Yacc parser for nested while loop - parsing

Question

i was trying to code a parser using yacc and lex that count the number of nested loops (while or for).I started the implementation for just while loops.But for some reason the parser gives me an error at the end of a closing brace.
Here is the code.
%{
#include<stdio.h>
/*parser for counting while loops*/
extern int yyerror(char* error);
int while_count=0;
extern int yylex();
%}
%token NUMBER
%token VAR
%token WHILE
%%
statement_list : statement'\n'
| statement_list statement'\n'
;
statement :
while_stmt '\n''{' statement_list '}'
| VAR '=' NUMBER ';'
;
while_stmt :
WHILE '('condition')' {while_count++;}
;
condition :
VAR cond_op VAR
;
cond_op : '>'
| '<'
| '=''='
| '!''='
;
%%
int main(void){
yyparse();
printf("while count:%d\n",while_count);
}
int yyerror(char *s){
printf("Error:%s\n",s);
return 1;
}
what is wrong with that code.And is there a way in yacc to mention optional arguments? like the "\n" after while?
here is the lexer code
%{
#include"y.tab.h"
/*lexer for scanning nested while loops*/
%}
%%
[\t ] ; /*ignore white spaces*/
"while" {return WHILE;}
[a-zA-Z]+ {return VAR;}
[0-9]+ {return NUMBER;}
'$' {return 0;}
'\n' {return '\n' ;}
. {return yytext[0];}
%%
VAR is a variable name with just ascii characters and WHILE is the keyword while.type is not taken into consideration on variable assignments

Answer 1

The problem you seem to be having is with empty loop bodies, not nested loops. As written, your grammar requires at least one statement in the while loop body. You can fix this by allowing empty statement lists:
statement_list: /* empty */
| statement_list statement '\n'
;
You also ask about making newlines optional. The easiest way is to make the lexer simply discard newlines (as whitespace) rather than returning them. Then just get rid of the newlines in the grammar, and newlines can appear between any two tokens and will be ignored.
If you really must have newlines in the grammar for some reason, you can add a rule like:
opt_newlines: /* empty */ | opt_newlines '\n' ;
and then use this rule wherever you want to allow for newlines (replace all the literal '\n' in your grammar.) You have to be careful not to use it redundantly, however. If you do something like:
statement_list: /* empty */
| statement_list statement opt_newlines
;
while_stmt opt_newlines '{' opt_newlines statement_list opt_newlines '}'
you'll get shift/reduce conflicts as newlines before the } in a loop could be either part of the opt_newlines in the while or the opt_newlines in the statement_list. Its pretty easy to deal such conflicts by just removing the redundant opt_newlines.