What's wrong with 2 hashes?

Maybe a HoH is what you need?

my %bidir= (
  column1 => {
    a => 'b',
  },
  column2 => {
    b => 'a',
  },
);
[download]

The problem about HoH is that I regularly update the content. In which case both hashes need updating. I do not like the idea of doing that for both especially as I access and manipulate the data at several locations. I think I will pick a Class structure giving me a getA and a getB method. This will at least make sure that data of a HoH is consistent as long as my setColumn method is fine.

Cheers,
PerlingTheUK

If you tie Skeeve's hash, you will have the chance to do sync when one of them updated...

$hash{ column_name1 }{ value1 } = value2;
$hash{ column_name2 }{ value2 } = value1;
[download]

$hash{ DATA }{ table1 }{ column_name1 }{ value1 } = value2;
$hash{ DATA }{ table1 }{ column_name2 }{ value2 } = value1;
[download]

#assume that %hash is already initialized
%hash = (%hash, reverse %hash);
[download]

I had thought of this too, but the OP claimed that each of the keys and the values are unique. He didn't exclude that any of the keys may also be a value, although it is definitely reasonable to doubt about this. I'm pointing this out just to let him/her know that this solution may fail in the case it is so:

use Data::Dumper;

my %hash=(foo => 'bar', bar => 'baz');
%hash=(%hash, reverse %hash);
print Dumper \%hash;
[download]

Incidentally I wouldn't have used your smart and elegant syntax -of which I may have thought, but but which didn't occur to me- and I would have done a clumsier

@hash{values %hash} = keys %hash;
[download]

instead.

He didn't exclude that any of the keys may also be a value, although it is definitely reasonable to doubt about this.

I took the OP to mean that each value would be unique in the hash. That is to say that for any key or value in the hash, there would not exist any other key or value that would have the same..erm...value.

thor

Feel the white light, the light within
Be your own disciple, fan the sparks of will
For all of us waiting, your kingdom will come

my %hash=(a => 'foo', b => 'bar', c => 'baz');
my %reverse;
@reverse{values %hash}=keys %hash; # if(f) values are unique, as you c
+laim.
[download]

To re-gain the psychological feeling of unity you may make them into a single HoH with keys qw/direct reverse/ associated to the "direct" and "reverse" hashrefs.

This sounds like a job for overload.

## untested "idea" code ##
package TwoWayHash;
use overload '%{}' => \&as_hash;

sub new {
    my $self = shift;
    my $hashref = shift;
    my %struct = ( 'data' => $hashref );
    
    bless \%struct, $self;
}

sub as_hash {
    my $self = shift;

    # this will only work for simple hashes, really
    my $temp = reverse $self->{data};
    return $temp{shift};
}
[download]

Which in turn might be used as:

use TwoWayHash;

my %a = get_results_from_query; # assume this populates %a;

# then, when you need $b{$a}:
{ 
    my $b = TwoWayHash->(\%a);
    print "I need ",$$b{'item'};  ## finds key where $a{key} eq 'item'
}
[download]

Just ideas, not tested, to get you thinking down this type of road. You could also do this with a function that implements the logic of as_hash(), and then merely call it like:

print "I need", rv_lookup('item',\%a);
[download]

1) How many records does the database table have?
2) What's the average size of each field?
3) What's the ratio of updates to searches?
4) Do you need to support multiple simultaneous searches? If so, up to how many?
5) What are your memory limits?