Question

Hi I have a small problem in my ANTLR tree grammar. I am using ANTLRWorks 1.4. In parser grammar I have the rule like this:

declaration
:       'variable' IDENTIFIER ( ',' IDENTIFIER)* ':' TYPE ';'
->    ^('variable' IDENTIFIER TYPE)+

So I wanted one tree per each IDENTIFIER.

And in the tree grammar I left only rewrite rules:

declaration
:     ^('variable' IDENTIFIER TYPE)+

But when I check grammar I got syntax error unexpected token +. And it is this + sign at the end of the declaration rule in the tree grammar. So what I am doing wrong?

Parser grammar works fine and builds AST tree as expected. I generated lexer and parser for C# and test it for some input.

Answer 1

When parsing the source:

variable a, b, c : int;

you’re trying to construct an AST that looks like:

variable variable variable
    /        |        \
   a         b         c
  /          |          \
int         int         int

But since 'variable' and TYPE are always the same token, I see no need to create all those duplicate nodes. Why not just do:

declaration
  :  'variable' IDENTIFIER ( ',' IDENTIFIER)* ':' TYPE ';' 
     -> ^('variable' TYPE IDENTIFIER+)
  ;

which will create an AST like:

 variable
  / | | \
int a b  c

?