Re: SQL like query over iterable in perl (updated x2)

[haukex]

Sounds to me like foreach (and perhaps map) and hashes will help you:

my @array_of_objects = (
    { sex=>'Male', occupation=>'foo' },
    { sex=>'Female', occupation=>'foo' },
    { sex=>'Male', occupation=>'bar' },
    { sex=>'Female', occupation=>'bar' },
    { sex=>'Female', occupation=>'foo' },
);

my %counts;
$counts{ $_->{sex} }++ for @array_of_objects;

my %grouped;
$grouped{ $_->{sex} }{ $_->{occupation} }++ for @array_of_objects;

use Data::Dumper;
print Dumper( \%counts, \%grouped );

__END__

$VAR1 = { 'Female' => 3, 'Male' => 2 };
$VAR2 = { 'Female' => { 'bar' => 1, 'foo' => 2 },
          'Male' => { 'foo' => 1, 'bar' => 1 } };

Update: Personally, I'd prefer the above, but if you really want a generic function, here's one option. Probably not the most efficient solution because it's recursive, I don't think the morning caffeine has fully kicked in yet ;-) Input and output is the same as above. (Update: huck's solution, posted before the below, is the non-recursive variation of this.)

sub count {
    my ($data, $fields) = @_;
    $fields = [$fields] unless ref $fields;
    my $count = {};
    _dive( $count, $_, @$fields ) for @$data;
    return $count;
}
sub _dive {
    my ($ref, $obj, @path) = @_;
    my $targ = \$ref->{ $obj->{ shift @path } };
    if (!@path) { $$targ++; return $ref }
    $$targ = _dive( $$targ, $obj, @path );
}

print Dumper( count(\@array_of_objects, 'sex') );
print Dumper( count(\@array_of_objects,
    ['sex','occupation']) );

Update 2: In the above, I'm working with hash references instead of objects. If you want to use real objects and method calls, then in the first example, replace { $_->{sex} }{ $_->{occupation} } with { $_->sex }{ $_->occupation }, and in the second example, replace $obj->{ shift @path } with $obj->${\shift @path}.