Flex and Bison C adder gives error after not executing expression - c

Question

I have been trying to implement a calculator, and I cannot find sources for this Flex, Bison and C program. I really can't tell what I'm doing wrong. Here are my files:
henry#FusionReactor:~/Downloads/YASPLANG$ ls
a.out compiler.output compiler.y lex.yy.c
compiler.l compiler.tab.c file.o README
My grammar:
henry#FusionReactor:~/Downloads/YASPLANG$ cat compiler.y
%{
#include <stdio.h>
void yyerror(const char *msg) {
fprintf(stderr, "%s\n", msg);
}
int yydebug = 1;
%}
%code requires
{
#define YYSTYPE double
}
/*
%union {
char c;
char *s;
double d;
}
*/
%define parse.error verbose
%define parse.lac full
%token NUM
%token PLUS
%token NEWLINE
%left PLUS
%%
answered: %empty {; }
| answered answer {; }//{ printf("%lg is answered.\n",$2); $$ = $1; }
;
answer: NEWLINE { $$ = 0.0; }
| expr NEWLINE { $$ = $1; }
;
expr: expr PLUS expr { $$ = $1 + $3; }//printf("%lg\n", $$ = $1 + $3); printf("Doing %lg + %lg.\n", $1, $3); }
| NUM { $$ = $1; }//printf("%lg\n", $$ = $1); printf("Found number: %lg.\n", $1); }
;
%%
And my lexer:
henry#FusionReactor:~/Downloads/YASPLANG$ cat compiler.l
%{
int yylex(void);
#include "compiler.tab.c"
%}
%%
[ \t] ;
'\n' ;//{ return NEWLINE; }
'+' { return PLUS; }
^[0-9]+(\.[0-9]+)? { sscanf(yytext, "%lf", &yylval); printf("%lf = %s\n", yylval, yytext); return NUM; }
%%
int main(const char **argv, int argc) {
return yyparse();
}
I compiled it with these commands (I keep everything in my Downloads, and the folder name is an acronym, let's not worry about that too much):
henry#FusionReactor:~/Downloads/YASPLANG$ bison --report=all --verbose --debug compiler.y
henry#FusionReactor:~/Downloads/YASPLANG$ flex compiler.l
henry#FusionReactor:~/Downloads/YASPLANG$ gcc lex.yy.c -lfl
When I execute and type "5 + 5 [NEWLINE]" twice, I get:
henry#FusionReactor:~/Downloads/YASPLANG$ ./a.out
Starting parse
Entering state 0
Reducing stack by rule 1 (line 26):
-> $$ = nterm answered ()
Stack now 0
Entering state 1
Reading a token: 5 + 5
5.000000 = 5
Next token is token NUM ()
Shifting token NUM ()
Entering state 3
Reducing stack by rule 6 (line 34):
$1 = token NUM ()
-> $$ = nterm expr ()
Stack now 0 1
Entering state 6
Reading a token: +5
5 + 5
5.000000 = 5
Next token is token NUM ()
LAC: initial context established for NUM
LAC: checking lookahead NUM: Err
Constructing syntax error message
LAC: checking lookahead $end: Err
LAC: checking lookahead NUM: Err
LAC: checking lookahead PLUS: S7
LAC: checking lookahead NEWLINE: S8
syntax error, unexpected NUM, expecting PLUS or NEWLINE
Error: popping nterm expr ()
Stack now 0 1
Error: popping nterm answered ()
Stack now 0
Cleanup: discarding lookahead token NUM ()
Stack now 0
henry#FusionReactor:~/Downloads/YASPLANG$
I am very puzzled, and if it would be possible to tell how and why it will not work, that would be very helpful.

Answer 1

In (f)lex, the pattern '+' matches a sequence of two or more ', because ' is just an ordinary character. If you want to match a single +, use "+" or \+ or [+].
Similarly, '\n' matches the three-character sequence 'ENTER'. Just use \n.
Finally, ^[0-9]+(\.[0-9]+)? will only match a number at the very beginning of a line, because the pattern starts with^. You want it to match anywhere, so lose the ^.
The reason no sensible error is reported is that (f)lex adds an implicit default rule which matches any single character and executes the ECHO action (which writes the character to stdout. That default rule is hardly ever what you want; I strongly recommend using
%option nodefault
to suppress it. You must then provide your own default rule, with a more sensible action.