Develop Reference

JS closure to return object instance as interface

I have the following code.
function Test() {
this.funct_1 = function() {
alert('funct_1');
}
this.funct_2 = function() {
alert('funct_2');
}
return this;}
function getTestObj() {
var testObj;
if (!testObj) {
testObj = new Test();
}
return function() {
return testObj;
}}
What I'm trying to accomplish is the following. I want to have a class Test which is not singleton. Then in some other places in my application I need to have a function which could return the same instance per script execution. I figured that I could use closure for that getTestObj.
However, when I try to use it
getTestObj().funct_1();
I'm getting the following error, saying the funct_1() is not found.
Cannot find function funct_1 in object function () {...}.
Clearly, I'm making some kind of mistake here, but I'm not able to find any solution over the net which could help me. Would appreciate any comments.
NOTE: I'm forced to use ECMA5

testObj is wrapped inside a function
So, either call it
getTestObj()().funct_1(); //notice two ()()
Save the value of getTestObj() in a variable
var singleTon = getTestObj();
var testObj = singleTon();
testObj.funct_1();
Or, simply return testObj (in case singleTon isn't required)
function getTestObj()
{
var testObj;
if (!testObj) {
testObj = new Test();
}
return testObj;
}
And invoke it as
getTestObj().funct_1(); //notice single ()

getTestObj() is returning a function i.e. :
function() {
return testObj;
}
So you have to call it again getTestObj()(), this will return the Test's object and now you can access it's properties.
getTestObj()().funct_1();
OR
You can change your getTestObj function as :
function getTestObj() {
var testObj;
if (!testObj) {
testObj = new Test();
}
return (function() {
return testObj;
}());
}

return whole self invoking functions

How can I return the whole object of the self-inv-function without returning every functions manually?
I want to try with the following solution which should normally work, however, it does not work:
var publish = (function() {
var pub = {};
pub.hello = function() {
return "test"
};
return pub;
}());
now "pub" must be callable by subscribe:
var subsribe = (function(pub) {
function hello() {
return pub.hello();
};
}(publish));
I loaded both files in the browser (pub first).
However:
Debugger says: ReferenceError: pub not defined.

I think you want to write them like this:
var publish = (function() {
var pub = {};
pub.hello = function() {
return "test"
};
return pub;
})();
var subsribe = (function(pub) {
function hello() {
return pub.hello();
};
console.log(hello());
})(publish);
However, keeping a global reusable collections of functions can be accomplished in other ways, more elegantly maybe :) (separate file with export, singleton decorated with those methods)

You can't.
There's no mechanism in JS to get a list of variables in the current scope.
Even if you could, it probably wouldn't be a good idea as there would be no way to distinguish between public and private variables.

There is no such a mechanism but you can do something like this:
var publish = (function() {
const me = this;
let publicMethods = ['hello', 'bye'];
// private
function _hello() {
return "test";
};
function _bye() {
return "end test";
};
publicMethods.forEach((methodName) => {
let privateMethod = eval('_' + methodName);
Object.defineProperty(me, methodName, {
get: function() {
return privateMethod;
}
});
});
return this;
}());
console.log(publish.hello);
console.log(publish.bye);
console.log(publish.hello());
console.log(publish.bye());

Is it more efficient to use a common empty function instead of creating a new one in each class instance?

Let's say I have a class that is designed to have some callbacks added to it later on.
function myclass() {
this.onSomething = function () {};
this.onOtherThing = function () {};
this.something = function () {
// stuff
this.onSomething();
};
this.otherThing = function () {
// other stuff
this.onOtherThing();
};
}
I can't have this.onSomething and this.onOtherThing being undefined or null because when they are called in something() and otherThing(), an error will be thrown, stating that their type is not a function.
Since those empty functions are needed, but they use memory, is the class going to be more memory efficient if I did this?
function myclass() {
this.onSomething = empty;
this.onOtherThing = empty;
...
}
function empty() {
}
This way each class instance's properties point to the same empty function, instead of creating new functions every time. I assume defining an empty method doesn't take a lot of memory, but still... is this technically better?

You are right about the fact that a new function is created for every instance of your class. In order to have this shared across all instances you can declare it on the prototype of the class:
var MyClass = function() {
this.something = function () {
// stuff
this.onSomething();
};
this.otherThing = function () {
// other stuff
this.onOtherThing();
};
}
MyClass.prototype.onSomething = function() {};
MyClass.prototype.onOtherThing = function() {};
This way, the methods will be shared by all instances.

why don't you try to return true or return false instead of returning empty functions.
or best you can use :
function myclass() {
this.onSomething = false;
this.onOtherThing = false;
...
}
as per your comment you can try :
function myclass() {
this.onSomething = empty();
this.onOtherThing = empty();
... }
function empty() {
//return something
return true;
}

avoid needing to declare 'var me = this' for javascript prototype functions

Currently, I create objects in javascript by declaring a construction (regular function) then add methods to the prototype like so
function Test(){
}
Test.prototype.test1 = function(){
var me = this;
}
However, I would like to avoid having to declare var me = this at the top of every function. The following seems to work, but seems like it would be very inefficient:
$(document).ready(function(){
var n = 0;
(function(){
function createTest(){
var me;
function Test(){
this.n = n;
this.testArr = [1, 2, 3, 4];
n++;
}
Test.prototype.test1 = function(){
me.test2();
};
Test.prototype.test2 = function(){
alert(me.n);
$.getJSON('test.php', {}, function(reply)
//want to be able to use 'me' here
me.newField = reply;
});
};
var t = new Test();
me = t;
return t;
}
window['createTest'] = createTest;
})();
var t = createTest();
t.test1();
var t2 = createTest();
t2.test1();
t.test1();
});
This code outputs the expected, but is it actually as inefficient as it looks (the Test object being re-declared every time you call createTest())?
Anyhoo, this would seem a bit hacky... is there a completely different way to do this that is better?
EDIT: The real reason I would like to do this is so that callbacks like the one in test2 will have references to the correct this.

What you can do is bind the current this value to a function and store a copy somewhere. (For the sake of efficiency.)
if (!Function.prototype.bind) {
// Most modern browsers will have this built-in but just in case.
Function.prototype.bind = function (obj) {
var slice = [].slice,
args = slice.call(arguments, 1),
self = this,
nop = function () { },
bound = function () {
return self.apply(this instanceof nop ? this : (obj || {}),
args.concat(slice.call(arguments)));
};
nop.prototype = self.prototype;
bound.prototype = new nop();
return bound;
};
}
function Test(n) {
this.n = n;
this.callback = (function () {
alert(this.n);
}).bind(this)
}
Test.prototype.test1 = function () {
this.test2();
}
Test.prototype.test2 = function () {
doSomething(this.callback);
}
function doSomething(callback) {
callback();
}
var t = new Test(2);
t.test1();

I realize your question was not tagged with jQuery, but you are using it in your example, so my solution also utilizes jQuery.
I sometimes use the $.proxy function to avoid callback context. Look at this simple jsfiddle example. Source below.
function Test(){
this.bind();
}
Test.prototype.bind = function(){
$('input').bind('change', $.proxy(this.change, this));
// you could use $.proxy on anonymous functions also (as in your $.getJSON example)
}
Test.prototype.change = function(event){
// currentField must be set from e.target
// because this is `Test` instance
console.log(this instanceof Test); // true
console.log(event.target == $('input')[0]); // true
this.currentField = event.target; // set new field
};
function createTest(){
return new Test();
}
$(function(){ // ready callback calls test factory
var t1 = createTest();
});

Most of the time, I just declare a local variable that references this, wherever I need a reference to this in a callback:
function Foo() {
}
Foo.prototype.bar = function() {
var that=this;
setTimeout(function() {
that.something="This goes to the right object";
}, 5000);
}
Alternatively, you can use bind() like this:
Function Foo() {
this.bar = this.bar.bind(this);
// ... repeated for each function ...
}
Foo.prototype.bar = function() {
}
What this gives you is that every time you create a new Foo instance, the methods are bound to the current instance, so you can use them as callback functions for setTimeout() et al.

Calling method using JavaScript prototype

Is it possible to call the base method from a prototype method in JavaScript if it's been overridden?
MyClass = function(name){
this.name = name;
this.do = function() {
//do somthing
}
};
MyClass.prototype.do = function() {
if (this.name === 'something') {
//do something new
} else {
//CALL BASE METHOD
}
};

I did not understand what exactly you're trying to do, but normally implementing object-specific behaviour is done along these lines:
function MyClass(name) {
this.name = name;
}
MyClass.prototype.doStuff = function() {
// generic behaviour
}
var myObj = new MyClass('foo');
var myObjSpecial = new MyClass('bar');
myObjSpecial.doStuff = function() {
// do specialised stuff
// how to call the generic implementation:
MyClass.prototype.doStuff.call(this /*, args...*/);
}

Well one way to do it would be saving the base method and then calling it from the overriden method, like so
MyClass.prototype._do_base = MyClass.prototype.do;
MyClass.prototype.do = function(){
if (this.name === 'something'){
//do something new
}else{
return this._do_base();
}
};

I'm afraid your example does not work the way you think. This part:
this.do = function(){ /*do something*/ };
overwrites the definition of
MyClass.prototype.do = function(){ /*do something else*/ };
Since the newly created object already has a "do" property, it does not look up the prototypal chain.
The classical form of inheritance in Javascript is awkard, and hard to grasp. I would suggest using Douglas Crockfords simple inheritance pattern instead. Like this:
function my_class(name) {
return {
name: name,
do: function () { /* do something */ }
};
}
function my_child(name) {
var me = my_class(name);
var base_do = me.do;
me.do = function () {
if (this.name === 'something'){
//do something new
} else {
base_do.call(me);
}
}
return me;
}
var o = my_child("something");
o.do(); // does something new
var u = my_child("something else");
u.do(); // uses base function
In my opinion a much clearer way of handling objects, constructors and inheritance in javascript. You can read more in Crockfords Javascript: The good parts.

I know this post is from 4 years ago, but because of my C# background I was looking for a way to call the base class without having to specify the class name but rather obtain it by a property on the subclass. So my only change to Christoph's answer would be
From this:
MyClass.prototype.doStuff.call(this /*, args...*/);
To this:
this.constructor.prototype.doStuff.call(this /*, args...*/);

if you define a function like this (using OOP)
function Person(){};
Person.prototype.say = function(message){
console.log(message);
}
there is two ways to call a prototype function: 1) make an instance and call the object function:
var person = new Person();
person.say('hello!');
and the other way is... 2) is calling the function directly from the prototype:
Person.prototype.say('hello there!');

This solution uses Object.getPrototypeOf
TestA is super that has getName
TestB is a child that overrides getName but, also has
getBothNames that calls the super version of getName as well as the child version
function TestA() {
this.count = 1;
}
TestA.prototype.constructor = TestA;
TestA.prototype.getName = function ta_gn() {
this.count = 2;
return ' TestA.prototype.getName is called **';
};
function TestB() {
this.idx = 30;
this.count = 10;
}
TestB.prototype = new TestA();
TestB.prototype.constructor = TestB;
TestB.prototype.getName = function tb_gn() {
return ' TestB.prototype.getName is called ** ';
};
TestB.prototype.getBothNames = function tb_gbn() {
return Object.getPrototypeOf(TestB.prototype).getName.call(this) + this.getName() + ' this object is : ' + JSON.stringify(this);
};
var tb = new TestB();
console.log(tb.getBothNames());

function NewClass() {
var self = this;
BaseClass.call(self); // Set base class
var baseModify = self.modify; // Get base function
self.modify = function () {
// Override code here
baseModify();
};
}

An alternative :
// shape
var shape = function(type){
this.type = type;
}
shape.prototype.display = function(){
console.log(this.type);
}
// circle
var circle = new shape('circle');
// override
circle.display = function(a,b){
// call implementation of the super class
this.__proto__.display.apply(this,arguments);
}

If I understand correctly, you want Base functionality to always be performed, while a piece of it should be left to implementations.
You might get helped by the 'template method' design pattern.
Base = function() {}
Base.prototype.do = function() {
// .. prologue code
this.impldo();
// epilogue code
}
// note: no impldo implementation for Base!
derived = new Base();
derived.impldo = function() { /* do derived things here safely */ }

If you know your super class by name, you can do something like this:
function Base() {
}
Base.prototype.foo = function() {
console.log('called foo in Base');
}
function Sub() {
}
Sub.prototype = new Base();
Sub.prototype.foo = function() {
console.log('called foo in Sub');
Base.prototype.foo.call(this);
}
var base = new Base();
base.foo();
var sub = new Sub();
sub.foo();
This will print
called foo in Base
called foo in Sub
called foo in Base
as expected.

Another way with ES5 is to explicitely traverse the prototype chain using Object.getPrototypeOf(this)
const speaker = {
speak: () => console.log('the speaker has spoken')
}
const announcingSpeaker = Object.create(speaker, {
speak: {
value: function() {
console.log('Attention please!')
Object.getPrototypeOf(this).speak()
}
}
})
announcingSpeaker.speak()

No, you would need to give the do function in the constructor and the do function in the prototype different names.

In addition, if you want to override all instances and not just that one special instance, this one might help.
function MyClass() {}
MyClass.prototype.myMethod = function() {
alert( "doing original");
};
MyClass.prototype.myMethod_original = MyClass.prototype.myMethod;
MyClass.prototype.myMethod = function() {
MyClass.prototype.myMethod_original.call( this );
alert( "doing override");
};
myObj = new MyClass();
myObj.myMethod();
result:
doing original
doing override

function MyClass() {}
MyClass.prototype.myMethod = function() {
alert( "doing original");
};
MyClass.prototype.myMethod_original = MyClass.prototype.myMethod;
MyClass.prototype.myMethod = function() {
MyClass.prototype.myMethod_original.call( this );
alert( "doing override");
};
myObj = new MyClass();
myObj.myMethod();