lexer.cpp

编译原理课程设计中

#include <cstdio>
#include <cctype>
#include <map>
#include <string>
#include <cstring>
#include <utility>

#define BSIZE 100 //规定记号的长度小于100
#define NONE -1
#define EOS '\0'

#define NUM 256
#define ID 257
#define CONST 258
#define VAR 259
#define EQU 260
#define LESS_EQU 261
#define MORE_EQU 262
#define NOT_EQU 263
#define CALL 264
#define BEGIN 265
#define END 266
#define ODD 267
#define IF 268
#define THEN 269
#define WHILE 270
#define DO 271
#define DONE 273
#define PROCEDURE 274
#define EVU 275  //:=
#define LESS 276
#define MORE 277
#define FLOAT 278
using namespace std;

map<string,pair<int,int> > symtable;

struct entry
{
	char lexptr[BSIZE];
	int token;
};

char token[500][BSIZE];
int token_num;


//#include "global.h"

int lexan();
void init();
void print(int t,int tt);
void error();
int lookahead;

entry s[18]={{"const",CONST},
{"var",VAR},
{"=",EQU},
{"<=",LESS_EQU},
{">=",MORE_EQU},
{"<>",NOT_EQU},
{">",MORE},
{"<",LESS},
{"call",CALL},
{"begin",BEGIN},
{"end",END},
{"odd",ODD},
{"if",IF},
{"then",THEN},
{"while",WHILE},
{"do",DO},
{"procedure",PROCEDURE},
{":=",EVU}
};
char buf[BSIZE];
int b;
int lineno=1;
int tokenval=NONE;
double tokenval1;
int lexan()
{
	char t;
	map<string,pair<int,int> >::iterator iter;
	while(1)
	{
		t=getchar();
		if(t==' '||t=='\t')
		{
			return 0;
		}
		else if(t=='\n')
		{
			lineno++;
			return 0;
		}
		else if(isdigit(t))
		{
			ungetc(t,stdin);
			scanf("%d",&tokenval);
			t=getchar();
			if(t=='.')
			{
				ungetc(t,stdin);
				while(tokenval)
				{
					ungetc(tokenval%10+'0',stdin);
					tokenval/=10;
				}
			//	ungetc(' ',stdin);
				scanf("%lf",&tokenval1);
				printf("%lf\n",tokenval1);
				t=getchar();
				if(isalpha(t))
				{
					error();
					ungetc(t,stdin);
					return 0;
				}
				if(t!=EOF)
					ungetc(t,stdin);
				return FLOAT;
			}
			if(isalpha(t))
			{
				error();
				ungetc(t,stdin);
				return 0;
			}
			ungetc(t,stdin);
			return NUM;
		}
		else if(isalpha(t))
		{
			b=0;
			memset(buf,'\0',sizeof(buf));
			while(isalnum(t))
			{
				buf[b++]=t;
				t=getchar();
				if(b>=BSIZE)
				{
					error();
					return 0;
				}
			}
			iter=symtable.find(buf);
			if(iter==symtable.end())
			{
				if(t!='\n'&&t!=' '&&t!=EOF)
				{
					error();
					return 0;
				}
				tokenval=token_num++;
				strcpy(token[tokenval],buf);
				symtable[buf]=make_pair(ID,tokenval);
			}
			tokenval=symtable[buf].second;
			if(t!=EOF)
				ungetc(t,stdin);
			return symtable[buf].first;
		}
		else if(t==EOF)
			return DONE;
		else if(t=='=')
			return EQU;
		else if(t==':')
		{
			t=getchar();
			if(t=='=')
				return EVU;
			else {error();return 0;}
		}
		else if(t=='<')
		{
			t=getchar();
			if(t=='=')
				return LESS_EQU;
			else if(t=='>'){
                 return NOT_EQU;
                 }
            else
			{
				ungetc(t,stdin);
				return LESS;
			}
		}
		else if(t=='>')
		{
			t=getchar();
			if(t=='=')
				return MORE_EQU;
			else
			{
				ungetc(t,stdin);
				return MORE;
			}
		}
		else
			return t;
	}
}


void init()
{
	int i;
	symtable.clear();
	for(i=0;i<18;i++)
		symtable[s[i].lexptr]=make_pair(s[i].token,NONE);
	token_num=0;
	lineno=1;
}

void print(int t)
{
	switch(t)
	{
	case '+':
	case '-':
	case '*':
	case '/':
		printf("<%c %d>\n",t,t);break;
	case EQU:
		printf("<= %d>\n",t);break;
	case LESS_EQU:
		printf("<<= %d>\n",t);break;
	case LESS: 
		printf("<< %d>\n",t);break;
	case MORE:
		printf("<> %d>\n",t);break;
	case MORE_EQU: 
		printf("<>= %d>\n",t);break;
	case NOT_EQU: 
		printf("<<> %d>\n",t);break;
	case CALL:
		printf("<call %d>\n",t);break;
	case BEGIN:
		printf("<begin %d>\n",t);break;
	case END: 
		printf("<end %d>\n",t);break;
	case ODD:
		printf("<odd %d>\n",t);break;
	case IF:
		printf("<if %d>\n",t);break;
	case THEN: 
		printf("<then %d>\n",t);break;
	case WHILE:
		printf("<while %d>\n",t);break;
	case DO: 
		printf("<do %d>\n",t);break;
	case PROCEDURE: 
		printf("<procedure %d>\n",t);break;
	case NUM:
		printf("<NUM %d>\n",t);break;
	case VAR:
		printf("<var %d>\n",t);break;
	case EVU:
		printf("<:= %d>\n",t);break;
	case CONST:
		printf("<const %d>\n",t);break;
	case FLOAT:
		printf("<float %d>\n",t);break;
	case ID:
		printf("<ID %d>\n",t);break;
	default :
		printf("<%c>\n",t);
		break;
	}
}

void error()
{
	//printf("%d\n",lookahead);
	printf("%d error\n",lineno);
}
int main()
{
	freopen("input.txt","r",stdin);
	//freopen("output.txt","w",stdout);
	init();
	int t;
	while(1)
	{
            t=lexan();
			if(t==DONE)
				break;
			if(t!=0)
			print(t);
			
	}
	//system("pause");
	return 0;
}