writing a function queue in javascript - javascript

I have an object which can only place 60 API calls per minute. So what I would like to do, is when a function call comes that I know I wont be allowed to place, add it to a queue, and call the function again at a more convenient time.
Heres how I thought to fix it
var API_caller = function(){
this.function_queue = [];
};
API_caller.prototype.make_api_call = function(){
if(this.can_make_call()){
this.make_call();
} else {
// If I cant place an API call then add
// the function to the function queue
this.function_queue.push(this.make_api_call);
}
};
API_caller.prototype.queue_call = function(){
// remove function from queue and call it
var func = this.function_queue.shift();
func();
}
This works fine for functions without parameters but what if make_api_call() had a parameter
API_caller.prototype.make_api_call = function(data){
if(this.can_make_call()){
this.make_call();
} else {
// If I cant place an API call then add
// the function to the function queue
this.function_queue.push(this.make_api_call(data));
}
};
In this case however, make_api_call(data) will be evaluated before it is pushed to function_queue and func will no longer hold a function causing queue_call() to error.
How can I get around this?

You can partially apply arguments to a function with bind:
this.function_queue.push(this.make_api_call.bind(this, data));
Check MDN for support in old browsers.

The queue entry should contain the function, f and the parameters as an array, p.
When you add to the queue you would do something like queue.push ([f, arguments]) and when the time comes to make that call it would be something like queue[0][0].apply (null, queue[0][1])

You can queue up an ad-hoc function that contains your API call with the argument already bound:
var that = this;
this.function_queue.push(function() {
that.make_api_call(data);
));
The aliasing of this to that is required, because inside the anonymous function, this would not be bound to the same object as outside.
Note this technique is similar to eclanrs' answer, but doesn't rely on bind method's availability.

Related

proper function calling etiquette?

I may be wording this title wrong but in javascript is it ok to call a nested function like so, if not why and what are some safer or more proper ways
function foo() {
return function poo() {
console.log("ew");
}
}
var fooPoo = foo()();
Yes, that's fine, and fairly normal, if you want poo to have access to information that's private within foo and you don't want the calling code to have access to that information. Or even just if foo is what knows how to create the poo function, even if private information isn't needed.
It's relatively rare to do it all in one expression, because usually when you return a function from another function, you want to keep the function around:
var p = foo();
var fp1 = p();
var fp2 = p();
...but only relatively unusual, not unusual.
Here's an example of using the private information held by the context of the original call to the function (allocator, here, is like your foo):
function allocator(seed) {
return function() {
return seed++;
};
}
var a = allocator(1);
console.log(a()); // 1
console.log(a()); // 2
console.log(a()); // 3
Note that the code calling a can't manipulate seed directly. It can only call a and use the value it returns.
Yes, it as a functional technique referred to as currying. it allows you to set parameters for the function in different places in your code
function foo(param1) {
return function poo(param2) {
console.log(param1, param2);
}
}
var fooPoo = foo('param1')('param2');
A common thing I do is use currying for passing in settings when running event listeners to allow greater reuse of functions
function setColor(color) {
return function (e) {
e.target.background = color
}
}
someElement.addEventLister('click', setColor('red'))
Here you can pass in your configuration when declaring your event listener but it won't be called until later when the event is fired and due to the closure you will have access to the color variable within the event listener callback. But now that I know the technique I use it quite a bit

JS asynchronous function in a loop

I've got some experience in PHP, but I'm starting out with javascript and jquery. I'm working on my first project. I thought that scripting is scripting, and there will be little difference between this and PHP. Well was I wrong. For the first time I saw that something which is first in the code executes last!
Please have a look at this function which is meant to get svg and store them in json object to use as inline svg later
var svgIcons = { "arrow_left": "", "arrow_right":"", } //json object with empty values
this.getIcons = function() {
for (var icon_name in svgIcons) {
if (svgIcons.hasOwnProperty(icon_name)) {
var url=PHP.plugin_url+'/includes/icons/'+icon_name+'.svg';
jQuery.get(url, function(data) {
svgIcons[icon_name]=data;
console.log('iterating');
console.log(svgIcons[icon_name]); //outputs svg
});
}
}
console.log('this should be after iteration');
console.log(svgIcons["arrow_left"]); //empty
}
this.getIcons(); //called at object initialization
But the output is:
this should be after iteration
iterating
#document (and svg inside it)
iterating
#document (and svg inside it)
What is the cause of this change of order? Is it the get() function? How do I avoid situations like this?
jQuery.get is asynchronous. You are iterating inside the callback for an AJAX call, so that gets executed whenever the AJAX call is completed.
AJAX callbacks, setTimeout and setInterval are some asynchronous Javascript functions. Some threads you might find useful:
How does Asynchronous Javascript Execution happen?
Are all javascript callbacks asynchronous? If not, how do I know which are?
Edit: Yes, the function call ends before any of the callback stuff happens. Basically the execution of your JS will be linear, placing functions on a call stack whenever they are called. On the call-stack they are executed one-by-one, line-by-line. However, when one of those lines calls an asynchronous function (like a setTimeout or AJAX), the current execution places the async function on the call-stack and immediately returns to complete itself. So something like:
function myFunc(){
console.log('a');
setTimeout(function(){
console.log('b');
},0)
console.log('c');
}
myFunc();
would always log:
a
c
b
...even though the setTimeout is 0.
So, in your case what must be happening is that you are assigning the AJAX-received data to svgIcons[icon_name] inside the async callback (obviously), while the rest of your code which uses the object svgIcons is in the sequential/normal execution. You either have to move the code that uses the object inside the async callback, or use promises (basically promises are functions that are executed after an async call is completed).
2nd Edit: So, the reason you are not able to set svgIcons[icon_name] inside the callback is related to the things I was mentioning in my comment. When synchronous functions are called, they are placed on top of the current stack and executed right away, before returning to the calling function. So if you called a sync function inside a loop:
function outer(){
function inner(){
console.log(i);
}
for(var i=0;i<3;i++)
inner();
}
outer();
the synchronous inner function would be executed right away inside each loop, and would have access to the current value of i, so it would output 0, 1, 2 (as expected).
If however, inner was asynchronous, e.g
function outer(){
for (var i=0;i<3;i++)
setTimeout(function(){console.log(i)},0);
}
Then you would get 3, 3, 3 as the output!
This is because the loop has already finished, including the final i++.
So now I think you can see the problem with your code. Upto calling jQuery.get you have access to the current value of icon_name, but once we are inside that asynchronous callback, the current value disappears and is replaced by the last value for it, because the loop already completed before any of the callbacks were executed.
Try something like this:
var svgIcons = {}
var props = ["arrow_left","arrow_right"];
this.getIcons = function() {
props.forEach(function(prop){
var url=PHP.plugin_url+'/includes/icons/'+prop+'.svg';
jQuery.get(url, function(data) {
svgIcons[prop]=data;
var fullyLoaded = false;
for(var i=0;i<props.length;i++) {
if(!svgIcons.hasOwnProperty(props[i])){
fullyLoaded = false;
break;
}
else fullyLoaded = true;
} // end for loop
if(fullyLoaded)
callMyFunctionWhereIUseSvgIconsData();
}); //end jQuery.get()
});//end forEach
}
this.getIcons()
This uses the forEach method, which is native to arrays (MDN reference). Inside the function passed to forEach, the first argument is always the current element of the array (which I named as prop). So there is no messy loop or i, and every executing function has access to its own prop property.
Then, inside the AJAX callback, I assign the current prop to the data received, and then loop through all the properties to check if the svgIcons object has received the properties. So fullyLoaded will only evaluate to true once all the callbacks have been executed and the global svgIcons has received all the properties and data. Hence, you can now call the function that uses the object.
Hope this helps, feel free to ask further or let me know if the console throws errors.
Any ajax calls are async therefore it can be run while the ajax call is taking place. If you want to call something after all calls are done then try this:
var svgIcons = { "arrow_left": "", "arrow_right":"", } //json object with empty values
var executing = 0;
this.getIcons = function() {
for (var icon_name in svgIcons) {
//store that this call has started
exectuing = executing + 1;
if (svgIcons.hasOwnProperty(icon_name)) {
var url=PHP.plugin_url+'/includes/icons/'+icon_name+'.svg';
console.log('this will run as you were expecting');
//this ajax call is then started and moves to next iteration
jQuery.get(url, function(data) {
//This is run after the ajax call has returned a response, not in the order of the code
svgIcons[icon_name]=data;
console.log('iterating');
console.log(svgIcons[icon_name]); //outputs svg
//if you want to call a function after evey call is comeplete then ignore the 'executing' part and just call the function here.
//decrement value as this call has finished
executing = executing - 1;
//if all have finished then call the function we want
if(executing === 0){
executeAfter();
}
});
}
}
console.log('this should be after iteration');
console.log(svgIcons["arrow_left"]); //empty
}
this.executeAfter(){
//This will be exectued after all of you ajax calls are complete.
}
this.getIcons(); //called at object initialization

Getting function arguments within another function

I've heard that it's possible to do something like this
this[func].apply(this, Array.prototype.slice.call(arguments, 1));
But to have access to the arguments object I need to be inside that function.
So, if I am, for example, running a code in the function function1, is there a way to get the arguments of a different function function2?
The whole problem is that I want to add events to any given element. But this element might already have another event attached to it. so, if I have, for example, an 'onclick' event attached to an element, I would do the following:
var event = 'onclick';
if(typeof currentElement[event] === 'function'){
cf = currentElement[event];
f = function(){
cf();
func();
}
}else f = func;
currentElement[event] = f;
Now both functions, the new function and the previous function, are being called. The problem is that the arguments being passed to the previous function were lost with this method.
Does anyone know if it is possible to not lose the arguments when we call a function dynamically like in my example?
OBS: jQuery is not an option :/
Sounds like using addEventListener would work better for you. It lets you attach multiple listeners (functions) to a single element:
elem.addEventListener("click", function() {
console.log("click listener 1");
});
elem.addEventListener("click", function() {
console.log("click listener 2");
});
Note, according to MDN, addEventListener is supported in IE >= 9.
If you did want to continue down your current path, you could do:
f = function(){
cf.apply(this, arguments);
func.apply(this, arguments);
}
For you specific case, it's not necessary to figure out what arguments are passed to the function since you know what it is - the event object in all browsers except (older) IE where its nothing (undefined).
So the code can simply be:
var event = 'onclick';
if(typeof currentElement[event] === 'function'){
cf = currentElement[event];
f = function(e){
cf(e);
func(e);
}
} else f = func;
currentElement[event] = f;
If possible use addEventListner and attachEvent as fallback to play nicely with other scripts that may run on your page.
Additional discussion:
In general, it is never necessary to figure out what has been passed to the callback function because that doesn't even make sense. The programmer cannot determine what to pass to the callback, it's the event emitter that determines what gets passed. In this case its the event object but it's a general rule.
Say for example we have an API that queries our database and returns some data:
function my_api (query) {}; // returns an object where a callback may
// be attached to the oncomplete property
The API documentation mentions:
my_api() returns an API object.
API object - has one property - oncomplete where a callback may be attached to handle data returned from the api call. The callback will be called with one argument - the returned data or no argument (undefined) if an error occured.
OK. So we use it like this:
var db = my_api('get something');
db.oncomplete = function (data) {alert(data)};
Now, if we want to wrap that oncomplete event in another event handler, we don't need to know what data it accepts since it's the API object that determines what to pass to the function, not the function! So we simply:
var tmp = db.oncomplete;
db.oncomplete = function (x) {
new_callback(x);
tmp(x);
}
We get the x argument not from querying the previous callback but from the documentation.
somefunction.arguments //Use the arguments property to get arguments of another function

JS bind the context of a function before passing

I'm trying to setup a function that stores a callback which can be fired later on an event, but I'm not getting the context to bind right. The setup is basically like this:
app.myObject = function(){
this.bindEvents();
};
app.myObject.prototype = {
bindEvents: function(){
var self = this;
this.library = new OutsideLibrary();
this.library.onMyEvent = function(data) {
OtherLibrary.processData(data, self.runCallback); // I'm not able to pass a context here...
}
},
sendToLibrary: function(message,callback) {
this.callback = callback;
this.library.send(message);
}
// ... and this doesn't work because it gets evaluated in the wrong context.
runCallback: function() {
if (this.callback) {
this.callback();
this.callback = null;
}
}
}
I hope the above code makes sense. The idea is, I'm trying to expose an API where I can call:
app.myObject.sendToLibrary("something", function() {...});
...and the callback function will run after the whole chain is complete.
The problem is, I can't pass the callback that is received in sendToLibrary directly through, because the callback need to be passed to another object when an event fires. Also since I'm having to store the currently passed callback, and the callback is optional, I need to unset it once it has fired, so I actually need to pass the runCallback function to run the callback once then unset it.
Anyway, the problem is that when runCallback eventually gets executed it is no longer in the correct context, so it doesn't work. My first idea for how to fix this was to change the runCallback function to include a context argument, so I could pass self.runCallback(self) -- but that executes immediately so it defeats the purpose.
Anyway, I'm sort of lost in the callback woods here, and would appreciate some help finding my way out :)
My question is: is there a way to wrap up a callback function and the correct context, without executing it now, so that when it eventually fires it does so from the correct context?
Thanks!
OtherLibrary.processData(data, function() {
self.runCallback();
)};
Then inside of runCallback this refers to an instance of app.myObject, if this is what you want.
Andrew, I think you might be trying a little too hard, here.
If all you're looking to do is bind context, you can do one of the following:
var working_method = context.method.bind(context);
// bind is IE9+, and does something similar to the below
// or, from whatever scope your context/method are available in, define this function
var working_method = function () { context.method(); };
If you require parameters, and you already know the footprint:
var working_method = function (a, b, c) { context.method(a, b, c); };
The solution for variable argument lengths (ie: a generalized "bind" method) is longer, but if you're defining stuff in line, what I've got here might be all you need.
Just make sure that you do not use this in your wrapping function (whether you save it to a var, or pass it directly into another function), because it will reference window.

can you say this is a right example of Javascript Closure.. Where the places we need to consider avoiding the closures?

Problem & Reason
One of my team mate ended up in messy situtaion implementing function hooking in javascript. this is the actual code
function ActualMethod(){
this.doSomething = function() {
this.testMethod();
};
this.testMethod = function(){
alert("testMethod");
};
}
function ClosureTest(){
var objActual= new ActualMethod();
var closeHandler = objActual.doSomething;
closeHandler();
closeHandler.apply(objActual,arguments); //the fix i have added
this.ActualTest = function() {
alert("ActualTest");
};
}
In the above code, var closeHandler is created in the context of ClosureTest(), but it holds the handler of the ActualMethod.doSomething. Whenever calling the closeHandler() ended up in "object doesnt support this method" error.
This is because doSomething() function calls another method inside called this.testMethod();. Here this refers to the context of the caller not callee.so i assume the closeHandler is bound to the environment(ClosureTest) actually created.Even though it holds the handler to the another context, it just exposes the properties of its own context.
Solution
To avoid this i suggest to use apply to specify the conext in which it needs to execute.
closeHandler.apply(objActual,arguments);
Questions
is it perfect scenario for closures..??
What are the intersting places you have encountered closures in javascript..?
UPDATE
Yes its simple i can call the method directly. but the problem is, in a particular scenario I need to intercept the call to actuall method and run some code before that, finally execute the actual method..
say for an example, am using 3rd party aspx grid library, and all the mouseclick events are trapped by their controls. In particular group by mouse click i need to intercept the call to their ilbrary method and hook my mthod to execute instead and redirect the call to actual library method
hope this helps
Update: Because you probably left out some details in your code, it is difficult to adapt it into something workable without missing the point of your actual code. I do think I understand your underlying problem as you describe it. I hope the following helps.
Suppose the following simple example:
// Constructor function.
function Example() {
// Method:
this.method = function() {
alert("original method");
}
}
// You would use it like this:
var obj = new Example();
obj.method(); // Calls original method.
To intercept such a method call, you can do this:
function wrap(obj) {
var originalMethod = obj.method;
obj.method = function() {
alert("intercepted call");
originalMethod.apply(this, arguments);
}
return obj;
}
var obj = wrap(new Example());
obj.method(); // Calls wrapped method.
Unfortunately, because method() is defined in the constructor function, not on a prototype, you need to have an object instance to wrap the object.
Answer to original question: The doSomething() function is used as a method on objects created with ActualMethod(). You should use it as a method, not detach it and use it as a function in a different context. Why don't you just call the method directly?
function ClosureTest(){
var objActual = new ActualMethod();
// Call method directly, avoid messy apply() calls.
objActual.doSomething();
this.ActualTest = function() {
alert("ActualTest");
};
}
If you assign a method (a function on some object) to a local variable in Javascript and call it, the context will be different (the value of this changes). If you don't want it to happen, don't do it.
When I want to hook a function, I use the following Function method which is also a fine piece of Closure demonstration:
Function.prototype.wrap = function (wrapper) {
var __method = this;
return function() {
var __obj = this;
var args = [ __method.bind(__obj) ];
for(var i=0; i<arguments.length; i++) args.push(arguments[i]);
return wrapper.apply(__obj, args);
}
};
Then do something like:
ActualMethod = ActualMethod.wrap(function (proceed, option) {
// ... handle option
proceed(); // calls the wrapped function
});
proceed is bound to its initial object, so you can safely call it.

Categories