use Hash::MultiValue; my $hash = Hash::MultiValue->new( foo => 'a', foo => 'b', bar => 'baz', ); # $hash is an object, but can be used as a hashref and DWIMs! my $foo = $hash->{foo}; # 'b' (the last entry) my $foo = $hash->get('foo'); # 'b' (always, regardless of context) my @foo = $hash->get_all('foo'); # ('a', 'b') keys %$hash; # ('foo', 'bar') not guaranteed to be ordered $hash->keys; # ('foo', 'foo', 'bar') guaranteed to be ordered
Why? Because the value could be just a scalar if there is one value and an array ref if there are multiple, depending on user input rather than how you code it. So your code should always be like this to be defensive:
my $p = $c->req->parameters; my @maybe_multi = ref $p->{m} eq 'ARRAY' ? @{$p->{m}} : ($p->{m}); my $must_single = ref $p->{m} eq 'ARRAY' ? $p->{m}->[0] : $p->{m};
Otherwise you'll get a random runtime exception of Can't use string as an ARRAY ref or get stringified array ARRAY(0xXXXXXXXXX) as a string, depending on user input and that is miserable and insecure.
This module provides a solution to this by making it behave like a single value hash reference, but also has an API to get multiple values on demand, explicitly.
This module does not use "tie" or overload and is quite fast.
Yes, there is Tie::Hash::MultiValue and this module tries to solve exactly the same problem, but using a different implementation.
my $hash = Hash::MultiValue->new(...); # WRONG $hash->{foo} = 'bar'; delete $hash->{foo}; # Correct $hash->add(foo => 'bar'); $hash->remove('foo');
See below for the list of updating methods.
$hash = Hash::MultiValue->new(@pairs);
Creates a new object that can be treated as a plain hash reference as well.
$value = $hash->get($key); $value = $hash->{$key};
Returns a single value for the given $key. If there are multiple values, the last one (not first one) is returned. See below for why.
Note that this always returns the single element as a scalar, regardless of its context, unlike CGI.pm's "param" method etc.
$value = $hash->get_one($key);
Returns a single value for the given $key. This method croaks if there is no value or multiple values associated with the key, so you should wrap it with eval or modules like Try::Tiny.
@values = $hash->get_all($key);
Returns a list of values for the given $key. This method always returns a list regardless of its context. If there is no value attached, the result will be an empty list.
@keys = $hash->keys;
Returns a list of all keys, including duplicates (see the example in the ``SYNOPSIS'').
If you want only unique keys, use "keys %$hash", as normal.
@values = $hash->values;
Returns a list of all values, in the same order as "$hash->keys".
$hash->set($key [, $value ... ]);
Changes the stored value(s) of the given $key. This removes or adds pairs as necessary to store the new list but otherwise preserves order of existing pairs. "$hash->{$key}" is updated to point to the last value.
$hash->add($key, $value [, $value ... ]);
Appends a new value to the given $key. This updates the value of "$hash->{$key}" as well so it always points to the last value.
$hash->remove($key);
Removes a key and associated values for the given $key.
$hash->clear;
Clears the hash to be an empty hash reference.
@pairs = $hash->flatten;
Gets pairs of keys and values. This should be exactly the same pairs which are given to "new" method unless you updated the data.
$hash->each($code); # e.g. $hash->each(sub { print "$_[0] = $_[1]\n" });
Calls $code once for each "($key, $value)" pair. This is a more convenient alternative to calling "flatten" and then iterating over it two items at a time.
Inside $code, $_ contains the current iteration through the loop, starting at 0. For example:
$hash = Hash::MultiValue->new(a => 1, b => 2, c => 3, a => 4); $hash->each(sub { print "$_: $_[0] = $_[1]\n" }); # 0: a = 1 # 1: b = 2 # 2: c = 3 # 3: a = 4
Be careful not to change @_ inside your coderef! It will update the tracking object but not the plain hash. In the future, this limitation may be removed.
$new = $hash->clone;
Creates a new Hash::MultiValue object that represents the same data, but obviously not sharing the reference. It's identical to:
$new = Hash::MultiValue->new($hash->flatten);
$copy = $hash->as_hashref;
Creates a new plain (unblessed) hash reference where a value is a single scalar. It's identical to:
$copy = +{%$hash};
$mixed = $hash->as_hashref_mixed; $mixed = $hash->mixed;
Creates a new plain (unblessed) hash reference where the value is a single scalar, or an array ref when there are multiple values for a same key. Handy to create a hash reference that is often used in web application frameworks request objects such as Catalyst. Ths method does exactly the opposite of "from_mixed".
$multi = $hash->as_hashref_multi; $multi = $hash->multi;
Creates a new plain (unblessed) hash reference where values are all array references, regardless of there are single or multiple values for a same key.
$hash = Hash::MultiValue->from_mixed({ foo => [ 'a', 'b' ], bar => 'c', });
Creates a new object out of a hash reference where the value is single or an array ref depending on the number of elements. Handy to convert from those request objects used in web frameworks such as Catalyst. This method does exactly the opposite of "as_hashref_mixed".
sub x { return ('a', 'b', 'c'); } my $x = x(); # $x = 'c' my %a = ( a => 1 ); my %b = ( a => 2 ); my %m = (%a, %b); # $m{a} = 2
When perl gets a list in a scalar context it gets the last entry. Also if you merge hashes having a same key, the last one wins.
if (ref $args eq 'HASH') { ... }
because this is a blessed hash reference it doesn't match and would fail. To avoid that you should call "as_hashref" to get a finalized (= non-blessed) hash reference.
You can also use UNIVERSAL::ref to make it work magically:
use UNIVERSAL::ref; # before loading Hash::MultiValue use Hash::MultiValue;
and then all "ref" calls to Hash::MultiValue objects will return HASH.
Aristotle Pagaltzis
Hans Dieter Pearcey
Thanks to Michael Peters for the suggestion to use inside-out objects instead of tie.