词法分析设计.cpp

编译原理中的词法分析程序

#include <stdio.h>
#include <ctype.h>
#include <string.h>

#define ID 12	//标识符
#define INT 13	//无符号整数
#define SIN 14	//单字符分界符
#define DOU 15	//双字节分界符
#define LT 16	//小于
#define EQ 17	//等于
#define GT 18	//大于
#define COLON 19	//冒号
#define SLASH 20	//斜竖

#define NULL 0

char *key[11]={"begin","end","if","then","else",
				"for","do","while","and","or","not"};	//11个保留字，类别编码为1-11
char TOKEN[20];	//最大扫描字符数为20个
FILE *fp,*fp1;


int lookup(char);
void out(int,char);
void report_error(char);
void scanner_example(FILE*);

int lookup(char* ch)
//以TOKEN中的字符串查保留字表，若查到，就返回相应关键字的类别码，否则返回0
{
	int i=0,j=0,len1=0,len2=0;
	len1=strlen(ch);
	for(i=0;i<11;i++)	//查保留字表，大小写不敏感
	{
		len2=strlen(key[i]);
		if(len1==len2)
		{
			for(j=0;j<len1;j++)
			{
				if((*(key[i]+j)-*(ch+j))==0||(*(key[i]+j)-*(ch+j))==32)	//与保留关键字的每个字母进行比较
				{continue;}
				else goto label;
			} 
			return (i+1);
		}
		label:continue;
	}
	return(0);
}


void out(int a,char* ch)
//输出二元式写到result中
{
	fp1=fopen("result.txt","a");
	fprintf(fp1,"(%d,%s)\n",a,ch);
}

void report_error(char ch)
//报告错误
{
	if(ch!='\n')
		printf("the letter '%c' is wrong,please check it out!\n",ch);
}

void scanner_example(FILE* fp)
//对扫描到的字母进行分析
{
	char ch;
	int i,c;
	while(ch!=EOF){
		ch=fgetc(fp);
		if(isalpha(ch))	//对读到的首字母是a-z、A-Z的情况的分析
		{
			TOKEN[0]=ch;
			ch=fgetc(fp);
			i=1;
			while(isalnum(ch))
			{
				TOKEN[i]=ch;
				i++;
				ch=fgetc(fp);
			}
			TOKEN[i]='\0';
			fseek(fp,-1,1);
			c=lookup(TOKEN);  //与关键字表进行匹配，匹配上返回关键字码，匹配不上返回零
			if(c==0)
				out(ID,TOKEN);   // out(int a,char* ch) 将结果写到result 中
			else
				out(c,TOKEN);
		}
		else
			if(isdigit(ch))	//对首字母是数字的情况的分析
			{
				TOKEN[0]=ch;
				ch=fgetc(fp);
				i=1;
				while(isdigit(ch))
				{
					TOKEN[i]=ch;
					i++;
					ch=fgetc(fp);
				}
				TOKEN[i]='\0';
				fseek(fp,-1,1);
				out(INT,TOKEN);   //类别码+字符
			}
			else
				switch(ch)	//对其他情况的分析
			{
				case '+':out(SIN,"+");break;
				case '-':out(SIN,"-");break;
				case '*':ch=fgetc(fp);
						 if(ch=='/')
							 out(DOU,"*/");
						 else
						 {
							fseek(fp,-1,1);
							out(SIN,"*");	 
						 }
						 break;
				case ';':out(SIN,";");break;
				case '(':out(SIN,"(");break;
				case ')':out(SIN,")");break;
				case '<':ch=fgetc(fp);
						 if(ch=='=')
							 out(DOU,"<=");
						 else if(ch=='>')
							 out(DOU,"<>");
						 else
						 {
							fseek(fp,-1,1);
							out(LT,"<");
						 }
						 break;
				case '=':out(EQ,"=");break;
				case '>':ch=fgetc(fp);
						 if(ch=='=')
							 out(DOU,">=");
						 else
						 {
							fseek(fp,-1,1);
							out(GT,">");
						 }
						 break;
				case ':':ch=fgetc(fp);
						 if(ch=='=')
							 out(DOU,":=");
						 else
						 {
							fseek(fp,-1,1);
							out(COLON,":");
						 }
						 break;
				case '/':ch=fgetc(fp);
						 if(ch=='*')
						 {
							 out(DOU,"/*");
							 ch=fgetc(fp);
							 while(ch!=EOF)	//对于注释部分不进行分析
							 {
								if(ch=='*'&&fgetc(fp)=='/')
								{
									out(DOU,"*/");
									break;
								}
								else
									ch=fgetc(fp);
							 }
						 }
						 else
						 {
							fseek(fp,-1,1);
							out(SLASH,"/");
						 }
						 break;
				default:if(ch!=' '&&ch!=EOF)
							report_error(ch);	//如果所扫描的字符不包括在所规定的语言单词符号中，则报告错误
						break;
		}
	}
	return;
}


void main()
{
	
	if((fp=fopen("test.txt","r"))==NULL)	//读取文件失败
		printf("读取文件失败!\n");
	else{
		scanner_example(fp);	
		printf("程序运行完毕！请查看result文件.\n");
	}
}