New Topic/Question
Reply
that is to construct lexer and parser for my own language.......
it is giving error and system will hang when i remove ? in output statement(just like teminal symbol ';' in c)..
how to construct next phase of compiler design i,e SDT(syntax directed translation)...
/
#include<stdio.h>
#include<ctype.h>
#include<stdlib.h>
FILE *fp;
char delim[10]={' ','\t','\n',',','{','}','[',']','(',')'};
char oper[5]={'+','-','*','/','%'};
char lop[3]={'<','>','#'};
char op[9]={'+','-','*','/','%','<','>','!','='};
char key[14][12]={"Integer","Real","Char","If","Else","Repeat","Upto","Inc","Begin","End","Input","Read","Output","Write"};
void skipcomment();
void analyze();
void checkwrite();
void check(char []);
int isdelim(char);
int isend(char c);
int isop(char);
int operator();
void apara();
void ipara();
void rpara();
int fop=0,numflag=0,f=0,k=0,y=0,z=0,flag,fop1=0;
char c,ch,sop,input[100];
char token[50];
int main()
{
char fname[12];
printf("\n enter filename:");
scanf("%s",fname);
fp=fopen(fname,"r");
if(fp==NULL)
printf("\n ERROR : The File Does Not Exist");
else
analyze(); //scan the input file
printf("\t%s\n",input); //print the output terminal array
//printf("\tk=%d\n",k);
return 0;
}
/*------ANALYZE THE INPUT ---------------*/
void analyze()
{
long int var;
while(!feof(fp))
{
c=getc(fp); //take character input
if(c=='!')
{
skipcomment();
}
else if(isalpha(c)) //to check if alphabet
{
if(numflag==1)
{
token[z]='\0'; //numflag=1 signifies a number
check(token);
z=0;
}
else
{
token[z]=c; //create token
z++;
}
if(f==0)
f=1; //f=1 signifies an identifier
}
else if(isalnum(c))
{
if(numflag==0) //to check if alphanumeric
numflag=1;
token[z]=c;
z++;
}
else
{
if(isend(c))
{ // to check if end of line character
input[k++]='s';
}
else if(isdelim(c))
{
if(numflag==1)
{
token[z]='\0';
check(token);
numflag=0;
}
if(f==1) //to check if identifier
{
token[z]='\0';
numflag=0;
check(token);
}
z=0;
f=0;
}
else if(isop(c))
{ // to check if operator
if(numflag==1)
{
token[z]='\0';
check(token);
z=0;
operator();
}
else if(f==1)
{
token[z]='\0';
z=0;
f=0;
numflag=0;
check(token);
operator();
}
if(fop==1)
{
fop=0;
sop='\0';
}
else if(fop1==1)
{
fop1=0;
}
}
else if(c=='.')
{
token[z]=c;
z++;
}
}
}
}
/*--------------- TO CHECK IF OPERATOR-------------------------- */
int isop(char c)
{
for(y=0;y<9;y++)
{
if(c==op[y]){return 1;}
}
}
/*--------- TO CHECK FOR END OF LINE ' ? '----------- */
int isend(char c)
{
if(c=='?')
{
ch=token[z-1];
if(isdigit(ch))
{
token[z]='\0';
z=0;
check(token);
return 1;
}
else if(isalpha(ch))
{
token[z]='\0';
z=0;
check(token);
return 1;
}
/* IF IF END OF LINE RETURN 1*/
return 1;
}
/* OTHER WISE RETURN 0 */
else return 0;
}
/*----- TO CHECK IF DELIMITOR--------*/
int isdelim(char c)
{
int i=0,j=0,f1,f2;
char str[20],temp1[6]={"Enter"},temp2[5]={"Exit"};
if(c=='[')
{
while((str[j]=getc(fp))!= ']')
j++;
str[j]='\0';
/* compare with 'Enter' */
while(i<j)
{
if (temp1[i] != str[i])
{
f1=1;
break;
}
else
i++;
}
if(f1!=1)
input[k++]='u'; /*---TOKEN FOR [enter] -----*/
i=0;
while(i<j)
{
if (temp2[i] != str[i])
{
f2=1;
break;
}
else
i++;
}
if(f2!=1)
input[k++]='z';/*------ TOKEN FOR [exit] -----*/
}
for(i=0;i<10;i++)
{
/* IF CHARACTER IS DELIMITOR RETURN 1 */
if(c==delim[i])
return 1;
}
return 0;
}
/*---------------- TO GET OPERATOR TOKEN---------*/
int operator()
{
int i,j;
ch=getc(fp);
/* TO CHECK FOR = and == */
if(c=='=')
{
if(ch=='=')
{ fop=1;
sop=ch;
input[k++]='l'; /*-- TOKEN FOR == */
return 1;
}
fop1=1;
input[k++]='a';/*---- TOKEN FOR = */
ungetc(ch,fp);
return 1;
}
/*---- TO CHECK FOR ARITHMETIC OPERATOR */
for(i=0;i<5;i++)
{
if(c==oper[i])
{
if(c=='-'&& ch=='-')
{
fop=1;
sop=ch;
input[k++]='a'; /* TOKEN FOR -- */
return 1;
}
else if(c=='+'&&ch=='+')
{
fop=1;
sop=ch;
input[k++]='a';/* TOKEN FOR ++ */
return 1;
}
else if(ch=='=')
{
fop=1;
sop=ch;
input[k++]='a';/* TOKEN FOR +|-|/|* = */
return 1;
}
else
{
fop1=1;
input[k++]='a';
ungetc(ch,fp);
return 1;
}
}
}
/* FOR LOGICAL OPERATORS */
for(i=0;i<3;i++)
{
if(c==lop[i])
{
if(ch=='=')
{
fop=1;
sop=ch;
input[k++]='l'; /* TOKEN FOR <= | >= | #= */
return 1;
}
else
{ fop1=1;
input[k++]='l';/* TOKEN FOR > | < | # */
ungetc(ch,fp);
return 1;
}
}
}
return 0;
}
/*---- FUNCTION TO SKIP THE COMMENT ---------*/
void skipcomment()
{
ch=getc(fp);
while(f==0)
{
ch=getc(fp);
if(ch=='!')
f=1;
}
f=0;
input[k++]='c';
}
/*----TO ASSIGN TOKEN NAMES TO KEYWORDS, NUMBERS AND IDENTIFIERS --*/
void check(char t[])
{
int i,count=0;
if(numflag==1)
{
input[k++]='n';
numflag=0;
return;
}
else if(strcmp(t,"Begin")==0)
{
input[k++]='b';
}
else if(strcmp(t,"End")==0)
{
input[k++]='f';
}
else if(strcmp(t,"Upto")==0)
{
input[k++]='t';
}
else if(strcmp(t,"Inc")==0)
{
input[k++]='j';
}
else if(strcmp(t,"Repeat")==0)
{
input[k++]='y';
}
else if(strcmp(t,"Else")==0)
{
input[k++]='e';
}
else if(strcmp(t,"Input")==0)
{
flag=1;
rpara();
}
else if(strcmp(t,"Read")==0)
{
apara();
}
else if(strcmp(t,"Integer")==0)
{
ipara();
}
else if(strcmp(t,"Char")==0)
{
ipara();
}
else if(strcmp(t,"Real")==0)
{
ipara();
}
else if(strcmp(t,"If")==0)
{
input[k++]='x';
}
else if(strcmp(t,"Output")==0)
{
checkwrite();
}
else if(strcmp(t,"Write")==0)
{
apara();
}
else
{
input[k++]='v';
f=0;
}
}
/*-------TO ASSIGN TOKEN NAMES TO ARRAY io FUNCTION -----*/
void apara()
{
int count=0;
c=getc(fp);
while(1)
{
z=0;
count++;
while(c!=',')
{
if(isalpha(c))
{
token[z++]=c;
}
c=getc(fp);
if(c=='?')
break;
}
if(c=='?')
break;
if(count>2)
{
printf("Error : Invalid Number Arguments In Array Function\n");
return;
}
c=getc(fp);
}
input[k++]='i';
input[k++]='s';
count=0;
}
/*----- TO ASSIGN TOKEN NAMES TO INPUT FUNCTIONS-------*/
void ipara()
{
int count=0;
c=getc(fp); /* Real a,b,b? */
while(1)
{
z=0;
count++;
while(c!=',')
{
if(isalpha(c))
{
token[z++]=c;
}
c=getc(fp);
if(c=='?')
{
flag=1;
break;
}
}
if(c=='?')
{
flag=1;
break;
}
c=getc(fp);
}
if(flag==1) input[k++]='d';
else
{
printf("Error : Missing End Of Statement ' ? ' \n");
return;
}
count=0;
}
void rpara()
{
int count=0;
c=getc(fp);
while(1)
{
z=0;
count++;
while(c!=',')
{
if(isalpha(c))
{
token[z++]=c;
}
c=getc(fp);
if(c=='?')
break;
}
if(c=='?')
break;
c=getc(fp);
}
if(flag==1) input[k++]='r';
input[k++]='s';
count=0;
}
/*--- TO ASSIGN TERMINAL VALUES FOR 'OUTPUT' FUNCTION -------*/
void checkwrite()
{
int count=0;
c=getc(fp);
while(1)
{
if(c=='"')
{
count++;
if(count>2){printf("Error : ' Output ' Syntax Error\n");return;}
c=getc(fp);
while(c!='"')
{
c=getc(fp);
if(c=='?'){printf("Error ; ' Output ' Syntax Error\n");return;}
}
}
else if(c==' ') c=getc(fp);
else if(count==0)
{
printf("Error ; ' Output ' Syntax Error\n");
return;
}
if(c=='"') {count++; c=getc(fp);}
if(c=='?') break;
}
input[k++]='w';
input[k++]='s';
}
My language specification is as fallows...
[Enter]
Real q,b,u?
Read a,b?
Write a,n?
Read s,d?
Input "hghbghhjb"?
Output "uuhu"?
If v<f
Begin
Output "jhsdbfjh"?
Repeat j=67 Upto t Inc 56
Begin
Output "hghg"?
End
End
[Exit]
MultiQuote
| 