Have you tried with Hash::Merge? This module can merge arbitrarily deep hashes into a single hash; it can also merge ARRAYs and SCALARs. Hope this helps.

Ciao, Valerio

use strict;
use Data::Dumper;
my %hash1;
# fill %hash1 here
print Dumper \%hash1;
[download]

Got it, in fact I use references (it's my question that wasn't clear enough), the thing is that I always stay with one value for the same keys, if I'll take my example from the question, then I get only one brown entry. The references always clober each other and not added, I didn't seem to solve that yet.

Hotshot

Thats what hashes do. map a unique key to a unique values. But you can make that value an array ref and store your multiple values.

It also looks like you want to apply it recursively. May I ask what you are trying to do?

-Lee

"To be civilized is to deny one's nature."
UPDATE

---

Here's enough rope to hang yourself with.
Merges N hashes arbitrarily deep. Making it merge arrays as well is left as an exercise for the reader.
Dies on circular references. Not heavily tested.

#!/usr/bin/perl

use warnings; 
use strict;
use Data::Dumper;

BEGIN {
        my %SeenMerged = ();

        sub mergehashes{     # Dies on circular references
            my @hashrefs = @_;
            die "Passed a non hashref" if grep { ref $_ ne 'HASH' } @h
+ashrefs;
            my %merged = ();

            my @seen = grep { ref $_ eq 'HASH' }  @SeenMerged{@hashref
+s}; # Break circular links..

            if (@seen){
                die "contains a circular reference! bailing...";
            }

            @SeenMerged{@hashrefs} = @hashrefs;

            foreach my $h (@hashrefs){
                while (my ($k,$v) = each %$h ){
                    push @{$merged{$k}}, $v;
                }
            }

            while (my ($k,$v) = each %merged){
                my @hashes = grep { ref $_ eq 'HASH' } @$v;
                $merged{$k} = $v->[0] if (@$v == 1 && !ref $v->[0]);  
+  
                $merged{$k}  = mergehashes(@hashes) if @hashes;
            }

            delete @SeenMerged{@hashrefs};
            return \%merged;
        }

}

my %hash1 = (
          red => 1,
          brown => {
                    green => 1,
                    blue => {
                             yellow => 1,
                            },
                    black => 1,
                   },
          gray => 1,
         );

my %hash2 = (
          white => 1,
          brown => {
                    purple => 1,
                   },
          
         );


my $merged = mergehashes(\%hash1, \%hash2) ;#or die "Couldn't merge!";

print Dumper($merged);
[download]

#!/usr/bin/perl -w

use Data::Dumper;

%hash1 = (
          red => 1,
          brown => {
                    green => 1,
                    blue => {
                             yellow => 1,
                            },
                    black => 1,
                   },
          gray => 1,
         );

%hash2 = (
          white => 1,
          brown => {
                    purple => 1,
                   },
         );

# merge ( \%hash1,  \%hash2 ) returns the merge
# If the two hashes have a common (possibly deep) key
# not pointing to hashes, it dies trying to force
# the reference to a hash reference
sub merge {
    my ( $h1, $h2 ) = @_;
    my %h3;
    for my $key ( keys %$h1 ) {
    if ( exists $h2->{$key} ) {
        $h3{$key} = merge( $h1->{$key}, $h2->{$key} );
    } else {
        $h3{$key} = $h1->{$key};
    }
    }
    for my $key ( keys %$h2 ) {
    unless ( exists $h1->{$key} ) {
        $h3{$key} = $h2->{$key};
    }
    }
    return \%h3;
}

print Dumper ( merge( \%hash1, \%hash2 ) );
[download]