Re: Help with a simple grammar I can't get to work, please

[Kelly Leahy]

I think your rule for the grammar should use END in place of the '\n' in 
endline.  The problem is that the lexical analyzer sends an END, but the 
grammar looks for a '\n', IMO.

Kelly

----- Original Message ----
From: Dustin Robert Kick <address@hidden>
To: address@hidden
Cc: Dustin Robert Kick <address@hidden>
Sent: Tuesday, September 26, 2006 8:41:17 AM
Subject: Help with a simple grammar I can't get to work, please

I'm trying to make what I thought would be the simplest grammar I  
could specify work before moving on to more productive programs, sort  
of a hello bison.  All it does is print some output with the input  
token passed to the grammar, but it works on every other line being  
input.  Here is some sample output with debugging enabled:

__lex_yacc test copy__08:29:33__BRUNOISE-TIGRE__./fixtest
Starting parse
Entering state 0
Reading a token: hall
matched string:hall
Next token is 257 (STRING)
Shifting token 257 (STRING), Entering state 1
Reading a token: matched \n
Next token is 258 (END)
Shifting token 258 (END), Entering state 2
Reducing via rule 1 (line 22), STRING END  -> yowza
YOWZA hall
: 5
state stack now 0
Entering state 3
Reading a token: hello
matched string:hello
Next token is 257 (STRING)
parse error:hello
Error: state stack now 0
Starting parse
Entering state 0
Reading a token: matched \n
Next token is 258 (END)
parse error:hello

Starting parse
Entering state 0
Reading a token: test
matched string:test
Next token is 257 (STRING)
Shifting token 257 (STRING), Entering state 1
Reading a token: matched \n
Next token is 258 (END)
Shifting token 258 (END), Entering state 2
Reducing via rule 1 (line 22), STRING END  -> yowza
YOWZA test
: 5
state stack now 0
Entering state 3
Reading a token: ^C

the parser continues with this pattern of parsing the way I expect on  
the odd parses, and outputting a syntax error on the evens.  I can't  
figure out why this won't work, can someone explain?  Here are the  
source files and .output file.

<lex_flex("rules.l")>
%{
#include "grammar.tab.h"
#include <stdlib.h>

%}

%%

.* {yylval.string = yytext;
    printf("matched string:%s\n", yytext);
    return STRING;}

\n {printf("matched \\n\n");
    return END;}

%%

int yywrap (void)
{
return 1;
}
</lex_flex("rules.l">

<bison("grammar.y")>

#include <libc.h>
#include <math.h>
%}


%union
{
    int label;
    double dval;
    char * string;
}

%token <string> STRING

%% /* beginning of rules section */

yowza : STRING endline {printf("YOWZA %s : %d\n", $1, strlen($1));}
;

endline : '\n' {printf("AND IN THE END!\n");}
;

%% /* beginning of functions section */
void yyerror(char *s)
{
        printf ("%s:%s\n", s, yylval.string);
        fflush(stdout);
}

int main(int argc, char **argv)
{
    while (!feof(stdin))
    {
        yyparse();
    }
    exit(0);
}
</bison("grammar.y">

<"grammar.output">

Grammar
rule 1    yowza -> STRING END

Terminals, with rules where they appear

$ (-1)
error (256)
STRING (257) 1
END (258) 1

Nonterminals, with rules where they appear

yowza (5)
     on left: 1


state 0

     STRING    shift, and go to state 1

     yowza    go to state 3



state 1

     yowza  ->  STRING . END   (rule 1)

     END     shift, and go to state 2



state 2

     yowza  ->  STRING END .   (rule 1)

     $default    reduce using rule 1 (yowza)



state 3

     $       go to state 4



state 4
<"grammar.output">

Thanks for any effort spent on this.


Dustin Kick


_______________________________________________
address@hidden http://lists.gnu.org/mailman/listinfo/help-bison