Question

I’m having a little difficulty coming up with the OOP design of a small php app I’m building. I have information for Restaurants in a database, split out into a restaurant table and a locations table. Both tables have a few common columns, such as phone, website, and logo url. Obviously the relation between locations and restaurants is many-to-one.

So here’s the issue: I want to create a Restaurant class that has all the information related to the global restaurant information, such as name, phone, website, logo, etc. Then I want to make a Location class that contains location-specific information such as address, phone, website, logo, etc.

The problem that I’m running into is that I want to be able to instantiate both object types, but also want to have the Location class fall back on the parent data if it’s not present in itself. Normally, you’d be able to write something like this (abbreviated):

class Restaurant {

    protected $phone;

    function __construct($restaurant_id) {
        // Perform db call here and set class attributes
    }

    public function getPhone() {
        return $this->phone;
    }
}

class Location extends Restaurant {

    function __construct($location_id) {
        // Perform db call here and set class attributes

        // $restaurant_id would be loaded from the DB above
        parent::__construct($restaurant_id)
    }
}

$location = new Location(123);
echo $location->getPhone();

$restaurant = new Restaurant(456);
echo $restaurant->getPhone();

But like I said, I want the getPhone() method to first check $this->phone, and if it doesn’t exist, the fall back to the parent. Would something like this be the correct way?

class Restaurant {

    private $phone;

    function __construct($restaurant_id) {
        // Perform db call here and set class attributes
    }

    public getPhone() {
        return $this->phone;
    }
}

class Location extends Restaurant {

    private $phone;

    function __construct($location_id) {
        // Perform db call here and set class attributes

        // $restaurant_id would be loaded from the DB above
        parent::__construct($restaurant_id)
    }

    public function getPhone() {
        if(!empty($this->phone)) {
            return $this->phone;
        }
        return parent::getPhone();
    }
}

$location = new Location(123);
echo $location->getPhone();

I feel like the above code is really hacky, and there’s probably a much better way of accomplishing this. Since the two have common attributes, would it be better for the Location class to not extend Restaurant but instead hold a variable of type Restaurant for the “parent” object? Then in the Location::getPhone() method, it performs a similar if(empty()) check?

Answer 1

Location should not extend Restaurant because it isn’t a restaurant itself; it’s one of many locations of that restaurant.

class Location {
    private $restaurant;
    private $phone;

    public function getPhone() {
        return $this->phone ?: $restaurant->getPhone();
    }
}

Now, with so many fields in common between the two classes, you may want to define a common base class that they each extend, such as CompanyInfoHolder that contains website, phone, and logo. In this case, Location would override getPhone exactly as above.