词法分析器.cpp

编译原理中的词法分析器

#include "stdafx.h"
#include "stdio.h"
#include "iostream.h"
#include "ctype.h"
#include "string.h"
typedef struct
{
	int num;
	char word[20];
}key;
int line=1,row=0,r;    
key keyword[15]=  
{
	{1,"PROGRAM"},{2,"CONST"},{3,"VAR"},
	{4,"INTEGER"},{5,"LONG"},{6,"PROCEDURE"},
	{7,"IF"},{8,"THEN"},{9,"WHILE"},
	{10,"DO"},{11,"READ"},{12,"WRITE"},
	{13,"BEGIN"},{14,"END"},{15,"ODD"}
};
void output(int m,char * s)  
{
	cout<<"["<<m<<","<<s<<"]"<<' '<<line<<' '<<row-strlen(s)+1<<endl;
}
char getch(FILE * fp)  
{
	char ch;
	ch=fgetc(fp);
	if(ch=='\n') 
	{
		line=line+1;
		row=0;      
	}
	else
	{
		row=row+1;
		r=row;
	}
	return ch;
}
void seek(FILE *fp,char ch)  
{
	if(ch=='\n')
	{
		line=line-1;
        row=r;
	}
	else
	{
		row=row-1;
		r=row;
	}
	fseek(fp,-1,1);
}
int compare(char *s1) 
{ 
	int i;
	for(i=0;i<20;i++)
	{
		if(strcmp(s1,keyword[i].word)==0)
			return keyword[i].num;
	}
	return -1;
}
void getsym(FILE * fp)  
{
	char ch;
	int n,m;
	char str[20];
	int val;
	ch=getch(fp);
	while(ch==' ')ch=getch(fp);
	if(isalpha(ch)) 
	{
		n=0;
		while(isalnum(ch))
		{
			if(n<20)
			{
				str[n]=ch;
				n=n+1;
			}
			ch=getch(fp);
		}
		seek(fp,ch);
		str[n]='\0';
		m=compare(str);
		if(m!=-1)
			output(m,keyword[m-1].word);  
		else
			output(34,str);  
	}
	else if(isdigit(ch))
	{
		val=0;
		int i=0;
		while(isdigit(ch))
		{
			val=val*10+ch-'0';
			ch=getch(fp);
			i=i+1;
		}
		seek(fp,ch);
		cout<<'['<<33<<','<<val<<']'<<line<<' '<<row-i+1<<endl; 
	}
	else         
		switch(ch)
	{
        case '+':output(16,"+");break;    
        case '-':output(17,"-");break;
		case '*':output(18,"*");break;
		case '/':output(19,"/");break;
		case '=':output(20,"=");break;
		case ',':output(26,",");break;
		case '.':output(27,".");break;
		case ';':output(28,";");break;
		case '(':output(31,"(");break;
		case ')':output(32,")");break;
		case '#':output(35,"#");break;
		case '<':         
			str[0]=ch;
			ch=getch(fp);
			if(ch=='>')
			{
				str[1]=ch;
				str[2]='\0';
				output(21,str);
			}
			else 
			{
				if(ch=='=')
				{
					str[1]=ch;
					str[2]='\0';
					output(23,str);
				}
				else 
				{
					str[1]='\0';
					seek(fp,ch);
					output(22,str);
				}
			}
			break;	
		case '>':
			str[0]=ch;
			ch=getch(fp);		
			if(ch=='=') 
			{
				str[1]=ch;
				str[2]='\0';
				output(25,str);
			}
			else 
			{
				str[1]='\0';
				seek(fp,ch);
				output(24,str);
			}
			break;
		case ':':
			str[0]=ch;
			ch=getch(fp);
			if(ch=='=')
			{
				str[1]=ch;
				str[2]='\0';
				output(30,str);
			}
			else 
			{
				str[1]='\0';
				seek(fp,ch);
				output(29,str);
			}
			break;
		default: 
			{ 
				if(ch!=EOF&&ch!=' '&&ch!='\n')
					cout<<"ERROR: "<<ch<<" line: "<<line<<" row: "<<row<<endl;
			}
	}
}
int main()
{
	FILE *fp;
    char wenjian[30];
    cout<<"Please input the file you want to transform:";
    cin>>wenjian;
    fp=fopen(wenjian,"r");
    while(!feof(fp))
    {
		getsym(fp);
    }
	return 0;
}