Update: I saw the post from johngg at Re^7: Sort files by date in the file name.. I had completely missed the fact that the date format is different in this second data set. If this is the format, then a simple numeric date comparison won't work and something more involved is required. However, when I see at date like 28062017102508, in 2017, that makes me think that this data was hand generated. Note to OP when you post example data, make it as close to the "real thing" as possible. These details matter.

Also, the sort method I show below is basic. Although straight forward, it has a built in inefficiency in that the the splits are run each time a pair of data in $a,$b is selected by the sort algorithm. The Schwartzian Transform as shown by kcott avoids this by calculating the split for all the data at one time. So it is faster, but at the cost of more complexity. I recommend to a beginner to master the basics first, then get fancy once you have a solid foundation.

#!/usr/bin/perl
use strict;
use warnings;

my @files;
while (<DATA>)
{
   chomp;
   push @files, $_;
}

@files =  sort by_first_third @files;

sub by_first_third
{
   my ($Afirst,$Athird) = (split '_|\.',$a)[0,2];
   my ($Bfirst,$Bthird) = (split '_|\.',$b)[0,2];
         $Afirst <=> $Bfirst
                  or
         $Athird <=> $Bthird
}

print join ("\n",@files), "\n\n";
print "first file: $files[0]", "\n";
print "last file : $files[-1]", "\n";

=prints

971305332_XXXXXX12345678765463565E_28062011102508.TXT 
971305332_AAAAAAAA12345678765463565E_28062011102508.TXT 
971305332_CCCC12345678765463565E_28062011102508.TXT 
971305332_CCCC12345678765463565E_28062020102508.TXT
981139804_ABCDEF12345678765463565E_28062016102508.TXT 
981139804_ABCDEF12345678765463565E_28062016112508.TXT 
981139804_ABCDEF12345678765463565E_28062016172508.TXT  
981139804_ABCDEF12345678765463565E_28062017102508.TXT 

first file: 971305332_XXXXXX12345678765463565E_28062011102508.TXT 
last file : 981139804_ABCDEF12345678765463565E_28062017102508.TXT  

=cut


__DATA__
981139804_ABCDEF12345678765463565E_28062016172508.TXT  
981139804_ABCDEF12345678765463565E_28062016112508.TXT 
981139804_ABCDEF12345678765463565E_28062016102508.TXT 
981139804_ABCDEF12345678765463565E_28062017102508.TXT 
971305332_XXXXXX12345678765463565E_28062011102508.TXT 
971305332_AAAAAAAA12345678765463565E_28062011102508.TXT 
971305332_CCCC12345678765463565E_28062011102508.TXT 
971305332_CCCC12345678765463565E_28062020102508.TXT
[download]

G'day leoberbert,

I think this does what you want:

#!/usr/bin/env perl -l

use strict;
use warnings;

my @files = qw{
    1020300000_XXXXXXXXX_20160707193000.TXT
    1020300000_XXXXXXXXX_20160707170000.TXT
    1020400000_XXXXXXXXX_20160707180000.TXT
    1020400000_XXXXXXXXX_20160707190000.TXT
};

my @sorted = map {
        $_->[0]
    }
    sort {
        $a->[1] <=> $b->[1] || $a->[3] <=> $b->[3]
    }
    map {
        [ $_ => split /[_.]/ ]
    } @files;

print join "\n", '@files:', @files;
print join "\n", '@sorted:', @sorted;
[download]

Output:

@files:
1020300000_XXXXXXXXX_20160707193000.TXT
1020300000_XXXXXXXXX_20160707170000.TXT
1020400000_XXXXXXXXX_20160707180000.TXT
1020400000_XXXXXXXXX_20160707190000.TXT
@sorted:
1020300000_XXXXXXXXX_20160707170000.TXT
1020300000_XXXXXXXXX_20160707193000.TXT
1020400000_XXXXXXXXX_20160707180000.TXT
1020400000_XXXXXXXXX_20160707190000.TXT
[download]

The construct I've used here is known as a Schwartzian Transform.

This sorts first on column1 and then on column3, both in ascending order; this should put your wanted file in $sorted[0]. If that's not what you want, modify the sort { ... } code to suit.

The following code finds the oldest file for each "first column" set, if there are multiple oldest files it will display only the first in lexical order. It is not clear whether the date format is YYYYMMDD as in the OP or DDMMYYYY as per the later posted data; I have gone with the latter.

johngg@shiraz:~/perl/Monks > perl -Mstrict -Mwarnings -E '
my @files = qw{
   981139804_ABCDEF12345678765463565E_28062016172508.TXT
   981139804_ABCDEF12345678765463565E_28062016192508.TXT
   981139804_ABCDEF12345678765463565E_28062016112508.TXT
   981139804_ABCDEF12345678765463565E_28062016102508.TXT
   981139804_ABCDEF12345678765463565E_28062017102508.TXT
   971305332_XXXXXX12345678765463565E_28062011102508.TXT
   971305332_AAAAAAAA12345678765463565E_28062011102508.TXT
   971305332_CCCC12345678765463565E_28062011102508.TXT
   971305332_CCCC12345678765463565E_28062020102508.TXT
   };

my @oldest = do {
   my %seen;
   grep { not $seen{ ( split m{_} )[ 0 ] } ++ }
   map  { substr $_, 21 }
   sort
   map  {
            my( $col1, $dt ) = ( split m{[_.]} )[ 0, 2 ];
            my( $d, $m, $y, $t ) = unpack q{a2 a2 a4 a6}, $dt;
            pack q{a9 a4 a2 a2 a6 a*}, $col1, $y, $m, $d, $t, $_;
        }
   @files;
   };

say for @oldest;'
08971305332_AAAAAAAA12345678765463565E_28062011102508.TXT
08981139804_ABCDEF12345678765463565E_28062016102508.TXT
[download]

I hope this guess at your intent is correct.

What is first column code ?

A number that must be used as a key ..... The first and last column should be the sort keys.

A number that must be used as a key ..... The first and last column should be the sort keys.

:) you gave sample input and sample output

the sample output does not appear in the sample input

its not clear what you mean by "first column code" and I cannot deduce/guess from the sample you provided

The word explanations do not clarify which part is which