!/usr/bin/env perl

use 5.014;
use warnings;
use Benchmark 'cmpthese';

my @data = (
    'XX', 'XY', 'X<ph />X', 'X<ph />Y',
    'X<bpt x="y">X', 'X<bpt x="y">Y',
    'X<ept x="z">X', 'X<ept x="z">Y',
    'X<i />X', 'X<i />Y',
    'X<bpt x="y">XX<i />XX<i />X',
    'X<bpt x="y">XX<i />XX<i />Y',
);

if (@ARGV && $ARGV[0] eq 'bench') {
    cmpthese 0 => {
        orig => sub { run_comp(\&compare, 1) },
        kens => sub { run_comp(\&kens_cmp, 1) },
    };
}
else {
    say '*** Original ***';
    say '-' x 60;
    sub { run_comp(\&compare) }->();
    say '*** Ken\'s ***';
    say '-' x 60;
    sub { run_comp(\&kens_cmp) }->();
}

sub run_comp {
    my ($cmp_func, $bench) = @_;

    for my $d1 (@data) {
        for my $d2 (@data) {
            if (defined $bench) {
                $cmp_func->($d1, $d2);
            }
            else {
                say "'$d1' and '$d2': ",
                    $cmp_func->($d1, $d2) ? "SAME" : "DIFFER";
            }
        }
    }
}

sub compare {
  # We ignore the standalone, opening and closing tags <ph />, <bpt x="y">, <ept x="z">, <i />

  my $cString1 = $_[0];
  my $cString2 = $_[1];

  $cString1 =~ s/\/?<(bpt|ept|ph|i)[^<>]*>//gsmi;
  $cString2 =~ s/\/?<(bpt|ept|ph|i)[^<>]*>//gsmi;

  ("$cString1" eq "$cString2") ? return 1 : return 0;
}

sub kens_cmp {
    state $stripped;

    ($stripped->{$_[0]} //= $_[0] =~ s/<(?:i|ph|bpt|ept) [^>]*?>//gr)
    eq
    ($stripped->{$_[1]} //= $_[1] =~ s/<(?:i|ph|bpt|ept) [^>]*?>//gr);
}