Tukukkk

I wrote a Perl 6 grammar to parse a C++ function. My final goal would be to parse an entire header. The aim is not to correct C++ syntax errors, but to parse valid C++.

Do you have some advice or improvements?

#!/usr/bin/env perl6



grammar FUNCTION {

    token TOP       { [ <attr> s+ ]? <type> [ s* <type_mod> ]? 

                        (s+) <fname> (s*) "("

                        (s*) [<parameter> [ "," (s*) <parameter> ]* ]? (s*) 

                        ')'(s*) ';'  

                    }



    token name      { w+ }

    token namespace { [ "::" ]? [ <name> "::" ]* }

    token attr      { <name>  }

    token type      {  <namespace>? <name> }

    token type_mod  { [ *|& ]+ }

    token fname     { <name> }

    token variable  { <name> }

    token parameter { <type> [s* <type_mod> ]? s+ <variable> }

}



my $str = "const ::one::std::string ** ma1n( int&& i, two::std::string va1e_ );";

my $parsed = FUNCTION.parse($str);

say $parsed;

I'd expect to see many more tests of any program that addresses a problem this gnarly.

Choosing a couple I've recently had cause to write (on Stack Overflow), I would immediately add

• 
  

  void (SENDER::*get_func())(double, double);
  
  

  
  
  (from Passing pointers to member function as returned values to QObject::connect())


• 
  

  constexpr std::size_t len(const T(&)[length]);
  
  

  
  
  (from Differentiate between Array and Pointer in function parameter)

Neither of these succeeded when I tried them.