A wrapper around node-fetch (or any other fetch-like function) that returns a single promise until it resolves.

Sometimes, you have to interface with an API that doesn't respond fast enough. Moreover, you might perform the same request multiple times. So:

• You overload the API with the same exact requests.
• You wait for additional time during the API response.

Return the same promise for the same exact requests until they resolve. This is more useful when you interface with stateless APIs, where you just consume data.

User (1) makes a request to the backend. The backend then performs a request to a third-party API and then before it resolves, user (2) makes another request to the backend. The backend then needs to perform the same request, as before, to the third-party API. With this package, instead of performing a new request, you can access and use the same promise for user's (1) request and have user (2) wait for the same request's resolution. This should shorten the wait time for user (2).

This API is a wrapper around node-fetch.

Install the module: $ npm i memoized-node-fetch

import memoizedNodeFetch from 'memoized-node-fetch';

const fetch = memoizedNodeFetch();

(async () => {
    const fetch1 = fetch('https://jsonplaceholder.typicode.com/todos/1');
    const fetch2 = fetch('https://jsonplaceholder.typicode.com/todos/1');

    // This should return true because both requests return the same promise.
    console.log(fetch1 === fetch2);

    const res1 = await fetch1;
    const res2 = await fetch2;

    console.log(await res1.json());
    console.log(await res2.json());
})();

No. This package only caches the promise until it resolves. After the promise resolves, it is removed from the cache, along with the data returned.

The parameters of the two fetch functions are compared (the url and the RequestOptions), the specific key used for comparing the requests is:

const key = stringToHash(url.toString() + JSON.stringify(options));

The parameters of the request are hashed and stored on a map.

Of course, you can use your own fetch like this:

function myOwnFetch(url: RequestInfo, options?: RequestInit | undefined): Promise<Response> {
    /* bla bla bla */
}

const fetch = memoizedNodeFetch(myOwnFetch);

/* Use the fetch... */

Yes! Each time you run the factory function, a new promise-cache is created.

No. For most cases, you shouldn't use this library instead of react-query or swr. Rather you could use it in tandem with those libraries by substituting the fetcher function with this. swr, although it implements caching, doesn't implement it while the fetch is loading (so if you perform the request two times, you'll get two different promises). react-query now does (v3+), but comes with a bunch of other setup and features you might not need. So if what you want is a simple way to not multi-ping with identical fetches, this package is for you.

1. I don't want to do request deduplication per-se but rather I want to return the same promise for each instance of the request. That won't work easily with the debounce function.
2. The debounce implementation of lodash/underscore, waits for a specific preset time before running the request. If my promise takes longer to resolve, then I wouldn't reap the benefits of it. In my case, I wait for the promise to resolve before making a duplicate request.

Thanks goes to these wonderful people (emoji key):

Nikos Polykandriotis 💻

Fernando van Loenhout 💻

Bonjour Comosava 📖

This project follows the all-contributors specification. Contributions of any kind welcome!