Lex and Yacc to make compiler? - yacc

Question

I am starting a toy compiler, and I am making the simplest thing I can imagine, but it won't work.
Lex compiles, and Yacc compiles, and they link together, but the outputted program does not do what I expected.
Lex:
%{
#include <stdlib.h>
void yyerror(char *);
#include "y.tab.h"
%}
%%
a {
yylval = atoi(yytext);
return AAA;
}
. yyerror("invalid character");
%%
int yywrap(void) {
return 1;
}
Yacc:
%{
void yyerror(char *);
int yylex(void);
int sym[26];
#include <stdio.h>
%}
%token AAA
%%
daaaa:
AAA {printf("%d\n", $1);}
%%
void yyerror(char *s) {
fprintf(stderr, "%s\n", s);
}
int main(void) {
yyparse();
return 0;
}
The program I am trying to compile with this compiler is a file containing: a. that's it.
I don't know what's happened!
Clarification: What I expected the compiled compiler to do was to accept a file into it, process the file, and spit out a compiled version of that file.

Answer 1

Can you explain, maybe in an answer, exactly what you did, and how it worked, because as far as I can tell, and as far as I have tested the question, it shouldn't work as you say.
I took your code verbatim, creating files grammar.y and lexer.l. I then compiled the code. I'm working on Mac OS X 10.11.4, using GCC 6.1.0, Bison 2.3 (disguised as yacc) and Flex 2.5.35 (disguised as lex).
$ yacc -d grammar.y
$ lex lexer.l
$ gcc -o gl y.tab.c lex.yy.c
$ ./gl <<< 'a'
0
$
I subsequently made two changes. In grammar.y, I changed main() to:
int main(void) {
#if YYDEBUG
yydebug = 1;
#endif
yyparse();
return 0;
}
and in lexer.l, I changed the default character rule to:
\n|. yyerror("invalid character");
(The . doesn't match newline, so the newline after the a in the input was echoed by default in the original output.)
With a similar compilation, the output becomes:
$ ./gl <<< 'a'
0
invalid character
$
With the compilation specifying -DYYDEBUG too:
$ gcc -DYYDEBUG -o gl lex.yy.c y.tab.c
$
the output includes useful debugging information:
$ ./gl <<< 'a'
Starting parse
Entering state 0
Reading a token: Next token is token AAA ()
Shifting token AAA ()
Entering state 1
Reducing stack by rule 1 (line 12):
$1 = token AAA ()
0
-> $$ = nterm daaaa ()
Stack now 0
Entering state 2
Reading a token: invalid character
Now at end of input.
Stack now 0 2
Cleanup: popping nterm daaaa ()
$ ./gl <<< 'aa'
Starting parse
Entering state 0
Reading a token: Next token is token AAA ()
Shifting token AAA ()
Entering state 1
Reducing stack by rule 1 (line 12):
$1 = token AAA ()
0
-> $$ = nterm daaaa ()
Stack now 0
Entering state 2
Reading a token: Next token is token AAA ()
syntax error
Error: popping nterm daaaa ()
Stack now 0
Cleanup: discarding lookahead token AAA ()
Stack now 0
$
The second a in the input correctly triggers a syntax error (it isn't allowed by the grammar). Other characters are permitted, generate a 'invalid character' message, and are otherwise ignored (so ./gl <<< 'abc' generates 3 invalid character messages, one for the b, one for the c, and one for the newline).
Changing the assignment to yylval in lexer.l to:
yylval = 'a'; // atoi(yytext);
changes the number printed from 0 to 97, which is the character code for 'a' in ASCII, ISO 8859-1, Unicode, etc.
I've been using a here string as the source of data. It would be equally feasible to have used a file as the input:
$ echo a > program
$ cat program
a
$ ./gl < a
Starting parse
Entering state 0
Reading a token: Next token is token AAA ()
Shifting token AAA ()
Entering state 1
Reducing stack by rule 1 (line 12):
$1 = token AAA ()
97
-> $$ = nterm daaaa ()
Stack now 0
Entering state 2
Reading a token: invalid character
Now at end of input.
Stack now 0 2
Cleanup: popping nterm daaaa ()
$
If you want to read files specified by name on the command line, you have to write more code in main() to process those files.

Answer 2

The program does not accept a file because it was not told to.
In the Yacc program,
extern FILE *yyin; must be added in the definitions section.
I believe that's it.