Re: multiple entry points in yacc grammer

nbhatia@netcom.com (Naresh Bhatia) writes:


>Can a yacc grammar have multiple entry points? For example, if I have
>a grammar that can parse C++ method declarations, can I use it to
>parse standalone method parameters? The knowledge of parsing the
>parameters is obviously embedded in the method parser, so how to use
>this knowledge to parse the parameters alone?
>[No, but you can fake it. (...). -John]


This kludge may help. I did it because in a way to handle C
instructions OR declarations at any point, there is a shift/reduce
conflict ( 'identifier;' can be a valid declaration or instruction )


At the time where you know which entry the parser must take,
make the lexer return the pseudo-token FAKEn


%{


#undef yylex()
#define yylex() my_lex()


int fake_token;


%}


%token FAKE1 FAKE2 FAKE3
%start start


%%


start: FAKE1 rule_1
          | FAKE2 rule_2
          | FAKE3 rule_3
          ;


/* those are your multiple entry points :: */


rule_1: .... ;
rule_2: .... ;
...


%%




#undef yylex


int my_lex()
{
          int ret;


          if( fake_token )
          {
                      ret = fake_token;
                      fake_token = 0;


                      return ret;
          }


          return yylex();
}


Hope this helps,


[Yup, that's how you fake it. -John]
--

Return to the comp.compilers page.
Search the comp.compilers archives again.