JavaScript的this,call(),apply(),bind()

为了建立一个scope chain, 每个JavaScript的代码执行上下文都提供了this关键字。In its most common usage, this serves as an identity function, providing our neighborhoods a way of referring to themselves. We can’t always rely on that behavior, however: Depending on how we get into a particular neighborhood, this might mean something else entirely. In fact, how we get into the neighborhood is itself exactly what this generally refers to. 需要注意特殊的四种情况:

• Calling an Object’s Method

  在典型的面向对象编程时，我们需要一种方式去指向和引用我们调用的对象. this serves the purpose admirably, providing our objects the ability to examine themselves, and point at their own properties.

   

  var deep_thought = {
     the_answer: 42,
     ask_question: function () {
      return this.the_answer;
     }
    };
  
    var the_meaning = deep_thought.ask_question();

  var deep_thought = { the_answer: 42, ask_question: function () { return this.the_answer; } }; var the_meaning = deep_thought.ask_question(); 


   

  This example builds an object named deep_thought, sets its the_answer property to 42, and creates an ask_question method. When deep_thought.ask_question() is executed, JavaScript establishes an execution context for the function call, setting this to the object referenced by whatever came before the last ”.”, in this case: deep_thought. The method can then look in the mirror via this to examine its own properties, returning the value stored in this.the_answer: 42.

• Constructor

  Likewise, when defining a function to be used as a constructor with the new keyword, this can be used to refer to the object being created. Let’s rewrite the example above to reflect that scenario:

   

  <mce:script type="text/javascript"><!--
    function BigComputer(answer) {
     this.the_answer = answer;
     this.ask_question = function () {
      return this.the_answer;
     }
    }
  
    var deep_thought = new BigComputer(42);
    var the_meaning = deep_thought.ask_question();
  
  // --></mce:script>

  <mce:script type="text/javascript"><!-- function BigComputer(answer) { this.the_answer = answer; this.ask_question = function () { return this.the_answer; } } var deep_thought = new BigComputer(42); var the_meaning = deep_thought.ask_question(); // --></mce:script>


   

  Instead of explicitly creating the deep_thought object, we’ll write a function to create BigComputer objects, and instantiate deep_thought as an instance variable via the new keyword. When new BigComputer() is executed, a completely new object is created transparently in the background. BigComputer is called, and its this keyword is set to reference that new object. The function can set properties and methods on this, which is transparently returned at the end of BigComputer’s execution.

  Notice, though, that deep_thought.the_question() still works just as it did before. What’s going on there? Why does this mean something different inside the_question than it does inside BigComputer? Put simply, we entered BigComputer via new, so this meant “the new object.” On the other hand, we entered the_question via deep_thought, so while we’re executing that method, this means “whatever deep_thought refers to”. this is not read from the scope chain as other variables are, but instead is reset on a context by context basis.

• Function Call

  What if we just call a normal, everyday function without any of this fancy object stuff? What does this mean in that scenario?

  <mce:script type="text/javascript"><!--
    function test_this() {
     return this;
    }
    var i_wonder_what_this_is = test_this();
  
  // --></mce:script>

  <mce:script type="text/javascript"><!-- function test_this() { return this; } var i_wonder_what_this_is = test_this(); // --></mce:script> 

   

   

  In this case, we weren’t provided a context by new, nor were we given a context in the form of an object to piggyback off of. Here, this defaults to reference the most global thing it can: for web pages, this is the window object.

• Event Handler

  For a more complicated twist on the normal function call, let’s say that we’re using a function to handle an onclick event. What does this mean when the event triggers our function’s execution? Unfortunately, there’s not a simple answer to this question.

  If we write the event handler inline, this refers to the global window object:

   

   <mce:script type="text/javascript"><!--
    function click_handler() {
     alert(this); // alerts the window object
    }
  
  // --></mce:script>

  <mce:script type="text/javascript"><!-- function click_handler() { alert(this); // alerts the window object } // --></mce:script> 
 ...
 

  <button id='thebutton' onclick='click_handler()'>Click me!</button>

  <button id='thebutton' onclick='click_handler()'>Click me!</button> 

   

  However, when we add an event handler via JavaScript, this refers to the DOM element that generated the event. (Note: The event handling shown here is short and readable, but otherwise poor. Please use a real addEvent function instead.):

   <script type="text/javascript">
  

  <mce:script type="text/javascript"><!--
    function click_handler() {
     alert(this); // alerts the button DOM node
    }
  
    function addhandler() {
     document.getElementById('thebutton').onclick = click_handler;
    }
  
    window.onload = addhandler;
  
  // --></mce:script>

  <mce:script type="text/javascript"><!-- function click_handler() { alert(this); // alerts the button DOM node } function addhandler() { document.getElementById('thebutton').onclick = click_handler; } window.onload = addhandler; // --></mce:script> 
 ...
 

  <button id='thebutton'>Click me!</button>

  <button id='thebutton'>Click me!</button> 

   

Complications

Let’s run with that last example for a moment longer. What if instead of running click_handler, we wanted to ask deep_thought a question every time we clicked the button? The code for that seems pretty straightforward; we might try this:

<mce:script type="text/javascript"><!--
 function BigComputer(answer) {
  this.the_answer = answer;
  this.ask_question = function () {
   alert(this.the_answer);
  }
 }

 function addhandler() {
  var deep_thought = new BigComputer(42),
   the_button = document.getElementById('thebutton');

  the_button.onclick = deep_thought.ask_question;
 }

 window.onload = addhandler;
// --></mce:script>

<mce:script type="text/javascript"><!-- function BigComputer(answer) { this.the_answer = answer; this.ask_question = function () { alert(this.the_answer); } } function addhandler() { var deep_thought = new BigComputer(42), the_button = document.getElementById('thebutton'); the_button.onclick = deep_thought.ask_question; } window.onload = addhandler; // --></mce:script> 

 

对上面的代码，我们期望点击按钮, deep_thought.ask_question被执行，我们得到返回结果“42.” 但为什么得到的结果反而是undefined？哪里错了?

The problem is simply this: We’ve passed off a reference to the ask_question method, which, when executed as an event handler, runs in a different context than when it’s executed as an object method. 简而言之，ask_question 中的this关键字是指向产生事件的DOM元素节点，而不是BigComputer对象. DOM元素节点并没有the_answer属性，所以返回结果是undefined而不是“42.” setTimeout exhibits similar behavior, delaying the execution of a function while at the same time moving it out into a global context.

This issue crops up all over the place in our programs, and it’s a terribly difficult problem to debug without keeping careful track of what’s going on in all the corners of your program, especially if your object has properties that do exist on DOM elements or the window object.

Manipulating Context With .apply() and .call()

We really do want to be able to ask deep_thought a question when we click the button, and more generally, we do want to be able to call object methods in their native context when responding to things like events and setTimeout calls. Two little-known JavaScript methods, apply and call, indirectly enable this functionality by allowing us to manually override the default value of this when we execute a function call. Let’s look at call first:

<mce:script type="text/javascript"><!--
 var first_object = {
  num: 42
 };
 var second_object = {
  num: 24
 };

 function multiply(mult) {
  return this.num * mult;
 }

 multiply.call(first_object, 5); // returns 42 * 5
 multiply.call(second_object, 5); // returns 24 * 5
// --></mce:script>

<mce:script type="text/javascript"><!-- var first_object = { num: 42 }; var second_object = { num: 24 }; function multiply(mult) { return this.num * mult; } multiply.call(first_object, 5); // returns 42 * 5 multiply.call(second_object, 5); // returns 24 * 5 // --></mce:script> 

 

In this example, we first define two objects, first_object and second_object, each with a num property. Then we define a multiply function that accepts a single argument, and returns the product of that argument, and the num property of its this object. If we called that function by itself, the answer returned would almost certainly be undefined, since the global window object doesn’t have a num property unless we explicitly set one. We need some way of telling multiply what its this keyword ought refer to; the call method of the multiply function is exactly what we’re looking for.

 call方法的第一个参数定义了this关键字在被调用方法的执行上下文中指向和对象，call方法的剩余参数则是被调用方法的参数。因此当multiply.call(first_object, 5)被执行， multiply函数被调用, 5 为传入方法的第一个参数， this 执行 first_object对象。 Likewise, when multiply.call(second_object, 5) is executed, the multiply function is called, 5 is passed in as the first argument, and the this keyword is set to refer to object second_object.

apply方法和 call方法基本一致,但是允许你以数组的形式向被调用的函数传递参数, which can be quite useful when programatically generating function calls. Replicating the functionality we just talked about using apply is trivial:

<mce:script type="text/javascript"><!--
 ...

 multiply.apply(first_object, [5]); // returns 42 * 5
 multiply.apply(second_object, [5]); // returns 24 * 5
// --></mce:script>

<mce:script type="text/javascript"><!-- ... multiply.apply(first_object, [5]); // returns 42 * 5 multiply.apply(second_object, [5]); // returns 24 * 5 // --></mce:script> 

 

apply and call are very useful on their own, and well worth keeping around in your toolkit, but they only get us halfway to solving the problem of context shifts for event handlers. It’s easy to think that we could solve the problem by simply using call to shift the meaning of this when we set up the handler:

function addhandler() {
 var deep_thought = new BigComputer(42),
  the_button = document.getElementById('thebutton');

 the_button.onclick = deep_thought.ask_question.call(deep_thought);
}

function addhandler() { var deep_thought = new BigComputer(42), the_button = document.getElementById('thebutton'); the_button.onclick = deep_thought.ask_question.call(deep_thought); } 

 

上面的代码仍然存在问题: call是立即执行函数的，因此我们提供的 onclick handler是函数的执行结果而不是函数本身.我们需要JavaScript的另一个特性来解决这个问题：bind方法。

The Beauty of .bind()

I’m not a huge fan of the Prototype JavaScript framework, but I am very much impressed with the quality of its code as a whole. In particular, one simple addition it makes to the Function object has had a hugely positive impact on my ability to manage the context in which function calls execute: bind performs the same general task as call, altering the context in which a function executes. The difference is that bind returns a function reference that can be used later, rather than the result of an immediate execution that we get with call.

If we simplify the bind function a bit to get at the key concepts, we can insert it into the multiplication example we discussed earlier to really dig into how it works; it’s quite an elegant solution:

<mce:script type="text/javascript"><!--
 var first_object = {
  num: 42
 };
 var second_object = {
  num: 24
 };

 function multiply(mult) {
  return this.num * mult;
 }

 Function.prototype.bind = function(obj) {
  var method = this,
   temp = function() {
    return method.apply(obj, arguments);
   };

  return temp;
 }

 var first_multiply = multiply.bind(first_object);
 first_multiply(5); // returns 42 * 5

 var second_multiply = multiply.bind(second_object);
 second_multiply(5); // returns 24 * 5
// --></mce:script>

<mce:script type="text/javascript"><!-- var first_object = { num: 42 }; var second_object = { num: 24 }; function multiply(mult) { return this.num * mult; } Function.prototype.bind = function(obj) { var method = this, temp = function() { return method.apply(obj, arguments); }; return temp; } var first_multiply = multiply.bind(first_object); first_multiply(5); // returns 42 * 5 var second_multiply = multiply.bind(second_object); second_multiply(5); // returns 24 * 5 // --></mce:script> 

 

First, we define first_object, second_object, and the multiply function, just as before. With those taken care of, we move on to creating a bind method on the Function object’s prototype, which has the effect of making bind available for all functions in our program. When multiply.bind(first_object) is called, JavaScript creates an execution context for the bind method, setting this to the multiply function, and setting the first argument, obj, to reference first_object. So far, so good.

The real genius of this solution is the creation of method, set equal to this (the multiply function itself). When the anonymous function is created on the next line, method is accessible via its scope chain, as is obj (this couldn’t be used here, because when the newly created function is executed, this will be overwritten by a new, local context). This alias to this makes it possible to use apply to execute the multiply function, passing in obj to ensure that the context is set correctly. In computer-science-speak, temp is a closure that, when returned at the end of the bind call, can be used in any context whatsoever to execute multiply in the context of first_object.

This is exactly what we need for the event handler and setTimeout scenarios discussed above. The following code solves that problem completely, binding the deep_thought.ask_question method to the deep_thought context, so that it executes correctly whenever the event is triggered:

function addhandler() {
 var deep_thought = new BigComputer(42),
  the_button = document.getElementById('thebutton');

 the_button.onclick = deep_thought.ask_question.bind(deep_thought);
}

function addhandler() { var deep_thought = new BigComputer(42), the_button = document.getElementById('thebutton'); the_button.onclick = deep_thought.ask_question.bind(deep_thought); } 

 

Beautiful.