scan.cpp

自己写的一个simple语言scanner

//#include<stdio>
//#include<stdlib>
#include<string>
//#include<malloc>
//#include<ctype>
using namespace std;


#define  MAXBUF 32
char ch=' ';   /*当前的字符，初始化为空*/
int lineno;  /*行号*/
int rowno;/*列号*/


/*33个关键字*/
string keyword[33]={"and","array","begin","bool","call","case","char","constant","do","else",
              "end","false","for","if","input","integer","not","of","or","output","procedure",
              "program","read","real","repeat","set","then","to","true","until","var","while",
			  "write", };

int token;


int search(string *word) /*用来查关键字*/
{
  int i;
  for( i=0; i<33; i++)
  {
     if(word==(keyword[i])
     {
        return i;
     }
  }
  return 0;
}    
 

void scanner(FILE *fpin,FILE *fpout) /*扫描字符*/
{
    
	
	char arr[MAXBUF];/*读出的最长的字符串不超过MAXBUF*/
    int i=0;/*分析含字母的字符串用*/
    int j=0;/*分析纯数字的字符串用*/
    
    string *num;
     
    ch=fgetc(fpin);
    while(ch!=EOF)
    {
        //putchar(ch); 
        if( ch=' ')  /*空格*/
             rowno++;

        else if(ch='\t')  /*tab */
             rowno+=12;

        else if( ch='\n')  /*回车换行符*/
             lineno++;

        if(isalpha(ch))  /*读入的是字母*/
        {
            arr[i] =ch;
            i++;
           
            ch=fgetc(fpin); /*读入下一个字符*/
            rowno++;
            while(isalpha(ch) || isdigit(ch))
            {
               arr[i] =ch;
               i++;
               ch=fgetc(fpin);
               rowno++;
            }
            fseek(fpin,-1L,SEEK_CUR); /*文件指针后退一个字节*/
            rowno--;
            string *word = (char*)malloc(i+1) ;
			memcpy(word,arr,i);
			//word[i] =''\0'';


            if(search(word))/*在关键字表中查找和word字符串相同的，找到就返回种别码*/
            {
                 printf("%s\t\t&d\n",word,search(word));
                 fprintf(fpout,"%s\t%d \n",word,search(word));
            }
            else
            {
                fprintf(fpout,"%s\t%d\n",word,34);/*标示符号*/
            }
            i=0;
        }

        else if( isdigit(ch))/*读入的是数字*/
        {
            arr[i] =ch;
            i++;
            ch=fgetc(fpin);/*读入下一个字符*/
            rowno++;
            while(isdigit(ch))
            {
             arr[j] = ch;
             j++;
             fscanf(fpin,"%c",&ch);
            }
            
            fseek(fpin,-1L,SEEK_CUR);/*文件指针后退一个字节*/

            num =(char*) malloc(j+1);
            memcpy(num,arr,j); /*把数组里面的内容拷贝到链外一个数组里面*/
            /*num[j] =''\0''; */
                        

            j=0;/*恢复初始状态*/
            fprintf(fpout,"%s\t\t%d\n",num,35);/*常数*/
        
            free(num);/*释放内存*/
        }


    
        
                else if(ch==':')
                {
                        ch=fgetc(fpin);
                        if(ch=='=') 
                        {
                           fprintf(fpout,"%s\t%d\n",":=",51);    /*如果是 := */
                        }
                        else
                        {
                           fprintf(fpout,"%s\t%d\n",":",50);   /* 如果是 :  */
                           fseek(fpin,-1L,SEEK_CUR);/* 文件指针后退一个字节*/
                        }
               }
               else if (ch=='>')
               {
                        ch=fgetc(fpin);
                        if(ch=='=')    /*如果是 >= */
                        {
                           fprintf(fpout,"%s\t%d\n",">=",58);
                        }
                        else
                        {
                           fprintf(fpout,"%s\t%d\n",">",57);  /*如果是 > */
                           fseek(fpin,-1L,SEEK_CUR);/*文件指针后退一个字节 */
                        }
               }
               else if(ch=='<')
               { 
                        ch=fgetc(fpin);
                        if(ch=='>')
                        {
                            fprintf(fpout,"%s\t%d\n","<>",55);  /* 如果是 <> */
                        }
                        else if(ch=='=')
                        {
                            fprintf(fpout,"%s\t%d\n","<=",54);   /* 如果是 <= */
                        }
                        else 
                        {
                            fprintf(fpout,"%s\t%d\n","<",53);   /* 如果是 < */
                            fseek(fpin,-1L,SEEK_CUR);/*文件指针后退一个字节*/
                        }
               }
               else if(ch=='/')
               { 
                        ch=fgetc(fpin);
                        if(ch=='*')
                        {
                           ch=fgetc(fpin);
              s:
                           while(ch!='*')
                           {
                              ch=fgetc(fpin);
                           }
                           while(ch=='*')
                           {
                              ch=fgetc(fpin);
                              while(ch!='/')
                              {
                                 goto s;   /*如果是注释*/
                              }
                           }
                       }
                       else if(ch=='/')
                       {
                           ch=fgetc(fpin);
                           while(ch!='\n')
                           {
                              ch=fgetc(fpin);   /* 如果是注释 */
                           }
                       }
                       else 
                       {  
                           fprintf(fpout,"%s\t%d\n","/",48);
                           fseek(fpin,-1L,SEEK_CUR);/*文件指针后退一个字节*/
                       }
               }
               else if(ch=='\''){fprintf(fpout,"%s\t%d\n","'",38);}
               else if(ch=='+'){fprintf(fpout,"%s\t%d\n","+",43);}
               else if(ch=='-'){fprintf(fpout,"%s\t%d\n","-",45);}
               else if(ch=='*'){fprintf(fpout,"%s\t%d\n","*",41);}
               else if(ch=='('){fprintf(fpout,"%s\t%d\n","(",39);}
               else if(ch==')'){fprintf(fpout,"%s\t%d\n",")",40);}
               else if(ch=='['){fprintf(fpout,"%s\t%d\n","[",59);}
               else if(ch==']'){fprintf(fpout,"%s\t%d\n","]",60);}
               else if(ch=='.'){fprintf(fpout,"%s\t%d\n",".",46);}
               else if(ch==';'){fprintf(fpout,"%s\t%d\n",";",52);}
               else if(ch=='='){fprintf(fpout,"%s\t%d\n","=",56);}
               else if(ch==','){fprintf(fpout,"%s\t%d\n",",",44);}
               else 
                      fprintf(fpout,"无法识别的字符 %c\n",ch)  ;

               ch=fgetc(fpin);
        }
        printf("\nScan end.") ;
}




 void main()
{

       char filenamein[10];
       char filenameout[10];
       FILE* fpin;
       FILE* fpout;
       char sour;
       char option;
       printf("***********************     SCANNER      ***********************\n\n\n");

 a:    printf("Please enter sourse file name:\n");
       scanf("%s",filenamein);
       fpin=fopen(filenamein,"r");
       while (fpin==NULL)
       {
            printf("cannot open this file!!!\n\n\n");
            goto a;

       }
       printf("The sourse programe is:\n\n");
       sour=fgetc(fpin); /*将原程序显示到屏幕上*/
       while (sour!=EOF)
       {
           putchar(sour);
           sour=fgetc(fpin);
        }
       fseek(fpin,0L,0);/*文件指针回到开始位置，以备扫描*/

       printf("\n\nReady to Scan!!!\n");
 b:    printf("Please enter a file name for save result:\n");
       scanf("%s",filenameout);
       fpout=fopen(filenameout,"w");
       if (fpout==NULL)
       {
           printf("cannot open this file!!!\n\n\n");
           goto b;
       }

       scanner(fpin,fpout); /*开始扫描*/
       fclose(fpin);
       fclose(fpout);

   

}