Final Update: When I replied, the code in the OP did not compile, as one of the four closing braces of $VAR1 was missing, leading me to have to make an assumption of what the data structure looks like. The unmarked update to the OP has now invalidated my code below. Plus, the first nested hash was as shown below instead of the current '19-4' => { 'cost' => '6300.00', 'cost2' => '630.00' }, so the sum of the cost keys was indeed originally 430 as OP expects. vaitor15: It is uncool to update a node in a way that renders replies confusing or meaningless.

Note what you've posted here is not runnable. See How do I post a question effectively? Update 2: Because it's not runnable, I've had to guess as to what your data structure looks like, but as LanX points out in his reply, other interpretations are possible. Please clarify what your real input data looks like. /Update

You should read the Perl Data Structures Cookbook (in particular Access and Printing of a HASH OF HASHES) and the Perl References Tutorial to learn how to work with nested data structures like these. Basically, you can use keys to loop over the keys of each level of hashes - this should be enough to get you started:

use warnings;
use strict;

my %result = (
  "1069-9" => {},
  "135-1"  => { "68-4" => { cost => "300.00", cost2 => "130.00" } },
  "153-1"  => { "19-4" => { cost => "100.00", cost2 =>  "30.00" } },
  "35-1"   => { "28-4" => { cost =>  "30.00", cost2 =>  "10.00" } },
);

for my $k1 ( keys %result ) {
    for my $k2 ( keys %{ $result{$k1} } ) {
        if ( exists $result{$k1}{$k2}{cost} ) {
            print "$k1 / $k2 / $result{$k1}{$k2}{cost}\n";
        }
    }
}
[download]

Update: I used exists in the above code to check for the existence of the cost key because you said "find sum of value when cost key is found". I should add that, depending on what values are possible in your input data, you may want to check for definedness or truth instead.

#!/usr/bin/perl

use strict;
use warnings;
use List::Util qw( sum0 );

my $hashref =
  {
    "153-1" => {
      "1069-9" => {},
      "135-1"  => { "68-4" => { cost => "300.00", cost2 => "130.00" } 
+},
      "19-4"   => { cost => "6300.00", cost2 => "630.00" },
      "35-1"   => { "28-4" => { cost => "30.00", cost2 => "10.00" } },
    },
  };

print costof( $hashref ), " is the total where cost is found\n";

sub costof
  {
  my $href = shift;
  sum0 map /cost/ ? $href->{$_} : costof( $href->{$_} ), keys %$href;
  }
[download]

Outputs:

7400 is the total where cost is found
[download]

G'day vaitor15,

Welcome to the Monastery.

In my code below, I've laid out the data (as I currently see it) in a manner that makes it more readable and comprehensible. Your current costs (6300+300+30) total 6630.

#!/usr/bin/env perl

use strict;
use warnings;

my $VAR1 = {
    '153-1' => {
        '19-4' => {
            'cost' => '6300.00', 'cost2' => '630.00'
        },
        '135-1' => {
            '68-4' => {
                'cost' => '300.00', 'cost2' => '130.00'
            }
        },
        '1069-9' => {
        },
        '35-1' => {
            '28-4' => {
                'cost' => '30.00', 'cost2' => '10.00'
            }
        },
    }
};

my $total = 0;
get_total_cost($VAR1, \$total);
printf "Total = %.2f\n", $total;

sub get_total_cost {
    my ($hash_data, $sub_total) = @_;

    for my $key (keys %$hash_data) {
        if (ref $hash_data->{$key} eq 'HASH') {
            next unless keys %{$hash_data->{$key}};
            get_total_cost($hash_data->{$key}, $sub_total);
        }
        else {
            if ($key eq 'cost') {
                $$sub_total += $hash_data->{cost};
            }
        }
    }

    return;
}
[download]

Note that this doesn't care how deep in the hash the 'cost' key is located. It also ignores 'cost2' keys, and any other variant of a key containing 'cost'; if that's not what you wanted, modify the $key eq 'cost' condition (if it's just keys starting with 'cost', index would be the most efficient; otherwise, a regex would probably be the best solution).

Output:

Total = 6630.00
[download]

> In the version that I'm seeing, 430 as a total is incorrect;

yeah but it's probably the total of

    '153-1' => {
        '135-1' => {
            '68-4' => {
                'cost' => '300.00', 'cost2' => '130.00'
[download]

let's wait for the next update ... ;)

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery}

I was thinking of chaining 3 maps but thats not possible here.

I recommend using Data::Diver or Data::Leaf::Walker or the alternative recursive solutions listed in threads mentioning these modules.

my $VAR1 =
  {
   '153-1' => {
               '19-4' => {
                          'cost' => '6300.00',
                          'cost2' => '630.00'
                         },
               '135-1' => {
                           '68-4' => {
                                      'cost' => '300.00',
                                      'cost2' => '130.00'
                                     }
                          },
               '1069-9' => {},
               '35-1' => {
                          '28-4' => {
                                     'cost' => '30.00',
                                     'cost2' => '10.00'
                                    }
                         },
              },
  };
[download]

your indentation is misleading, the cost entries are at different levels.

You're right that it's misleading, for my reply I assumed there were two closing braces missing right before '135-1' and '1069-9', in which case the cost keys would be at the same level, but your interpretation is possible as well.

vaitor15: Please clarify your input data.

Update: The code in the OP did not compile when I replied because there were missing closing braces. It seems that has now been fixed and you are correct about the nesting.

anyway I don't understand the requirement "sums up to 430" either.

FWIW: This old node shows a recursive walker to dive into a randomly nested structure.

The OP might want to adapt it to his needs.

Cheers Rolf
_{(addicted to the Perl Programming Language :)
Wikisyntax for the Monastery}

use strict;
use warnings;
use List::Util qw(sum);

my $VAR1 = {
    '153-1' => {
        '19-4' => {
            'cost'  => '6300.00',
            'cost2' => '630.00'
        },
        '135-1' => {
            '68-4' => {
                'cost'  => '300.00',
                'cost2' => '130.00'
            }
        },
        '1069-9' => {},
        '35-1'   => {
            '28-4' => {
                'cost'  => '30.00',
                'cost2' => '10.00'
            }
        },

    },
};

my $cost_key = '68-4';
foreach my $temp1 (values %$VAR1) {
    foreach my $temp2 (values %$temp1) {
        next if !exists $temp2->{$cost_key};
        print sum( values(%{$temp2->{$cost_key}}));
    }
}
[download]

OUTPUT:

430
[download]