Digging into a big structure is pretty simple, once you know the rules:
- Wrap hash keys in
{}
- Wrap array indexes in
[]
- If your top level variable is a reference, use
->
before the first identifier.
- After the first set of braces or brackets, additional arrows (
->
) are optional.
So:
* $data->{person}{name}
returns 'Joe Smith'
* $data->{person}->{name}
also returns 'Joe Smith'
* $data->{pets}{cats}[0]{age}
returns 6
.
For way more detail on this topic, see the Perl Data Structures Cookbook (perldoc perldsc)
When you work with big structures like this there are some important things to be aware of. The biggest of these is autovivification
. Autoviv means that Perl will automatically make data structure elements pop into existence for you to make your life easier. Unfortunately it can also make things difficult.
For example, autoviv is great when I do this:
my $data;
$data->{horse}[0]{color} = 'brown';
Autoviv magically turns $data
into a hashref that contains the key horse
with an array ref as its value. The array ref gets populated by a hash ref. The final hash ref then gets the key value pair of color => brown
.
The problem comes in when you are walking a structure and do deep tests for existence:
# Code from above continues:
if( exists $data->{cat}[5]{color} ) {
print "Cat 5 has a color\n";
}
use Data::Dumper;
print Dumper $data;
Here, autovivification burns you by creating a bunch of junk in data, here's the program output:
$VAR1 = {
'cat' => [
undef,
undef,
undef,
undef,
undef,
{}
],
'horse' => [
{
'color' => 'brown'
}
]
};
Now you can guard against this kind of thing by carefully testing each layer of your structure for existence, but it's a huge pain in the butt. Instead, I prefer to use Data::Diver.
use Data::Diver qw( Dive );
my $dog_20_color = Dive( $data, 'dog', 20, 'color' );
print "Dog 20 is $dog_20_color\n" if defined $dog_20_color;
$data
is unchanged here.
Also, you may have noticed that since Dive
takes a list of keys or indexes, that means its easy to programatically build up a list of keys/indexes and descend an arbitrary path in your code.
Data::Diver can be a real life saver when you have to do a lot of manipulation of big, wonky data structures.