given what you've said I'd try something like this:

#!/usr/bin/perl
use strict;

open IN, 'file';
open OUT, ">outfile";
my %HASH = ();
my @RUNNING = ();
my $INSIDE = "";        # keep track of where we are
while(local $_ = <IN>) {
        chomp;
        if(/^\s*$/) {
                # we have an empty line
                $HASH{$INSIDE} = @RUNNING if $INSIDE;
                @RUNNING = ();          # empty running now that its c
+opied
                $INSIDE = "";           # no longer inside anything
                next;                   # keep going
        } elsif(/^\s*(pos\s*\=\s*\d+)$/i) {
                # this should be the "header"
                $INSIDE = $1;
                next;
        } elsif($INSIDE) {
                push @RUNNING, $_;
        }
}
close IN;
for(sort keys %HASH) {
        print OUT "$_\n", map { $_ .= "\n"; } @{$HASH{$_}};
}
close OUT;
[download]

*shrugs* it passed strict and warnings.. but is untested.. also, if you're opening a BIG textfile.. IE megs large.. you might run into problems.. it'd get a tad more complex to fix that duplicate records problem with a large file but look at seek(), tell() and keeping records of where you are at the begining of each "id" in the OUT file.. have fun..

-brad..