Recipe 11.1. Taking References to Arrays

Problem

You need to manipulate an array by reference.

Solution

To get a reference to an array:

$aref               = \@array;
$anon_array         = [1, 3, 5, 7, 9];
$anon_copy          = [ @array ];
@$implicit_creation = (2, 4, 6, 8, 10);

To deference an array reference, precede it with an at sign (@):

push(@$anon_array, 11);

Or use a pointer arrow plus a bracketed subscript for a particular element:

$two = $implicit_creation->[0];

To get the last index number by reference, or the number of items in that referenced array:

$last_idx  = $#$aref;
$num_items = @$aref;

Or defensively embracing and forcing context:

$last_idx  = $#{ $aref };
$num_items = scalar @{ $aref };

Discussion

Here are array references in action:

# check whether $someref contains a simple array reference
if (ref($someref) ne 'ARRAY') {
    die "Expected an array reference, not $someref\n";
}

print "@{$array_ref}\n";        # print original data

@order = sort @{ $array_ref };  # sort it

push @{ $array_ref }, $item;    # append new element to orig array  

If you can't decide whether to use a reference to a named array or to create a new one, here's a simplistic guideline that will prove right more often than not. Only take a reference to an existing array either to return the reference out of scope, thereby creating an anonymous array, or to pass the array by reference to a function. For virtually all other cases, use [@array] to create a new array reference with a copy of the old values.

Automatic reference counting and the backslash operator make a powerful combination:

sub array_ref {
    my @array;
    return \@array;
}

$aref1 = array_ref();
$aref2 = array_ref();

Each time array_ref is called, the function allocates a new piece of memory for @array. If we hadn't returned a reference to @array, its memory would have been freed when its block, the subroutine, ended. However, because a reference to @array is still accessible, Perl doesn't free that storage, and we end up with a reference to a piece of memory that can no longer be accessed through the symbol table. Such a piece of memory is called anonymous because no name is associated with it.

To access a particular element of the array referenced by $aref, you could write $$aref[4], but writing $aref->[4] is the same thing, and it is clearer.

print $array_ref->[$N];         # access item in position N (best)
print $$array_ref[$N];          # same, but confusing
print ${$array_ref}[$N];        # same, but still confusing, and ugly to boot

If you have an array reference, you can only access a slice of the referenced array in this way:

@$pie[3..5];                    # array slice, but a little confusing to read
@{$pie}[3..5];                  # array slice, easier (?) to read

Array slices, even when accessed through array references, are assignable. In the next line, the array dereference happens first, and then the slice:

@{$pie}[3..5] = ("blackberry", "blueberry", "pumpkin");

An array slice is exactly the same as a list of individual array elements. Because you can't take a reference to a list, you can't take a reference to an array slice:

$sliceref = \@{$pie}[3..5];     # WRONG!

To iterate through the entire array, use either a foreach loop or a for loop:

foreach $item ( @{$array_ref} ) {   
    # $item has data
}

for ($idx = 0; $idx <= $#{ $array_ref }; $idx++) {  
    # $array_ref->[$idx] has data
}

See Also

Chapter 4 of Programming Perl; perlref (1) and perllol (1); Recipe 2.14; Recipe 4.5