Written by Maciej Cieślar✏️
Proxy
is one of the most overlooked concepts introduced in the ES6 version of JavaScript.
Admittedly, it isn’t particularly useful on a day-to-day basis, but it is bound to come in handy at some point in your future.
The basics
The Proxy
object is used to define a custom behavior for fundamental operations such as property lookup, assignment, and function invocation.
The most basic example of a proxy would be:
const obj = {
a: 1,
b: 2,
};
const proxiedObj = new Proxy(obj, {
get: (target, propertyName) => {
// get the value from the "original" object
const value = target[propertyName];
if (!value && value !== 0) {
console.warn('Trying to get non-existing property!');
return 0;
}
// return the incremented value
return value + 1;
},
set: (target, key, value) => {
// decrement each value before saving
target[key] = value - 1;
// return true to indicate successful operation
return true;
},
});
proxiedObj.a = 5;
console.log(proxiedObj.a); // -> incremented obj.a (5)
console.log(obj.a); // -> 4
console.log(proxiedObj.c); // -> 0, logs the warning (the c property doesn't exist)
We have intercepted the default behavior of both get
and set
operations by defining the handlers with their respective names in the object provided to the proxy constructor. Now each get
operation will return the incremented value of the property, while set
will decrement the value before saving it in the target object.
What’s important to remember with proxies is that once a proxy is created, it should be the only way to interact with the object.
Different kinds of traps
There are many traps (handlers that intercept the object’s default behavior) aside from get
and set
, but we won’t be using any of them in this article. With that being said, if you are interested in reading more about them, here’s the documentation.
Having fun
Now that we know how proxies work, let’s have some fun with them.
Observing object’s state
As it has been stated before it is very easy to intercept operations with proxies. To observe an object’s state is to be notified every time there’s an assignment operation.
const observe = (object, callback) => {
return new Proxy(object, {
set(target, propKey, value) {
const oldValue = target[propKey];
target[propKey] = value;
callback({
property: propKey,
newValue: value,
oldValue,
});
return true;
}
});
};
const a = observe({ b: 1 }, arg => {
console.log(arg);
});
a.b = 5; // -> logs from the provided callback: {property: "b", oldValue: 1, newValue: 5}
And that’s all we have to do — invoke the provided callback every time the set
handler is fired.
As an argument to the callback
, we provide an object with three properties: the name of the changed property, the old value, and the new value.
Prior to executing the callback
, we assign the new value in the target object so the assignment actually takes place. We have to return true
to indicate that the operation has been successful; otherwise, it would throw a TypeError
.
Here’s a live example.
Validating properties on set
If you think about it, proxies are a good place to implement validation — they are not tightly coupled with the data itself. Let’s implement a simple validation proxy.
As in the previous example, we have to intercept the set
operation. We would like to end up with the following way of declaring data validation:
const personWithValidation = withValidation(person, {
firstName: [validators.string.isString(), validators.string.longerThan(3)],
lastName: [validators.string.isString(), validators.string.longerThan(7)],
age: [validators.number.isNumber(), validators.number.greaterThan(0)]
});
In order to achieve this, we define the withValidation
function like so:
const withValidation = (object, schema) => {
return new Proxy(object, {
set: (target, key, value) => {
const validators = schema[key];
if (!validators || !validators.length) {
target[key] = value;
return true;
}
const shouldSet = validators.every(validator => validator(value));
if (!shouldSet) {
// or get some custom error
return false;
}
target[key] = value;
return true;
}
});
};
First we check whether or not there are validators
in the provided schema for the property that is currently being assigned — if there aren’t, there is nothing to validate and we simply assign the value.
If there are indeed validators
defined for the property, we assert that all of them return true
before assigning. Should one of the validators return false
, the whole set
operation returns false, causing the proxy to throw an error.
The last thing to do is to create the validators
object.
const validators = {
number: {
greaterThan: expectedValue => {
return value => {
return value > expectedValue;
};
},
isNumber: () => {
return value => {
return Number(value) === value;
};
}
},
string: {
longerThan: expectedLength => {
return value => {
return value.length > expectedLength;
};
},
isString: () => {
return value => {
return String(value) === value;
};
}
}
};
The validators
object contains validation functions grouped by the type they should validate. Each validator on invocation takes the necessary arguments, like validators.number.greaterThan(0)
, and returns a function. The validation happens in the returned function.
We could extend the validation with all kinds of amazing features, such as virtual fields or throwing errors from inside the validator to indicate what went wrong, but that would make the code less readable and is outside the scope of this article.
Here’s a live example.
Making code lazy
For the final — and hopefully most interesting — example, let’s create a proxy that makes all the operations lazy.
Here’s a very simple class called Calculator
, which contains a few basic arithmetic operations.
class Calculator {
add(a, b) {
return a + b;
}
subtract(a, b) {
return a - b;
}
multiply(a, b) {
return a * b;
}
divide(a, b) {
return a / b;
}
}
Now normally, if we ran the following line:
new Calculator().add(1, 5) // -> 6
The result would be 6.
The code is executed on the spot. What we would like is to have the code wait for the signal to be run, like a run
method. This way the operation will be postponed until it is needed — or not executed at all if there is never a need.
So the following code, instead of 6, would return the instance of the Calculator
class itself:
lazyCalculator.add(1, 5) // -> Calculator {}
Which would give us another nice feature: method chaining.
lazyCalculator.add(1, 5).divide(10, 10).run() // -> 1
The problem with that approach is that in divide
, we have no clue of what the result of add
is, which makes it kind of useless. Since we control the arguments, we can easily provide a way to make the result available through a previously defined variable — $
, for example.
lazyCalculator.add(5, 10).subtract($, 5).multiply($, 10).run(); // -> 100
$
here is just a constant Symbol
. During execution, we dynamically replace it with the result returned from the previous method.
const $ = Symbol('RESULT_ARGUMENT');
Now that we have a fair understanding of what do we want to implement, let’s get right to it.
Let’s create a function called lazify
. The function creates a proxy that intercepts the get
operation.
function lazify(instance) {
const operations = [];
const proxy = new Proxy(instance, {
get(target, propKey) {
const propertyOrMethod = target[propKey];
if (!propertyOrMethod) {
throw new Error('No property found.');
}
// is not a function
if (typeof propertyOrMethod !== 'function') {
return target[propKey];
}
return (...args) => {
operations.push(internalResult => {
return propertyOrMethod.apply(
target,
[...args].map(arg => (arg === $ ? internalResult : arg))
);
});
return proxy;
};
}
});
return proxy;
}
Inside the get
trap, we check whether or not the requested property exists; if it doesn’t, we throw an error. If the property is not a function, we return it without doing anything.
Proxies don’t have a way of intercepting method calls. Instead, they are treating them as two operations: the get
operation and a function invocation. Our get
handler has to act accordingly.
Now that we are sure the property is a function, we return our own function, which acts as a wrapper. When the wrapper function is executed, it adds yet another new function to the operations array. The wrapper function has to return the proxy to make it possible to chain methods.
Inside the function provided to the operations array, we execute the method with the arguments provided to the wrapper. The function is going to be called with the result argument, allowing us to replace all the $
with the result returned from the previous method.
This way we delay the execution until requested.
Now that we have built the underlying mechanism to store the operations, we need to add a way to run the functions — the .run()
method.
This is fairly easy to do. All we have to do is check whether the requested property name equals run. If it does, we return a wrapper function (since run acts as a method). Inside the wrapper, we execute all the functions from the operations array.
The final code looks like this:
const executeOperations = (operations, args) => {
return operations.reduce((args, method) => {
return [method(...args)];
}, args);
};
const $ = Symbol('RESULT_ARGUMENT');
function lazify(instance) {
const operations = [];
const proxy = new Proxy(instance, {
get(target, propKey) {
const propertyOrMethod = target[propKey];
if (propKey === 'run') {
return (...args) => {
return executeOperations(operations, args)[0];
};
}
if (!propertyOrMethod) {
throw new Error('No property found.');
}
// is not a function
if (typeof propertyOrMethod !== 'function') {
return target[propKey];
}
return (...args) => {
operations.push(internalResult => {
return propertyOrMethod.apply(
target,
[...args].map(arg => (arg === $ ? internalResult : arg))
);
});
return proxy;
};
}
});
return proxy;
}
The executeOperations
function takes an array of functions and executes them one by one, passing the result of the previous one to the invocation of the next one.
And now for the final example:
const lazyCalculator = lazify(new Calculator());
const a = lazyCalculator
.add(5, 10)
.subtract($, 5)
.multiply($, 10);
console.log(a.run()); // -> 100
If you are interested in adding more functionality I have added a few more features to the lazify
function — asynchronous execution, custom method names, and a possibility to add custom functions through the .chain()
method. Both versions of the lazify
function are available in the live example.
Summary
Now that you have seen proxies in action, I hope that you could find a good use for them in your own codebase.
Proxies have many more interesting uses than those covered here, such as implementing negative indices and catching all the nonexistent properties in an object. Be careful, though: proxies are a bad choice when performance is an important factor.
Editor's note: Seeing something wrong with this post? You can find the correct version here.
Plug: LogRocket, a DVR for web apps
LogRocket is a frontend logging tool that lets you replay problems as if they happened in your own browser. Instead of guessing why errors happen, or asking users for screenshots and log dumps, LogRocket lets you replay the session to quickly understand what went wrong. It works perfectly with any app, regardless of framework, and has plugins to log additional context from Redux, Vuex, and @ngrx/store.
In addition to logging Redux actions and state, LogRocket records console logs, JavaScript errors, stacktraces, network requests/responses with headers + bodies, browser metadata, and custom logs. It also instruments the DOM to record the HTML and CSS on the page, recreating pixel-perfect videos of even the most complex single-page apps.
Try it for free.
The post Having fun with ES6 proxies appeared first on LogRocket Blog.
Top comments (1)
kind of hard to follow, but very powerful. I see why solidJS uses them for most things behind the scenes