Thank you for reply.

Please take a look for my example:

        (...)
        my $ref = $entry->get_value ( 'changes', asref => 1 );
        print Dumper($ref);
[download]

Execution:

perl test.pl
$VAR1 = [
          'delete: employeeType
employeeType: Intern
-
add: employeeType
employeeType: No Longer Employed
-
add: description
description: On 30-07-2013 05:17 removed from source.
-
add: nsAccountLock
nsAccountLock: true
-
replace: modifiersname
modifiersname: cn=mmariusz,ou=people,ou=testENV
-
replace: modifytimestamp
modifytimestamp: 20130730031746Z
-
'
[download]

So, this is stil "multiline" result in one variable. Probably I need to create some kind of function/object and make a parser for that:

function magic() {
       my $string = shift;
       
      # some magic should be placed here :)

      #sample return
      return @replaced_attrs,@added_attrs,@delete_attrs;

}
[download]

I can't imagine that its necessary that you parse this yourself, see http://search.cpan.org/grep?cpanid=MARSCHAP&release=perl-ldap-0.57&string=change&i=1&n=1&C=0, Net::LDAP::Control::EntryChange, and employ more dumpering :)

Uff solved:

(...)
       my $chang = &_changes($entry->get_value("changes"));
       my %changes = %$chang;

sub _changes {
        my $changesSTR = shift;
        my (%changes, @tmp);
        $changesSTR =~ s/\n/ /g;
        foreach my $element(split(/\s-\s/,$changesSTR)){
            if ($element =~ m/(replace|add|delete):\s+([^\s]*)\s*\w+\s
+*:\s*(.*)$/) {
                  my ($var1,$var2,$var3) = ($1, $2 ,$3);
                  push @tmp,$var3 ;
                  $changes{$var1}{$var2} .= "@tmp" ;
                  @tmp=undef;
             } elsif ($element =~ m/(replace|add|delete):\s+([^\s]*)$/
+){
                 $changes{$1}{$2} .= "undef" ;
             }
        }

        return \%changes;
}
[download]

Maybe solution is not pretty, but works ;)