lexical_analysis.cpp

词法分析的又一个例子 这个将词法分析做成了一个子程序便于后来与语法分析及语法制导连起来

#include<stdio.h>
#include<string.h>
#include<conio.h>
#define MAX 20

char RW[]="program input output begin end var integer real for to if then else do while write array procedure " ;/* Reserved word */
char ASSIGN[]="+ - * / = > < >= <= <> := .. ";
char BOS[]=", ; : { } [ ] ( ) ";/* Box off the sign */
char EB[]="COMMA  SEMI   COLON  LBRACE RBRACE LMREN  RMREN  LPAREN RPAREN ";/* Explain for BOS */
int LINE=1,L;

int Judgment(char J[],char B[])
{
    int i,j=0,k;
    while(J[j]!='\0'){
        i=j;
        k=0;
    while(B[k]==J[i]){
        i++;
        k++;
    }   
    if((B[k]=='\0')&&(J[i]==' '))
    return j;
    else
        while(J[j++]!=' ');     
    }
    return -1;
}

int Judge_char(char J[],char ch1,char ch2)
{
    char O[2];
    O[0]=ch1;
    O[1]=ch2;
    O[2]='\0';
    return Judgment(J,O);
}

int Judge_number(char B[])
{
    int m=-1,n=0;
    while(((B[n]!='\0')&&(B[n]<58)&&(B[n]>47))||(B[n]=='.')){
	if(B[n]=='.')
	m++;
   	n++;
    }
    if((B[n]=='\0')&&(m<=0))
    return 1;
    else return 0;
}

void fview(FILE *fp,char B[],int n)/* printf B from B[n] to next blank */
{
    int i;
    char V[10];
    for(i=0;B[n]!=' ';i++,n++)
    V[i]=B[n];
    V[i]='\0';
    fprintf(fp,"%s",V);
}

int Get_buffer(FILE *fp,char B[])
{
    int i=0;
	char ch1,ch2;
    ch1=fgetc(fp);
    if(ch1==EOF)
	return -1;
	if((ch1=='\n')||(ch1==9)||(ch1==' ')){
	    if(ch1=='\n')
     LINE++;
     return 0;
	} 
    ch2=fgetc(fp);		
	if((ch2==EOF)||(ch2=='\n')||(ch2==9)||(ch2==' ')){
	    B[0]=ch1;
	    B[1]='\0';
	    fseek(fp,-1L,1);
	    if(Judge_char(ASSIGN,ch1,'\0')>=0)
             return 3;
         if(Judge_char(BOS,ch1,'\0')>=0)
             return 4;
	    return 1;
    }	
	else{
     fseek(fp,-2L,1);  
     while((ch1!=' ')&&(ch1!=9)&&(ch1!='\n')){
         ch1=fgetc(fp);
         ch2=fgetc(fp);
         B[i]=ch1;
         B[i+1]=ch2;
         i++;
         if(ch2==EOF)
             return 1;
         if((ch1=='/')&&(ch2=='/')){
             B[i-1]='\0';
             while((ch2!='\n')&&(ch2!=EOF))
             ch2=fgetc(fp);
            if(ch2=='\n')
                 fseek(fp,-1L,1);
             return 1;
         }/* delete "//......" */  
         if((ch1=='/')&&(ch2=='*')){
             B[i-1]='\0';
             L=LINE;
             while(ch2!=EOF){
                 ch1=fgetc(fp);
                 ch2=fgetc(fp);
                 fseek(fp,-1L,1); 
                 if(ch1=='\n')
                     LINE++;
                 if((ch1=='*')&&(ch2=='/')){
                     fseek(fp,1L,1); 
                     return 1;   
                }     
             }// delete "/*...*/"
             return -2;
         }
         fseek(fp,-1L,1);
         if(Judge_char(ASSIGN,ch1,ch2)>=0){
             B[i]=ch2;
             B[i+1]='\0';
             fseek(fp,1L,1);
             return 2;
         } 
         if(Judge_char(ASSIGN,ch1,'\0')>=0){
             B[i]='\0';
             return 3;
         }    
         if(Judge_char(BOS,ch1,'\0')>=0){
             B[i]='\0';
             return 4;
         }
     }/* end while */
     if(ch1=='\n')
     fseek(fp,-1L,1);
     B[i-1]='\0';
     return 1;  
    }/* end else */ 
} /* -1:end -2:error 0:jump this time 1:get buffer  2-4:get sign */

void Print_result(FILE *fp,char B[],int n)
{
    int i=0,line;
    char ch1,ch2;
    char V[10];
    while(B[i]!='\0')
    i++;
    ch1=B[i-1];
    ch2=B[i-2];
    if(B[i+1]=='*')
    line=L;
    else
    line=LINE;
    switch(n){
        case 1:{
            if(B[0]=='\0')
            break;
            if(Judgment(RW,B)>=0){
                fprintf(fp,"     <token lineno=%d   string=\"%s\"   type=%s ></token>\n",line,B,B);
            }    
            else if(Judge_number(B)&&(B[0]!='.'))
            fprintf(fp,"     <token lineno=%d   string=\"%s\"   type=NUMBER ></token>\n",line,B);   
            else if(B[0]!='.')
            fprintf(fp,"     <token lineno=%d   string=\"%s\"   type=ID ></token>\n",line,B); 
            break;
        }
        case 2:{
            B[i-2]='\0';
            Print_result(fp,B,1);
            fprintf(fp,"     <token lineno=%d   string=\"%c%c\"   type=ASSIGN ></token>\n",LINE,ch2,ch1);
            break;
        } 
        case 3:{
            B[i-1]='\0';
            Print_result(fp,B,1);
            fprintf(fp,"     <token lineno=%d   string=\"%c\"   type=ASSIGN ></token>\n",LINE,ch1);
            break;
        }    
        case 4:{
            B[i-1]='\0';
            Print_result(fp,B,1);
            fprintf(fp,"     <token lineno=%d   string=\"%c\"   type=",LINE,ch1);
            fview(fp,EB,Judge_char(BOS,ch1,'\0')/2*7);
            fprintf(fp," ></token>\n");
            break;
        }   
    }
}

int Lexical_analysis()
{
    FILE *fp1,*fp2;
	int ES;
	char ch;
	char filename[10],buffer1[MAX],buffer2[MAX];
	printf("Enter name of the program file? ");
	gets(filename);
	if((fp1=fopen(filename,"a"))==NULL){
     printf("Can not open file!\n");
     return 0;
	}
	fputc('/',fp1);
	fputc('/',fp1);
	fclose(fp1);
	fp1=fopen(filename,"r");
	fseek(fp1,-3L,2);
    if(fgetc(fp1)!='.'){
            printf("Program not ended by '.'!");
            fclose(fp1);
            return 0;
    }
    fseek(fp1,0L,0);
    printf("Enter name of the result file:");
	scanf("%s",&filename);
	fp2=fopen(filename,"a");
 	fprintf(fp2,"<?xml version=\"1.0\"?>\n<root>\n");
	do{
     ES=Get_buffer(fp1,buffer1);
     if(ES&&(ES!=-1)&&(buffer1[0]!='\0'))
     Print_result(fp2,buffer1,ES);
	}while(ES>=0);
    if(ES==-2){
        printf("Error in line %d: Explain dismatch!",L);
        Print_result(fp2,buffer1,1);	
        fprintf(fp2,"</root>\nError in line %d: Explain dismatch!",L);
    } 
    else   
    fprintf(fp2,"</root>\n");
    fclose(fp1);
    fclose(fp2);
}

main()
{
	Lexical_analysis();
	getch();
}