How to know which alternative rule ANTLR parser is currently in during visit

Question

If we look at the bash sources, and the yacc grammar specifically, we can see that all redirections are defined as such :

redirection
        :   GREATER WORD
        |   LESS WORD
        |   NUMBER GREATER WORD
        |   NUMBER LESS WORD
        |   REDIR_WORD GREATER WORD
        |   REDIR_WORD LESS WORD
        |   GREATER_GREATER WORD
        |   NUMBER GREATER_GREATER WORD
        |   REDIR_WORD GREATER_GREATER WORD
        |   GREATER_BAR WORD
        |   NUMBER GREATER_BAR WORD
        |   REDIR_WORD GREATER_BAR WORD
        |   LESS_GREATER WORD
        |   NUMBER LESS_GREATER WORD
        |   REDIR_WORD LESS_GREATER WORD
        |   LESS_LESS WORD
        |   NUMBER LESS_LESS WORD
        |   REDIR_WORD LESS_LESS WORD
        |   LESS_LESS_MINUS WORD
        |   NUMBER LESS_LESS_MINUS WORD
        |   REDIR_WORD  LESS_LESS_MINUS WORD
        |   LESS_LESS_LESS WORD
        |   NUMBER LESS_LESS_LESS WORD
        |   REDIR_WORD LESS_LESS_LESS WORD
        |   LESS_AND NUMBER
        |   NUMBER LESS_AND NUMBER
        |   REDIR_WORD LESS_AND NUMBER
        |   GREATER_AND NUMBER
        |   NUMBER GREATER_AND NUMBER
        |   REDIR_WORD GREATER_AND NUMBER
        |   LESS_AND WORD
        |   NUMBER LESS_AND WORD
        |   REDIR_WORD LESS_AND WORD
        |   GREATER_AND WORD
        |   NUMBER GREATER_AND WORD
        |   REDIR_WORD GREATER_AND WORD
        |   GREATER_AND DASH
        |   NUMBER GREATER_AND DASH
        |   REDIR_WORD GREATER_AND DASH
        |   LESS_AND DASH
        |   NUMBER LESS_AND DASH
        |   REDIR_WORD LESS_AND DASH
        |   AND_GREATER WORD
        |   AND_GREATER_GREATER WORD
        ;

In my visitor, when visitRedirection is called this feels almost impossible to know which alternative the visitor is currently in easily.. I could label each alternatives using # and labels but adding 43 visit methods for just a single production rules seems somewhat excessive.

Usually I would just do some null checks by doing ctx.GREATER() != null to know if the first alternative was chosen, but in this example there's almost always 2 clashing alternatives, for instance :

GREATER WORD
NUMBER GREATER WORD

So should I do ctx.NUMBER() != null && ctx.GREATER() != null to match the second alternative and ctx.NUMBER() == null && ctx.GREATER() != null to match the first ?

Is there an easier or cleaner way to be able to know which specific alternative the visitor is currently in ?

user_1137174 · Answer

Restructure your grammar to have less alternatives. Many of them have common leading or trailing parts, for example:

redirection
        :   GREATER WORD
        |   LESS WORD
        |   NUMBER (GREATER | LESS) WORD
        |   REDIR_WORD (GREATER | LESS | LESS_LESS_MINUS) WORD
        |   ...

This way you have a unique first token in each alt, which you can then assign to a local variable:

redirection
        :   op = GREATER WORD
        |   op = LESS WORD
        |   op = NUMBER subOp= (GREATER | LESS) WORD
        |   op = REDIR_WORD subOp =(GREATER | LESS | LESS_LESS_MINUS) WORD
        |   ...

With that you can easily check in which alt you are in your listener/visitor:

public exitRedirection(RedirectionContext ctx) {
    switch (ctx.op.getType()) {
        case YourParser.GREATER_WORD: {
            break;
        }

        case YourParser.REDIR_WORD: {
            switch (ctx.supOp.getType()) {
                case YourParser.LESS_LESS_MINUS: {
                    break;
                }
            }
            break;
        }
    }

How to know which alternative rule ANTLR parser is currently in during visit

1 Answers