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

    # define consitituent characters.
    my $con = 'A-Za-z0-9\'\$!,.-';

    # first pass. split tokens on any non-consitituent characters.
    my @words = map {s/[,.]$//; $_} split /[^$con]+/, $msg;

    # second pass. split on , or . that aren't surrounded by digits.
    my %words;
    foreach (@words) { $words{$_}++ foreach split /(?<=\D)[,.](?=\D)/, $_; }

    return keys %words;
}