With thanks to tye and jdporter, the length method has been rewritten as follows:

sub length
{
    my ($self, $uri);
    my $length = 1;
    
    if (ref($_[0]))
    {
        $self = shift;
        $uri = $self->{uri};
    }
    else
    {
        $uri = shift;
    }
    
    while ($uri =~ m/((?:\{[^\}]+\})|(?:\[[^\]]+\]))/ig)
    {
        if    ($1 =~ m/\{([^\}]+)\}/i)
        {
            my $temp = $1;
            
            $length *= ($temp =~ tr/,/,/);
        }
        elsif ($1 =~ m/\[([^\]]+)\]/i)
        {
            my ($start, $stop) = sort split('-',$1);
            $stop = $start+10 unless ($stop);
            
            if ($start =~ /\d+/)
            {    #There is one or more digit, so we'll remove anything
+ else that might be there
                $start =~ s/\D+//g;
                $stop =~ s/\D+//g;
            }
            
            $length *= $stop - $start + 1;
        }
    }
        
    return $length;
}
[download]

However, this still isn't entirely perfect. These are some examples from my current log that generate inconsistent results:

scalar($uriobj->as_list)
        $uriobj->length
                $uri
1       0       foo{011-047}bar
0       -398    foo[800-1199]bar
0       -348    foo[850-1199]bar
0       -298    foo[900-1199]bar
0       50      foo[001-50]bar
0       -97     foo[2-100]bar
0       -97     foo[2-100]bar
0       -7      foo[2-10]bar
1       0       foo{001-188}bar
0       -140    foo[01-20]bar
0       -3      foo[1-05]bar
0       -46     foo[53-100]bar
975     -23     foo[001+-+025]bar
9       6       foo{one,two,three}/[01-03]bar
27      6       foo{one[01-03],two,three}/[01-03]bar
0       -120    foo[2-13]/[01-12]bar
0       -120    foo[2-13]/[1-12]bar
0       -90     foo[2-13]/[1-9]bar
0       101     foo[0000-100]bar
0       41      foo[0000-40]bar
0       -10     foo[8-19]bar
0       -399    foo[600-1000]bar
1       0       foo{011-047}bar
[download]