/mementojs

💾 Execute request, cache it and serve it again

Primary LanguageTypeScriptMIT LicenseMIT

Travis npm npm

mementojs

A lightweight and dependency-free cache system for manager like a downloader, texture renderer, ect... Execute request, store the data and re-serve it with at next request.

You can use it on node, browser in JS or TS :)

Explanations

  • If you make a Downloader (own need), you will wait during the first request during download. But at next request, it will be more fast because data is already downloaded and cached.
  • Multiple request on same object can be made but only one download will be launch. However, all listeners will be notified when first request will be done.

Installation

Install by npm

npm install --save mementojs

or install with yarn

yarn add mementojs

Usage

Before all, you need to extend the class Memento (extend only) and implement the load function.

export class JSONDownloader extends Memento<HTMLImageElement> {
    
    // Load function need to be implemented ! 
    load(id: string, request: Request<string>): void {
        
        // Execute the first request
        fetch('http://all-city-jsons.com/' + id + '.json').then(result: string => {

            // Store the data in the cache & notify every listeners
            request.setData(result);
        }).catch(() => {
            // Store the error, preventing next loading
            request.setError("Big fail, sorry, we can't download your file !")
        });
    }
}

Now, in your main code, you can just make request like this:

let jsonDownloader = new JSONDownloader();

jsonDownloader.request("json-city", (data: HTMLImageElement) => {
    // Get your data here ! 
}, error => {
    // Get your error here too ;) !
});

jsonDownloader.request("json-city", (data: HTMLImageElement) => {
    // Get your data again, without waiting because it's already loaded :) in the memento class !
});

jsonDownloader.request("json-country", (data: HTMLImageElement) => {
    // Request another data :)
});

All request will be executed one time, but notify every listeners when they will be done.

Options

  • maxSimultaneousRequest: Determine the number of conccurency request