I have a question regarding subrules within grammars used with Parse::RecDescent - The following code best illustrates the best test case that I can identify which highlights my problem.

#!/usr/bin/perl

use Parse::RecDescent;
use Test::More 'no_plan';

my $grammar = q
{
argument:       element ((comma element)(s))(?)

element:        /[\w.-_]+/

comma:          ','
};

my $sql = Parse::RecDescent->new( $grammar );
while ( chomp( my $test = <DATA> ) )
{
    my ( $result, $statement ) = split /,/, $test, 2;
    ok( defined $sql->argument( $statement ) == $result, $test );
}

__DATA__
0,
1,arg1
0,arg1,
1,arg1,arg2
0,arg1,arg2,
1,arg1,arg2,arg3
[download]

Note that with this test code, each of the lines within the __DATA__ section are tested against the grammar and the expected result, either pass or fail, is indicated by the 1 or 0 at the start of the line respectively.

However, despite these expected results, the grammar does not reject the lines arg1, and arg1,arg2,, which ideally should be rejected due to the incomplete match of the subrule (comma element) - In these cases, a trace shows that the subrule terminal comma is matched and consumed, despite the entire subrule, consisting of comma and element, not being matched. I am sure this is a relatively straight-forward oversight within the grammar on my part, but I am at a loss as to how to correct this.

 

perl -le 'print+unpack("N",pack("B32","00000000000000000000001000110011"))'

Edited 2003-02-22 by Ovid