","private","true","bool","explicit",
	"protected","try","break","extern","public","typedef","case","false","register",
	"typeid","catch","float","reinterpret_cast","typename","char","for","return","union",
	"class","friend","short","unsigned","const","goto","signed","using","const_cast","if",
	"sizeof","virtual","continue","inline","static","void","default","int","static_cast",
	"volatile","delete","long","struct","wchar_t","do","mutable","switch","while","double",
	"namespace","template","dynamic_cast","new","this"};
	for(int i=0;i<62;i++){
		if(!strcmp(token,keyword[i])){
			return 1;
			break;
		}		
	}
	return 0;
}
void other(){
	switch(ch){
	case'<':
		ch=fgetc(fp);
		if(ch=='<'){
			ch=fgetc(fp);
			if(ch=='=')
				cout<<"(ASSLSHIFT,<<=)";
			else{
				fseek(fp,-1,1);
				cout<<"(SHIFTOP,<<)";
			}
		}
		else{
			if(ch=='=')
				cout<<"(LE,<=)";
			else{
				fseek(fp,-1,1);
				cout<<"(LT,<)";
			}
		}
		break;
	case'=':
		ch=fgetc(fp);
		if(ch=='=')
			cout<<"(EQUOP,==)";
		else{
			fseek(fp,-1,1);
			cout<<"(EQ,=)";
		}
		break;
	case'>':
		ch=fgetc(fp);
		if(ch=='>'){
			ch=fgetc(fp);
			if(ch=='=')
				cout<<"(ASSRSHIFT,>>=)";
			else{
				fseek(fp,-1,1);
				cout<<"(SHIFTOP,>>)";
			}
		}
		else{
			if(ch=='=')
				cout<<"(GE,>=)";
			else{
				fseek(fp,-1,1);
				cout<<"(GT,>)";
			}
		}
		break;
	case'+':
		ch=fgetc(fp);
		if(ch=='+')
			cout<<"(INCOP,++)";
		else{
			if(ch=='=')
				cout<<"(ASSIGNOP,+=)";
			else{
				fseek(fp,-1,1);
				cout<<"(PLUS,+)";
			}
		}
		break;
	case'-':
		ch=fgetc(fp);
		switch(ch){
		case'-':cout<<"(INCOP,--)";
			break;
		case'=':cout<<"(ASSIGNOP,-=)";
			break;
		case'>':cout<<"(STRUCTOP,->)";
			break;
		default:fseek(fp,-1,1);
			cout<<"(MINUS,-)";
			break;
		}
	case'*':
		ch=fgetc(fp);
		if(ch=='=')
			cout<<"(ASSSTAR,*=)";
		else{
			fseek(fp,-1,1);
			cout<<"(STAR,*)";
		}
		break;
	case'&':
		ch=fgetc(fp);
		if(ch=='&'){
			ch=fgetc(fp);
			if(ch=='=')
				cout<<"(ASSANDAND,&&=)";
			else{
				fseek(fp,-1,1);
				cout<<"(ANDAND,&&)";
			}
		}
		else{
			fseek(fp,-1,1);
			cout<<"(AND,&)";
		}
		break;
	case'|':
		ch=fgetc(fp);
		if(ch=='|'){
			ch=fgetc(fp);
			if(ch=='=')
				cout<<"(ASSOROR,||=)";
			else{
				fseek(fp,-1,1);
				cout<<"(OROR,||)";
			}
		}
		else{
			if(ch=='=')
				cout<<"(ASSOR,|=)";
			else{
				fseek(fp,-1,1);
				cout<<"(OR,|)";
			}
		}
		break;
	case'^':
		ch=fgetc(fp);
		if(ch=='=')
			cout<<"(ASSXOR,^=)";
		else{
			fseek(fp,-1,1);
			cout<<"(XOR,^)";
		}
		break;
	case'!':
		ch=fgetc(fp);
		if(ch=='=')
			cout<<"(NE,!=)";
		else{
			fseek(fp,-1,1);
			cout<<"(UNOP,!)";
		}
		break;
	case'~':cout<<"(UNOP,~)";
		break;
	case'.':cout<<"(POINT,.)";
		break;
	case'(':cout<<"(LP,()";
		break;
	case')':cout<<"(RP,))";
		break;
	case'{':cout<<"(LC,{)";
		break;
	case'}':cout<<"(RC,})";
		break;
	case'[':cout<<"(LB,[)";
		break;
	case']':cout<<"(RB,])";
		break;
	case',':cout<<"(COMMA,,)";
		break;
	case';':cout<<"(SEMI,;)";
		break;
	case'?':
		ch=fgetc(fp);
		if(ch==':')
			cout<<"(QC,?:)";
		else{
			fseek(fp,-1,1);
			cout<<"(QUEST,?)";
		}
		break;
	case':':
		ch=fgetc(fp);
		if(ch==':')
			cout<<"(CC,::)";
		else{
			fseek(fp,-1,1);
			cout<<"(COLON,:)";
		}
		break;
	case'/':
		ch=fgetc(fp);
		switch(ch){
		case'=':cout<<"(ASSDIV,/=)";
			break;
		case'/':
			while(ch!='\n'){
		    	ch=fgetc(fp);
			}
			break;
		case'*':{
s1:             while(ch!='*'){
				    ch=fgetc(fp);
				}
			    ch=fgetc(fp);
			    if(ch!='/')
				    goto s1;
			}
			break;
		default:
			fseek(fp,-1,1);			
            cout<<"(IDIV,/)";
			break;
		}
		break;
	case'\\':
		cout<<"(FANXIEGANG,\\)";
		break;
	case'%':
		ch=fgetc(fp);
		if(ch=='=')
			cout<<"(ASSMOD,%=)";
		else{
			fseek(fp,-1,1);
			cout<<"(DIVOP,%)";
		}
		break;
	case'@':cout<<"(@,@)";
		break;
	case'#':cout<<"(JINGHAO,#)";
		break;
	case'"':cout<<"(YINGHAO,\")";
	case' ':
		 break;
	case'\n':
		break;
	case'	':
		break;
	default:report_error();
		break;
	}
}
void report_error(){
	cout<<"Error finds!";
}