Question

I have a parser generated using Flex/Bison – it parses each line of a file and returns some output for that line. My input data is a little corrupt, and what I’d like to do is add a special character (like a #) to the beginning of lines that I’d like bison to ignore, and merely echo the line to the output.

So if my input looked like

apples 3 ate
oranges 4 consumed
# rhino ten

my output after parsing the lines might be

I ate three apples
I consumed four oranges
# rhino ten

Is there some easy way to do this ?

Answer 1

You can do this lexically in the flex scanner.

Something like:

^#.*\n   { fputs(yytext, stdout); /* increment line number */ }

Or in the parser:

^#.*\n   { yystype.lexeme = strdup(yytext);
           return HASH_ECHO; /* token type defined in parser */ }

In the parser just generate from your top level grammar a production for this:

/* in top section */
%union {
   /* ... */
   char *lexeme;
   /* ... */
}

%token<lexeme> HASH_ECHO
/*...*/

/* make sure this rule is hooked into your grammar, of course */
hash_echo : HASH_ECHO { fputs($1, stdout); free($1); }
          ;

Not sure about including that newline; I have no idea how you’re handling those. So it may not be appropriate.