# Classes

# Class Constructor

The fundamental part of most classes is its constructor, which sets up each instance's initial state and handles any parameters that were passed when calling new.

It's defined in a class block as though you're defining a method named constructor, though it's actually handled as a special case.

class MyClass {
    constructor(option) {
        console.log(`Creating instance using ${option} option.`);
        this.option = option;
    }
}

Example usage:

const foo = new MyClass('speedy'); // logs: "Creating instance using speedy option"

A small thing to note is that a class constructor cannot be made static via the static keyword, as described below for other methods.

# Class Inheritance

Inheritance works just like it does in other object-oriented languages: methods defined on the superclass are accessible in the extending subclass.

If the subclass declares its own constructor then it must invoke the parents constructor via super() before it can access this.

class SuperClass {

    constructor() {
        this.logger = console.log;
    }

    log() {
        this.logger(`Hello ${this.name}`);
    }

}

class SubClass extends SuperClass {

    constructor() {
        super();
        this.name = 'subclass';
    }

}

const subClass = new SubClass();

subClass.log(); // logs: "Hello subclass"

# Static Methods

Static methods and properties are defined on the class/constructor itself, not on instance objects. These are specified in a class definition by using the static keyword.

class MyClass {
    static myStaticMethod() {
        return 'Hello';
    }

    static get myStaticProperty() {
        return 'Goodbye';
    }
}

console.log(MyClass.myStaticMethod()); // logs: "Hello"
console.log(MyClass.myStaticProperty); // logs: "Goodbye"

We can see that static properties are not defined on object instances:

const myClassInstance = new MyClass();

console.log(myClassInstance.myStaticProperty); // logs: undefined

However, they are defined on subclasses:

class MySubClass extends MyClass {};

console.log(MySubClass.myStaticMethod()); // logs: "Hello"
console.log(MySubClass.myStaticProperty); // logs: "Goodbye"

# Getters and Setters

Getters and setters allow you to define custom behaviour for reading and writing a given property on your class. To the user, they appear the same as any typical property. However, internally a custom function you provide is used to determine the value when the property is accessed (the getter), and to preform any necessary changes when the property is assigned (the setter).

In a class definition, a getter is written like a no-argument method prefixed by the get keyword. A setter is similar, except that it accepts one argument (the new value being assigned) and the set keyword is used instead.

Here's an example class which provides a getter and setter for its .name property. Each time it's assigned, we'll record the new name in an internal .names_ array. Each time it's accessed, we'll return the latest name.

class MyClass {
    constructor() {
        this.names_ = [];
    }

    set name(value) {
        this.names_.push(value);
    }

    get name() {
        return this.names_[this.names_.length - 1];
    }
}

const myClassInstance = new MyClass();
myClassInstance.name = 'Joe';
myClassInstance.name = 'Bob';

console.log(myClassInstance.name); // logs: "Bob"
console.log(myClassInstance.names_); // logs: ["Joe", "Bob"]

If you only define a setter, attempting to access the property will always return undefined.

const classInstance = new class {
    set prop(value) {
        console.log('setting', value);
    }
};

classInstance.prop = 10; // logs: "setting", 10

console.log(classInstance.prop); // logs: undefined

If you only define a getter, attempting to assign the property will have no effect.

const classInstance = new class {
    get prop() {
        return 5;
    }
};

classInstance.prop = 10;

console.log(classInstance.prop); // logs: 5

# Private Members

JavaScript does not technically support private members as a language feature. Privacy - described by Douglas Crockford (opens new window) - gets emulated instead via closures (preserved function scope) that will be generated each with every instantiation call of a constructor function.

The Queue example demonstrates how, with constructor functions, local state can be preserved and made accessible too via privileged methods.

class Queue {

  constructor () {                    // - does generate a closure with each instantiation.

    const list = [];                  // - local state ("private member").

    this.enqueue = function (type) {  // - privileged public method
                                      //   accessing the local state
      list.push(type);                //   "writing" alike.
      return type;
    };
    this.dequeue = function () {      // - privileged public method
                                      //   accessing the local state
      return list.shift();            //   "reading / writing" alike.
    };
  }
}


var q = new Queue;            //
                              //
q.enqueue(9);                 // ... first in ...
q.enqueue(8);                 //
q.enqueue(7);                 //
                              //
console.log(q.dequeue());     // 9 ... first out.
console.log(q.dequeue());     // 8
console.log(q.dequeue());     // 7
console.log(q);               // {}
console.log(Object.keys(q));  // ["enqueue","dequeue"]

With every instantiation of a Queue type the constructor generates a closure.

Thus both of a Queue type's own methods enqueue and dequeue (see Object.keys(q)) still do have access to list that continues to live in its enclosing scope that, at construction time, has been preserved.

Making use of this pattern - emulating private members via privileged public methods - one should keep in mind that, with every instance, additional memory will be consumed for every own property method (for it is code that can't be shared/reused). The same is true for the amount/size of state that is going to be stored within such a closure.

# Dynamic Method Names

There is also the ability to evaluate expressions when naming methods similar to how you can access an objects' properties with []. This can be useful for having dynamic property names, however is often used in conjunction with Symbols.

let METADATA = Symbol('metadata');

class Car {
  constructor(make, model) {
    this.make = make;
    this.model = model;
  }
  
  // example using symbols
  [METADATA]() {
    return {
      make: this.make,
      model: this.model
    };
  }

  // you can also use any javascript expression

  // this one is just a string, and could also be defined with simply add()
  ["add"](a, b) {
    return a + b;
  }

  // this one is dynamically evaluated
  [1 + 2]() {
    return "three";
  }
}

let MazdaMPV = new Car("Mazda", "MPV");
MazdaMPV.add(4, 5); // 9
MazdaMPV[3](); // "three"
MazdaMPV[METADATA](); // { make: "Mazda", model: "MPV" }

# Methods

Methods can be defined in classes to perform a function and optionally return a result.
They can receive arguments from the caller.

class Something {
    constructor(data) {
        this.data = data
    }

    doSomething(text) {
        return {
            data: this.data,
            text
        }
    }
}

var s = new Something({})
s.doSomething("hi") // returns: { data: {}, text: "hi" }

# Managing Private Data with Classes

One of the most common obstacles using classes is finding the proper approach to handle private states. There are 4 common solutions for handling private states:

# Using Symbols

Symbols are new primitive type introduced on in ES2015, as defined at MDN (opens new window)

A symbol is a unique and immutable data type that may be used as an identifier for object properties.

When using symbol as a property key, it is not enumerable.

As such, they won't be revealed using for var in or Object.keys.

Thus we can use symbols to store private data.

const topSecret = Symbol('topSecret'); // our private key; will only be accessible on the scope of the module file
export class SecretAgent{
    constructor(secret){
        this[topSecret] = secret; // we have access to the symbol key (closure)
        this.coverStory = 'just a simple gardner';
        this.doMission = () => {
            figureWhatToDo(topSecret[topSecret]); // we have access to topSecret
        };
    }
}

Because symbols are unique, we must have reference to the original symbol to access the private property.

import {SecretAgent} from 'SecretAgent.js'
const agent = new SecretAgent('steal all the ice cream');
// ok lets try to get the secret out of him!
Object.keys(agent); // ['coverStory'] only cover story is public, our secret is kept.
agent[Symbol('topSecret')]; // undefined, as we said, symbols are always unique, so only the original symbol will help us to get the data.

But it's not 100% private; let's break that agent down! We can use the Object.getOwnPropertySymbols method to get the object symbols.

const secretKeys = Object.getOwnPropertySymbols(agent);
agent[secretKeys[0]] // 'steal all the ice cream' , we got the secret.

# Using WeakMaps

WeakMap is a new type of object that have been added for es6.

As defined on MDN (opens new window)

The WeakMap object is a collection of key/value pairs in which the keys are weakly referenced. The keys must be objects and the values can be arbitrary values.

Another important feature of WeakMap is, as defined on MDN (opens new window).

The key in a WeakMap is held weakly. What this means is that, if there are no other strong references to the key, the entire entry will be removed from the WeakMap by the garbage collector.

The idea is to use the WeakMap, as a static map for the whole class, to hold each instance as key and keep the private data as a value for that instance key.

Thus only inside the class will we have access to the WeakMap collection.

Let's give our agent a try, with WeakMap:

const topSecret = new WeakMap(); // will hold all private data of all instances.
export class SecretAgent{
    constructor(secret){
        topSecret.set(this,secret); // we use this, as the key, to set it on our instance private data
        this.coverStory = 'just a simple gardner';
        this.doMission = () => {
            figureWhatToDo(topSecret.get(this)); // we have access to topSecret
        };
    }
}

Because the const topSecret is defined inside our module closure, and since we didn't bind it to our instance properties, this approach is totally private, and we can't reach the agent topSecret.

# Define all methods inside the constructor

The idea here is simply to define all our methods and members inside the constructor and use the closure to access private members without assigning them to this.


  export class SecretAgent{
        constructor(secret){
            const topSecret = secret;
            this.coverStory = 'just a simple gardner';
            this.doMission = () => {
                figureWhatToDo(topSecret); // we have access to topSecret
            };
        }
    }

In this example as well the data is 100% private and can't be reached outside the class, so our agent is safe.

# Using naming conventions

We will decide that any property who is private will be prefixed with _.

Note that for this approach the data isn't really private.

export class SecretAgent{
    constructor(secret){
        this._topSecret = secret; // it private by convention
        this.coverStory = 'just a simple gardner';
        this.doMission = () => {
            figureWhatToDo(this_topSecret); 
        };
    }
}

# Class Name binding

ClassDeclaration's Name is bound in different ways in different scopes -

The scope in which the class is defined - let binding
The scope of the class itself - within { and } in class {} - const binding

class Foo {
  // Foo inside this block is a const binding
}
// Foo here is a let binding

For example,

class A {
  foo() {
    A = null; // will throw at runtime as A inside the class is a `const` binding
  }
}
A = null; // will NOT throw as A here is a `let` binding

This is not the same for a Function -

function A() {
  A = null; // works
}
A.prototype.foo = function foo() {
  A = null; // works
}
A = null; // works

# Syntax

class Foo {}
class Foo extends Bar {}
class Foo { constructor() {} }
class Foo { myMethod() {} }
class Foo { get myProperty() {} }
class Foo { set myProperty(newValue) {} }
class Foo { static myStaticMethod() {} }
class Foo { static get myStaticProperty() {} }
const Foo = class Foo {};
const Foo = class {};

# Remarks

class support was only added to JavaScript as part of the 2015 es6 standard.

Javascript classes are syntactical sugar over JavaScript's already existing prototype-based inheritance. This new syntax does not introduce a new object-oriented inheritance model to JavaScript, just a simpler way to deal with objects and inheritance. A class declaration is essentially a shorthand for manually defining a constructor function and adding properties to the prototype of the constructor. An important difference is that functions can be called directly (without the new keyword), whereas a class called directly will throw an exception.

class someClass {
    constructor () {}
    someMethod () {}
}
 
console.log(typeof someClass);               
console.log(someClass);
console.log(someClass === someClass.prototype.constructor);                         
console.log(someClass.prototype.someMethod);
 
// Output:
// function
// function someClass() { "use strict"; }
// true
// function () { "use strict"; }

If you are using an earlier version of JavaScript you will need a transpiler like babel or google-closure-compiler in order to compile the code into a version that the target platform will be able to understand.

← Prototypes, objects Namespacing →