Your explanation is difficult to understand. It would have been easier for you and clearer to someone trying to answer your questions if you had just written:

I have this string:

my $str = 'XXXX - AAAA - - XXXXXXX XX - XXXXX XX TNBBBBB ';
[download]

...and I want to end up with this hash:

my $data = (
   'keyA' => 'AAAAA',
   'keyB' => 'BBBBB'
);
[download]

...or whatever it is you want.

If you have this string:

my $str = "XXXX - 12345 - - XXXXXXX XX - XXXXX XX TN67890";
[download]

and you want to get 12345 and 67890 into a hash, you can do this:

use strict;
use warnings;
use 5.010;

my $str = 'XXXX - 12345 - - XXXXXXX XX - XXXXX XX TN6789';
my @pieces = split /\s+/, $str;

my $val1 = $pieces[2];
(my $val2 = $pieces[-1]) =~ s/^TN//;

say $val1;
say $val2;

my %hash = (
    'keyA' => $val1,
    'keyB' => $val2
);

say %hash;

--output:--
12345
6789
keyB6789keyA12345
[download]

If you don't know which letters are on the front of "TN6789", or even how many letters are on the front of that string, you can change the substitution to this:

 s/^\D*//
[download]

...where ^ matches the start of the string, and \D matches "not a digit", and the * applies to "not a digit" and means "0 or more times". Because the * is greedy, it will match as many non-digits as it can find at the start of the string.

#!/usr/bin/perl

use strict;

my %record;
my $a = 'XXXXval - XXXX - - XXXXXXX XX - XXXXX XX TNkeyXXXXXXX';

my ($key, $val) = $a =~ /^(\w+)\s-.*\sTN(\w+)/;

$record{$key} = $val;
print "Key: $key\n";
print "Value: $record{$key}\n";
[download]