CODEWITHSUNDEEP
parsingyacc
statguy
statguy

Reputation: 1717

How do I format a grammar rule that requies two tokens to be equal?

How can I write a Yacc grammar that matches two tokens? For example:

START some_random_id
stuff stuff stuff
END some_random_id

I would like to make the requirement that the some_random_id matches at both places to match the entire block. So it would be something like:

block <- START ID block_body END ID

with the additional requirement that both IDs are equal.

Upvotes: 1

Views: 59

Answers (1)

Gene
Gene

Reputation: 47020

So long as some_random_id is drawn from a set of arbitrary size, this is not possible to do with grammar rules alone. There is a classic mathematical proof of this. You can only do it with parser action code that checks whether the ids are the same. But this is not very hard. Define the yylval union to have a field id_string that is filled in by the scanner. Then you'll have something like:

%union {
  char *id_string;
  ...
} 
%token <id_string> ID KW_START KW_END 
%%
...
block : KW_START ID stuff KW_END ID  { 
    if (strcmp($2.id_string, $5.id_string) != 0) YYERROR; }

Upvotes: 3

Related Questions