DEV Community

Cache API in JavaScript - with just 20 lines of code.

Rajesh Royal on May 15, 2022

let cache = {}; async function getData(url){ let result = ""; if(cache[url] !== undefined) return cache[url].value; await fetch(url...

Read full post

John P. Rouillard • May 15 '22

Just out of curiosity why are you calling:

let currentTime = new Date();

inside of .forEach? Why not assign it before the call:

 let currentTime = new Date();
 Object.keys(cache).forEach(key => {
            let seconds = currentTime - cache[key].time;
            ...

this way you only pay for the overhead of calling Date once.

I expect new Date() to be practically a constant (maybe 100ms difference) over the time spent clearing the cache. Since every three seconds you are clearing anything in the cache older than 10 seconds.

Calling new Date() IIUC is expensive since it has to reach down to the operating system to get the date value.

True some item might linger in the cache for 100 ms longer but....

Rajesh Royal • May 16 '22

Good correction, I will update the code.

Also I didn't write it to be optimised. I was just experimenting how caching works simply and pasted same code here 😀.

Dino Winchester • May 17 '22

It's good that you shared interesting material, but remember that not all developers know, are thinking and/or working on optimizing their scripts. There are beginners, and perhaps not only beginners, who, seeing interesting code at first glance, copy it into their applications and use it without checking or optimizing, and then other people may suffer from sudden errors that can come out absolutely by accident.

It is necessary to approach operations related to cycles carefully and try to think through the work of the function in advance, since even any small error can entail a bunch of others that can cause unnecessary memory costs or something else, perhaps. Lol.

I wish you good luck.

Rajesh Royal • May 17 '22

Sure brother.

Omid • May 19 '22

Can I ask you why are you doing this:

await fetch(url)
    .then(response => response.json())
    .then(json => cache[url] = {time: new Date(), value: json})

Instead of this:

const response = await fetch(url)
cache[url] = { time: new Date(), value: response.json()

If you use await you can replace those .then and I think this is much more cleaner :D

Rajesh Royal • May 21 '22

Good Improvement. I will update the code.

Omid • May 21 '22 • Edited

I think you can append .json() to this:

const response = await fetch(url).json()

(I didn't test it but I think it should work)
Happy to help you! :D

Rajesh Royal • May 22 '22

No it cannot be done.

Gohomewho • May 23 '22

Actually that doesn't work and you'll get an error.
Uncaught TypeError: fetch(...).json is not a function
Because fetch() return a promise, and it doesn't have a json method on it.

It should be like this.

const url = "https://jsonplaceholder.typicode.com/todos/1"
const data = await fetch(url).then((response)=>response.json())

or this

const response = await fetch(url)
const data = await response.json()

Note that calling response.json() also return a promise so you need to await it to get the result.

Eyal Shalev • May 20 '22 • Edited

Note: if you need caching for production projects, I highly recommend searching for existing implementations (or at least existing algorithms).

Otherwise I have some notes:

The current implementation only accepts the URL. But usually the same URL is used with different parameters, http methods or headers. Because not all of the possible keys can be mapped, I recommend accepting a key, and a function (to be used if the key is missing).
Some data will need to be cached for different amount of time, so accepting a parameter for it is a good idea. That would mean using a setInterval for each key, or replacing the cache[key].time with a validUntil time object.
Because you are deleting the cache after a fixed point of time, there is no need to keep the cache centralized, and it will be easier to just implement a memoize function that deleted the cache after a certain amount of time.

function memoize(fn, delAfter = 1000000) {
  const missing = Symbol("missing");
  let cache = missing;
  return async () => {
    if (cache === missing) {
       cache = Promise.resolve(fn());
       setTimeout(() => cache = missing, delAfter);
    }
    return await cache;
  }
}
const foo = memoize(() => fetch("https://foo/bar", { method: " POST", body: JSON.stringify({bar:1}) }))

// Will only do a single fetch.
console.log(await foo(), await foo())

Rajesh Royal • May 21 '22

restep++

Jose Gascon • May 21 '22 • Edited

Nice post, thanks for sharing.
What do you think about integrating the clearing of the cache in the getData function?
Performance wise it should be very similar and the result is even more compact (just 10 lines 😄)

const cache = {};

async function getData(url){
  const currentTime = new Date();
  if(cache[url] !== undefined) {
    if((currentTime - cache[url].time) < 10000) return cache[url].value;
    delete cache[url];
  }
  await fetch(url)
    .then(response => response.json())
    .then(json => cache[url] = {time: currentTime, value: json});
  return cache[url].value;
}

Rajesh Royal • May 22 '22

➕➕

Shikhar • May 16 '22

Nice idea to use Memoized results.Thanks!!

Ian Klek • May 20 '22 • Edited

First rule of software development: see if it's been done before - don't reinvent the wheel. Check out the Cache interface, which would provide you with caching across browser sessions: developer.mozilla.org/en-US/docs/W...