The first thing to do is to split your data into an array of arrays.

As your data doesn't contain any duplicated values, this doesn't show that it will correctly sort by two or more fields, but it will. Translating column names to numbers is left as an exercise.

#! perl -slw
use strict;

my @inventory = (
 [ qw'2003 abc rel 999' ],
 [ qw'1999 hds sdf 100' ],
 [qw'2010 kls pol 1400' ],
);;

while( my @order = split ' ', <STDIN> ) {
    print "@$_" for  sort{
        my $r;
        for(@order) {
            if( $a->[ $_ ] =~ m[^\d+$] &&  $b->[ $_ ] =~ m[^\d+$] ) {
                 ( $r = $a->[ $_ ] <=> $b->[ $_ ] ) and last;
            }
            else {
                 ( $r = $a->[ $_ ] cmp $b->[ $_ ] ) and last;
            }
        }
        $r
    } @inventory;
}

__END__
c:\test>junk42
0
1999 hds sdf 100
2003 abc rel 999
2010 kls pol 1400
1
2003 abc rel 999
1999 hds sdf 100
2010 kls pol 1400
2
2010 kls pol 1400
2003 abc rel 999
1999 hds sdf 100
3
1999 hds sdf 100
2003 abc rel 999
2010 kls pol 1400
1 2
2003 abc rel 999
1999 hds sdf 100
2010 kls pol 1400
^Z
[download]