sub tokenize_msg_w_oneregex {
    my ($msg) = @_;

    my $re =  qr{(?:[^\w\'\$!,.-]+|(?:(?<=\D)[.,])|(?:[.,](?=\D|$)))+};

    my %words = map {$_=>1} split $re, $msg;

    return keys %words;
}

##</code><code>##

Benchmark: timing 10000 iterations of Lists, One Regex, Strings...
     Lists:  4 wallclock secs ( 4.15 usr +  0.00 sys =  4.15 CPU) @ 2409.64/s (n=10000)
 One Regex:  4 wallclock secs ( 3.56 usr +  0.00 sys =  3.56 CPU) @ 2808.99/s (n=10000)
   Strings:  2 wallclock secs ( 2.33 usr +  0.00 sys =  2.33 CPU) @ 4291.85/s (n=10000)
            Rate     Lists One Regex   Strings
Lists     2410/s        --      -14%      -44%
One Regex 2809/s       17%        --      -35%
Strings   4292/s       78%       53%        --