lex.cpp

一个词法分析器的源程序

#include"lex.h"
#include"error.h"
#include"Table.h"
#include<stdlib.h>
#include<iostream.h>
///////公用量区//////////////////////////////////////////////////////////
FILE*fd,*fw;
char strout1[2000];
int c1=0;
char *guanjian[]={"and","array","begin","bool","call","case","char","constant","dim","do","else",
"end","false","for","if","input","integer","not","of","or","output","procedure","program","read",
"real","repeat","set","stop","then","to","true","until","var","while","write"};
///////////lex.cpp内部公用量//////////////////////////////////////////////
static char infile[30];
static char error[1000];
static int c3=0;

//////////////////////////////////////////////////////////////////////////
static FILE*init()
{
  cout<<"\t\tPlease input the address of infile\n";
  cout<<"\t\t";
  cin>>infile;
  FILE*filename1;
  if((filename1=fopen(infile,"r"))==NULL)
  {
    return(NULL);
  }
  else
  {
    return(filename1);
  }
}

static void startl()
{
  char string[MAX/2],currentch;
  int xb,Isg;
  int signid;
  cout<<"\n\n\n\n\n\n\t\tSIMPLE语言词法分析器\n\n";
  cout<<"\t\t\t\t--计算机02(2)张斌\n";
  cout<<"\t\t\t\t200245003122\n";
  fd=init();
  fw=fopen("token.txt","w");
  system("cls");
  do
  {
    currentch=get_ch();
	if(currentch==' ')
	{
	  continue;
	}
	else if(currentch=='\n')
	{
	  fprintf(fw," \n");
	}
	else if(IsLetter(currentch))
	{
	  string[0]=currentch;
	  xb=1;
	  while(IsLetter(currentch)||IsDigit(currentch))
	  {
	    currentch=get_ch();
		string[xb]=currentch;
		xb++;
	  }
	  string[xb-1]='\0';
	  retract();
	  if(!IsSign(currentch)&&currentch!=' '
		  &&currentch!='\n')//出现非法字符，识别失效
	  {
	    continue;
	  }
	  Isg=IsGJZ(string);
	  if(Isg==0)
	  {
        signid=AddTab(36,string);
	    fprintf(fw,"<36\t");
		fprintf(fw,"%d>\n",signid);
	  }
	  else
	  {
	    fprintf(fw,"<%d\t",Isg);
		fprintf(fw,"->\n");
	  }
	}
	else if(IsDigit(currentch))
	{
	  string[0]=currentch;
	  xb=1;
	  while(IsDigit(currentch))
	  {
	    currentch=get_ch();
		string[xb]=currentch;
		xb++;
	  }
	  string[xb-1]='\0';
	  retract();
	  signid=AddTab(37,string);
      fprintf(fw,"<37\t");
	  fprintf(fw,"%d>\n",signid);
	}
	else if(currentch=='\'')
	{
	  currentch=get_ch();
	  xb=0;
L1:	  while(IsSign(currentch)&&(currentch!='\'')
		  &&(currentch!='\n')&&(currentch!=' '))
	  {
	    string[xb++]=currentch;
		currentch=get_ch();
	  }
	  if(!IsSign(currentch)&&currentch!='\n'
		  &&currentch!=' ')
	  {
		Error(col,1);
		currentch=get_ch();
		goto L1;
	  }
	  else if(currentch=='\'')
	  {
	    string[xb]='\0';
		signid=AddTab(38,string);
		fprintf(fw,"<38\t");
		fprintf(fw,"%d>\n",signid);
	  }
	  else
	  {
	    Error(col,2);
	  }
	}
	else if(currentch=='+')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<43\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch=='-')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<45\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch=='(')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<39\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch==')')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<40\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch==',')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<44\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch==';')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<52\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch=='=')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<56\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch=='[')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<59\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch==']')
	{
	  string[0]=currentch;
	  string[1]='\0';
	  fprintf(fw,"<60\t");
	  fprintf(fw,"->\n");
	}
	else if(currentch=='*')
	{
	  string[0]=currentch;
	  currentch=get_ch();
	  if(currentch=='/')
	  {
	    string[1]=currentch;
		string[2]='\0';
	  }
	  else
	  {
	    string[1]='\0';
		retract();
		fprintf(fw,"<41\t");
		fprintf(fw,"->\n");
	  }
	}
	else if(currentch=='.')
	{
	  string[0]=currentch;
	  currentch=get_ch();
	  if(currentch=='.')
	  {
	    string[1]=currentch;
		string[2]='\0';
		fprintf(fw,"<47\t");
		fprintf(fw,"->\n");
	  }
	  else
	  {
	    string[1]='\0';
		retract();
		fprintf(fw,"<46\t");
		fprintf(fw,"->\n");
	  }
	}
	else if(currentch==':')
	{
	  string[0]=currentch;
	  currentch=get_ch();
	  if(currentch=='=')
	  {
	    string[1]=currentch;
		string[2]='\0';
		fprintf(fw,"<51\t");
		fprintf(fw,"->\n");
	  }
	  else
	  {
	    string[1]='\0';
		retract();
		fprintf(fw,"<50\t");
		fprintf(fw,"->\n");
	  }
	}
	else if(currentch=='<')
	{
	  string[0]=currentch;
	  currentch=get_ch();
	  if(currentch=='=')
	  {
	    string[1]=currentch;
		string[2]='\0';
		fprintf(fw,"<54\t");
		fprintf(fw,"->\n");
	  }
	  else
	  {
	    if(currentch=='>')
		{
		  string[1]=currentch;
		  string[2]='\0';
		  fprintf(fw,"<55\t");
		  fprintf(fw,"->\n");
		}
	    else
		{
	      string[1]='\0';
	   	  retract();
		  fprintf(fw,"<53\t");
		  fprintf(fw,"->\n");
		}
	  }
	}
	else if(currentch=='>')
	{
	  string[0]=currentch;
	  currentch=get_ch();
	  if(currentch=='=')
	  {
	    string[1]=currentch;
		string[2]='\0';
		fprintf(fw,"<58\t");
		fprintf(fw,"->\n");
	  }
	  else
	  {
	    string[1]='\0';
		retract();
		fprintf(fw,"<57\t");
		fprintf(fw,"->\n");
	  }
	}
	else if(currentch=='/')
	{
      string[0]=currentch;
	  currentch=get_ch();
      if(currentch=='*')
	  {
	    string[1]=currentch;
		string[2]='\0';	
        currentch=get_ch();
L2:		while(currentch!='*'&&currentch!='\n')
		{
		  currentch=get_ch();
		}
		if(currentch=='*')
		{
		  currentch=get_ch();
		  if(currentch=='/')
		  {
            retract();
			retract();
			continue;
		  }
		  else
		  {
		    goto L2;
		  }
		}
		if(currentch=='\n')
		{
		  Error(col,3);
		}
	  }
      else
	  {
	    string[1]='\0';
		retract();
		fprintf(fw,"<48\t");
		fprintf(fw,"->\n");
	  }
	}
	else
	{
	  if(currentch!=EOF)
	  {
		Error(col,1);
	  }
	}
  }while(currentch!=EOF);
  fclose(fd);
  fclose(fw);
  fclose(err);
  fclose(sign);
  fclose(nmsign);
  strout2[c2]='\0';         //源文件已经全部读入strout2中
  fw=fopen("token.txt","r");//把分析的结果读入strout1中，供输出用
  while(!feof(fw))
  {
    strout1[c1++]=fgetc(fw);
  }
  strout1[c1]='\0';
  fclose(fw);
  err=fopen("error.txt","r");//把错误信息读入error中，供输出用
  while(!feof(err))
  {
    error[c3++]=fgetc(err);
  }
  error[c3]='\0';
  fclose(err);
}

static void PrintWord()
{
  system("cls");
  cout<<"\t\t\tTHE CODE\n";
  cout<<strout2<<endl;
  cout<<"\t\t\tRESULT\n";
  cout<<error<<endl;
  cout<<strout1<<endl;
}

void main()
{
  startl();
  PrintWord();
  getch();
}