Reputation: 147146
I have a child object that inherits from a parent. Both have a static variable which has a different value in each object; I want to add that variable from both parent and child to an array when I instantiate the child. To save duplicating code, I've written a method (addFoo
) in the parent which is called from both the parent and the child constructors. However, I can't seem to find a way to distinguish between the calls when the parent constructor is called from the child constructor (as you can see below, the output from the method is the same in both cases whether using $this
, self
or static
).
class A {
public static $foo = 'foo';
public $thisvars = array();
public $selfvars = array();
public $staticvars = array();
public function __construct() {
$this->addFoo();
}
public function addFoo() {
$this->selfvars[] = self::$foo;
$this->staticvars[] = static::$foo;
$this->thisvars[] = $this::$foo;
}
}
class B extends A {
public static $foo = 'bar';
public function __construct() {
parent::__construct();
$this->addFoo();
}
}
$b = new B;
print_r($b->selfvars);
print_r($b->staticvars);
print_r($b->thisvars);
Output:
Array
(
[0] => foo
[1] => foo
)
Array
(
[0] => bar
[1] => bar
)
Array
(
[0] => bar
[1] => bar
)
I can workaround this by passing the calling class through to the addFoo
function (see below), but is there a better (correct?) way?
class C {
public static $foo = 'foo';
public $vars = array();
public function __construct() {
$this->addFoo(__CLASS__);
}
public function addFoo($class) {
$this->vars[] = $class::$foo;
}
}
class D extends C {
public static $foo = 'bar';
public function __construct() {
parent::__construct();
$this->addFoo(__CLASS__);
}
}
$d = new D;
print_r($d->vars);
Output:
Array
(
[0] => foo
[1] => bar
)
Upvotes: 1
Views: 1304
Reputation: 14927
I would alternatively consider another, more straightforward approach:
class A
{
public static $foo = 'foo';
public function getVars()
{
return [self::$foo];
}
}
class B extends A
{
public static $foo = 'bar';
public function getVars()
{
return array_merge(parent::getVars(), [self::$foo]);
}
}
class C extends B
{
public static $foo = 'baz';
public function getVars()
{
return array_merge(parent::getVars(), [self::$foo]);
}
}
$a = new A;
print_r($a->getVars()); // ['foo']
$b = new B;
print_r($b->getVars()); // ['foo', 'bar']
$c = new C;
print_r($c->getVars()); // ['foo', 'bar', 'baz']
You do have to redefine getVars
on each subclass in this case, but after all, it makes sense for each class to decide which variables should be exposed. Your code becomes a bit less obscure / easier to maintain in the process.
And if a class doesn't need/want to "contribute", you can simply omit both the static property and the getVars
extension for that class.
Notes:
$vars
property inside A
if needed,array_merge
if needed,$foo
static properties need to be public
, but I've left them like they were in your question.Upvotes: 1
Reputation: 14927
Instead of having addFoo
be called by every sub-constructor, one way would be to have a single addFoos
method in the base class that is called by the base constructor, that would append all the $foo
values starting from the late static binding class:
class A
{
public static $foo = 'foo';
public $vars = [];
public function __construct()
{
$this->addFoos();
}
private function addFoos()
{
$class = static::class;
do {
$this->vars[] = $class::$foo;
} while ($class = get_parent_class($class));
}
}
class B extends A
{
public static $foo = 'bar';
}
class C extends B
{
public static $foo = 'baz';
}
$a = new A;
print_r($a->vars); // ['foo']
$b = new B;
print_r($b->vars); // ['bar', 'foo']
$c = new C;
print_r($c->vars); // ['baz', 'bar', 'foo']
That method is marked private
as it's not supposed to be extended in this scenario (nor called from the outside).
Upvotes: 3