write ANTLR regex when declaring variables list

Question

I've wrote a grammar rule for a language in ANTLR as below:

variable: idlist COLON type (EQUAL explist)? SEMI;
idlist: identifier (COMMA identifier)*;
explist: exp (COMMA exp)*;

COLON: ':';
EQUAL: '=';
SEMI: ';';
COMMA: ',';

This input is valid for above grammar:

a, b, c: integer = 3, 4, 6;

But now if I want this input:

a, b, c, d: integer = 3, 4, 6;

or this:

a, b, c: integer = 3, 4, 6, 1;

becomes invalid due to inequality between amount of ID in idlist and value in explist, how I rewrite my grammar? Tks

Answer 1

Don't let your grammar handle this semantic task. The syntax is correct in all those cases. The constraint that the number of left-hand-values and right-hand-values must be equal is a semantic rule you should enforce in the semantic phase, following the parse step (the syntactic phase). This is usually done by evaluating the generated parse tree.

Answer 2

I agree with the previous answer that the proper way to do it would be later during the evaluation of the generated parse tree/AST.

But if you insist, here is an example of the grammar which matches the number of declarations and initializations. Please note that it is more an exercise in demonstrating the power of context-free grammar than something I would be glad to find in the production code.

grammar Test;

list: IDENTIFIER middle  exp ';';

middle :
   | ',' IDENTIFIER middle exp ','
   | ':' type '='
;

exp:
   NUMBER
;

type:
    IDENTIFIER
;

NUMBER : [0-9]+;
IDENTIFIER : [a-z]+;
COLON: ':';
EQUAL: '=';
SEMI: ';';
COMMA: ',';

In case you have a mismatch in the list sizes the error message is confusing:

line 1:11 mismatched input ',' expecting ';'

This is why I would not recommend using this approach.