SuicideJunkie hints at the solution for you in Re: Sort a Hash 3 Levels Deep, but you may need a little elaboration. The following mocks up your data structure, generates a sorted list of keys, then uses your sample for loop to print the result:

use strict;
use warnings;

my %siteInfo;

while (<DATA>) {
    chomp;

    my @parts = split;
    my $site  = ($siteInfo{"Site$."} = {});

    @{$site}{qw(region country city)} = @parts;
    $site->{totalS5s} = $site->{totalVulns} = ++$site->{totalDevices};
}

my @sorted = sort {
    my $siteA = $siteInfo{$a};
    my $siteB = $siteInfo{$b};

       $siteA->{region} cmp $siteB->{region}
    or $siteA->{country} cmp $siteB->{country}
    or $siteA->{city} cmp $siteB->{city}
    } keys %siteInfo;

for my $site1 (@sorted) {
    my $site = $siteInfo{$site1};
    print "$site->{region},$site->{country},$site1,$site->{totalDevice
+s}, ,$site->{totalVulns},$site->{totalS5s}\n";
}

__DATA__
...
[download]

Prints (if you replace ... above with the sample data):

APAC,Australia,Site3,1, ,1,1
APAC,China,Site7,1, ,1,1
APAC,China,Site2,1, ,1,1
APAC,India,Site1,1, ,1,1
APAC,Japan,Site4,1, ,1,1
EMEA,France,Site9,1, ,1,1
EMEA,Germany,Site8,1, ,1,1
EMEA,Italy,Site15,1, ,1,1
EMEA,Italy,Site10,1, ,1,1
EMEA,Russia,Site6,1, ,1,1
EMEA,SouthAfrica,Site5,1, ,1,1
NLAM,Argentina,Site11,1, ,1,1
NLAM,Brazil,Site13,1, ,1,1
NLAM,USA,Site14,1, ,1,1
NLAM,USA,Site16,1, ,1,1
NLAM,USA,Site12,1, ,1,1
[download]

Thank you. That is what I was trying to do.


Dru

Perl, the Leatherman of Programming languages. - qazwart

foreach my $key1 (sort keys( %hash ) )
{
    foreach my $key2 (sort keys( %{$hash{$key1}} ) )
    {
        foreach my $key3 (sort keys( %{$hash{$key1}{$key2}} ) )
        {
        }
    }
}
[download]

@SortedArrayOfHashes = sort 
            { 
            return $a->{'key1'} cmp $b->{'key1'} or  # if these are eq
+ual try the comparison below
                    $a->{'key2'} cmp $b->{'key2'} or # if these are eq
+ual try below
                    $a->{'key3'} cmp $b->{'key3'} #if these are equal,
+ then we give up; they're the same.
            }
            @ArrayOfHashes;
[download]

It looks to me as if you are not doing a sort, but rather grouping only. The order is the order returned by the keys statement, which is undefined (disclaimer: it is snowing here; I had to battle Jack Frost twice, and the snowplow once this morning, so I may be as hazy as the sky outside of my window).

To sort, you could nested loops on sort keys... (as noted by SuicideJunkie above) at each level of the data structure. This might be the simplest option.

You could also extract all levels of the structure's keys ([[reg1,ctry1,city1][reg2,ctry2,city2]...]), sort these, and then extract the data from the hash given these sorted values.

You could also join the keys into a single string for each entry (["reg1:ctry1:city1","reg2:ctry2:city2"...]), sort these, and extract the data from the hash using these sorted values (be careful of what you choose to use as the delimiter here).

On the last two, a Schwarzian Transform might (or might not) buy you some performance or understanding for your maintenance programmers, but you would need to benchmark the options to be sure what works best in your situation.

This all depends on your data structures (see kennethk above), how you build them, and how large your data set is.

Update: Added some clarification and credit. I was way too slow posting my response ;-)