Is this an ambiguity in antlr4 grammar? - parsing

Question

I am trying to parse input like this:
(0002,0980);
(000a,f987);
(0001,[foo]00);
The pattern is ( g, e);
where g is a four-digit hex number. If g is even, e is a four-digit hex number either even or odd. If g is odd, e has the pattern '[IDENT] hex-digit hex-digit'.
I have tried many variations but this summarizes my thinking...
grammar Post;
script : statement (statement)* EOF ;
statement : tag ';' ;
tag : even_tag | odd_tag ;
even_tag : '(' g_even ',' e_even ')' ;
odd_tag : '(' g_odd ',' e_odd ')' ;
g_even : HEXDIGIT HEXDIGIT HEXDIGIT EVEN_HEXDIGIT ;
g_odd : HEXDIGIT HEXDIGIT HEXDIGIT ODD_HEXDIGIT ;
e_even : HEXDIGIT HEXDIGIT HEXDIGIT HEXDIGIT ;
e_odd : '[' IDENT ']' HEXDIGIT HEXDIGIT ;
HEXDIGIT : ODD_HEXDIGIT | EVEN_HEXDIGIT ;
ODD_HEXDIGIT : ['1','3','5','7','9', 'b', 'B', 'd', 'D', 'f', 'F'];
EVEN_HEXDIGIT : ['0','2','4','6','8', 'a', 'A', 'c', 'C', 'e', 'E'];
IDENT : LETTER (LETTER | DIGIT | ' ')*;
fragment LETTER : ('a'..'z' | 'A'..'Z') ;
fragment DIGIT : ('0'..'9');
It fails with the un-inspiring error
line 2:12 token recognition error at: '\n'
line 3:4 no viable alternative at input '(0001'
Modifying this to
grammar P2;
script : statement (statement)* EOF ;
statement : tag ';' ;
tag : even_tag | odd_tag ;
even_tag : '(' g_even ',' e_even ')' ;
odd_tag : '(' g_odd ',' e_odd ')' ;
g_even : G_EVEN ;
g_odd : G_ODD ;
e_even : E_EVEN ;
e_odd : E_ODD ;
G_EVEN : HEXDIGIT HEXDIGIT HEXDIGIT EVEN_HEXDIGIT ;
G_ODD : HEXDIGIT HEXDIGIT HEXDIGIT ODD_HEXDIGIT ;
E_EVEN : HEXDIGIT HEXDIGIT HEXDIGIT HEXDIGIT ;
E_ODD : '[' IDENT ']' HEXDIGIT HEXDIGIT ;
ODD_HEXDIGIT : ['1','3','5','7','9', 'b', 'B', 'd', 'D', 'f', 'F'];
EVEN_HEXDIGIT : ['0','2','4','6','8', 'a', 'A', 'c', 'C', 'e', 'E'];
HEXDIGIT : ODD_HEXDIGIT | EVEN_HEXDIGIT ;
IDENT : LETTER (LETTER | DIGIT | ' ')*;
fragment LETTER : ('a'..'z' | 'A'..'Z') ;
fragment DIGIT : ('0'..'9');
helps alot but the problem looks more clearly like an ambiguity in e.
line 2:5 mismatched input ',098' expecting ','
line 2:12 token recognition error at: '\n'
I suspect the problem results from the fact that g_even and e_even are ambiguous and g_odd and e_even are ambiguous. However, the pattern is such that this ambiguity can be avoided because g is always parsed first and g_even and g_odd are not ambiguous. Once g is known, there is no ambiguity left. There is only ambiguity if the parser doesn't know it is always looking for g first. There is only ambiguity if parsing might begin with e and that is never the case.
Perhaps the problem isn't ambiguity at all. I'm new to this game.
How can I parse this so that the parse tree labels g_even, g_odd, e_even, e_odd?
Thanks!

Answer 1

I got this to work with the following:
grammar P2;
script : statement (statement)* EOF ;
statement : tag ';' ;
tag : even_tag | odd_tag ;
even_tag : '(' g_even ',' e_even ')' ;
odd_tag : '(' g_odd ',' e_odd ')' ;
g_even : G_EVEN ;
g_odd : G_ODD ;
e_even : G_EVEN | G_ODD;
e_odd : E_ODD ;
G_EVEN : HEXDIGIT HEXDIGIT HEXDIGIT EVEN_HEXDIGIT ;
G_ODD : HEXDIGIT HEXDIGIT HEXDIGIT ODD_HEXDIGIT ;
//E_EVEN : HEXDIGIT HEXDIGIT HEXDIGIT HEXDIGIT ;
E_ODD : '[' IDENT ']' HEXDIGIT HEXDIGIT ;
ODD_HEXDIGIT : [13579bBdDfF];
EVEN_HEXDIGIT : [02468aAcCeE];
HEXDIGIT : ODD_HEXDIGIT | EVEN_HEXDIGIT ;
IDENT : LETTER (LETTER | DIGIT | ' ')*;
fragment LETTER : ('a'..'z' | 'A'..'Z') ;
fragment DIGIT : ('0'..'9');
There was no need to define the overlapping E_EVEN. Plus, the list-of-charcters-to-match syntax was wrong and caused unintended matching on commas. Doh.