I'm writing a test for a component that takes a few props such as "isLoading", "clients" and "refreshClients". refreshClients is defined like this:
function refreshClients() {
setIsLoading(true)
getClients().then(response => {
setClients(response)
setIsLoading(false)
})
}
clients and isLoading are states from the parent component that are used as props of the child component. The client component also gets the refreshClients prop, which allows it to update its own props (isLoading and clients), through the function passed in by the parent component.
This is one of the use cases: after taking an action over a client, like deletion, the component will call refreshClients, which should take care of reloading the clients table displayed in the component. While the new listing is being loaded the table isn't displayed (isLoading is true). The component works well, however I'm unsure on how to properly test it using testing-library. I'm basically calling rerender in the tests but I feel there should be a way to replicate this behavior in the tests...
Is there a way to create states to pass as props to the tested component? Or is there another recommended approach to handle cases like this?
In case it makes it easier to visualize the idea, here is a complete simple example of how it would work:
import {createRoot} from 'react-dom'
import React, {useState} from 'react';
export function App() {
const [isLoading, setIsLoading] = useState(false);
const [clients, setClients] = useState(['david', 'sara']);
const refreshClients = () => {
setIsLoading(true)
setTimeout(() => {
setClients(['john', 'mary'])
setIsLoading(false)
}, 1000)
}
return <ClientsTable isLoading={isLoading} clients={clients} refreshClients={refreshClients} />
}
function ClientsTable({ isLoading, clients, refreshClients}) {
const deleteClient = () => {
console.log('TODO: delete client')
refreshClients()
}
return (
<div>
{isLoading && <p>Loading... please wait</p>}
{!isLoading && clients.map(client => (
<div>{client} <button onClick={deleteClient}>delete</button></div>
))}
</div>
);
}
createRoot(
document.getElementById('root')
).render(<App />)
In my Next.js app I can't seem to access window:
Unhandled Rejection (ReferenceError): window is not defined
componentWillMount() {
console.log('window.innerHeight', window.innerHeight);
}
̶A̶n̶o̶t̶h̶e̶r̶ ̶s̶o̶l̶u̶t̶i̶o̶n̶ ̶i̶s̶ ̶b̶y̶ ̶u̶s̶i̶n̶g̶ ̶p̶r̶o̶c̶e̶s̶s̶.̶b̶r̶o̶w̶s̶e̶r ̶ ̶t̶o̶ ̶j̶u̶s̶t̶ ̶e̶x̶e̶c̶u̶t̶e̶ ̶ ̶y̶o̶u̶r̶ ̶c̶o̶m̶m̶a̶n̶d̶ ̶d̶u̶r̶i̶n̶g̶ ̶r̶e̶n̶d̶e̶r̶i̶n̶g̶ ̶o̶n̶ ̶t̶h̶e̶ ̶c̶l̶i̶e̶n̶t̶ ̶s̶i̶d̶e̶ ̶o̶n̶l̶y̶.
But process object has been deprecated in Webpack5 and also NextJS, because it is a NodeJS variable for backend side only.
So we have to use back window object from the browser.
if (typeof window !== "undefined") {
// Client-side-only code
}
Other solution is by using react hook to replace componentDidMount:
useEffect(() => {
// Client-side-only code
})
Move the code from componentWillMount() to componentDidMount():
componentDidMount() {
console.log('window.innerHeight', window.innerHeight);
}
In Next.js, componentDidMount() is executed only on the client where window and other browser specific APIs will be available. From the Next.js wiki:
Next.js is universal, which means it executes code first server-side,
then client-side. The window object is only present client-side, so if
you absolutely need to have access to it in some React component, you
should put that code in componentDidMount. This lifecycle method will
only be executed on the client. You may also want to check if there
isn't some alternative universal library which may suit your needs.
Along the same lines, componentWillMount() will be deprecated in v17 of React, so it effectively will be potentially unsafe to use in the very near future.
If you use React Hooks you can move the code into the Effect Hook:
import * as React from "react";
export const MyComp = () => {
React.useEffect(() => {
// window is accessible here.
console.log("window.innerHeight", window.innerHeight);
}, []);
return (<div></div>)
}
The code inside useEffect is only executed on the client (in the browser), thus it has access to window.
With No SSR
https://nextjs.org/docs/advanced-features/dynamic-import#with-no-ssr
import dynamic from 'next/dynamic'
const DynamicComponentWithNoSSR = dynamic(
() => import('../components/hello3'),
{ ssr: false }
)
function Home() {
return (
<div>
<Header />
<DynamicComponentWithNoSSR />
<p>HOME PAGE is here!</p>
</div>
)
}
export default Home
The error occurs because window is not yet available, while component is still mounting. You can access window object after component is mounted.
You can create a very useful hook for getting dynamic window.innerHeight or window.innerWidth
const useDeviceSize = () => {
const [width, setWidth] = useState(0)
const [height, setHeight] = useState(0)
const handleWindowResize = () => {
setWidth(window.innerWidth);
setHeight(window.innerHeight);
}
useEffect(() => {
// component is mounted and window is available
handleWindowResize();
window.addEventListener('resize', handleWindowResize);
// unsubscribe from the event on component unmount
return () => window.removeEventListener('resize', handleWindowResize);
}, []);
return [width, height]
}
export default useDeviceSize
Use case:
const [width, height] = useDeviceSize();
componentWillMount() lifecycle hook works both on server as well as client side. In your case server would not know about window or document during page serving, the suggestion is to move the code to either
Solution 1:
componentDidMount()
Or, Solution 2
In case it is something that you only want to perform in then you could write something like:
componentWillMount() {
if (typeof window !== 'undefined') {
console.log('window.innerHeight', window.innerHeight);
}
}
In the constructor of your class Component you can add
if (typeof window === 'undefined') {
global.window = {}
}
Example:
import React, { Component } from 'react'
class MyClassName extends Component {
constructor(props){
super(props)
...
if (typeof window === 'undefined') {
global.window = {}
}
}
This will avoid the error (in my case, the error would occur after I would click reload of the page).
global?.window && window.innerHeight
It's important to use the operator ?., otherwise the build command might crash.
Best solution ever
import dynamic from 'next/dynamic';
const Chart = dynamic(()=> import('react-apexcharts'), {
ssr:false,
})
A bit late but you could also consider using Dynamic Imports from next turn off SSR for that component.
You can warp the import for your component inside a dynamic function and then, use the returned value as the actual component.
import dynamic from 'next/dynamic'
const BoardDynamic = dynamic(() => import('../components/Board.tsx'), {
ssr: false,
})
<>
<BoardDynamic />
</>
I have to access the hash from the URL so I come up with this
const hash = global.window && window.location.hash;
Here's an easy-to-use workaround that I did.
const runOnClient = (func: () => any) => {
if (typeof window !== "undefined") {
if (window.document.readyState == "loading") {
window.addEventListener("load", func);
} else {
func();
}
}
};
Usage:
runOnClient(() => {
// access window as you like
})
// or async
runOnClient(async () => {
// remember to catch errors that might be raised in promises, and use the `await` keyword wherever needed
})
This is better than just typeof window !== "undefined", because if you just check that the window is not undefined, it won't work if your page was redirected to, it just works once while loading. But this workaround works even if the page was redirected to, not just once while loading.
I was facing the same problem when i was developing a web application in next.js This fixed my problem, you have to refer to refer the window object in a life cycle method or a react Hook. For example lets say i want to create a store variable with redux and in this store i want to use a windows object i can do it as follows:
let store
useEffect(()=>{
store = createStore(rootReducers, window.__REDUX_DEVTOOLS_EXTENSION__ &&
window.__REDUX_DEVTOOLS_EXTENSION__())
}, [])
....
So basically, when you are working with window's object always use a hook to play around or componentDidMount() life cycle method
I wrapped the general solution (if (typeof window === 'undefined') return;) in a custom hook, that I am very pleased with. It has a similiar interface to reacts useMemo hook which I really like.
import { useEffect, useMemo, useState } from "react";
const InitialState = Symbol("initial");
/**
*
* #param clientFactory Factory function similiar to `useMemo`. However, this function is only ever called on the client and will transform any returned promises into their resolved values.
* #param deps Factory function dependencies, just like in `useMemo`.
* #param serverFactory Factory function that may be called server side. Unlike the `clientFactory` function a resulting `Promise` will not be resolved, and will continue to be returned while the `clientFactory` is pending.
*/
export function useClientSideMemo<T = any, K = T>(
clientFactory: () => T | Promise<T>,
deps: Parameters<typeof useMemo>["1"],
serverFactory?: () => K
) {
const [memoized, setMemoized] = useState<T | typeof InitialState>(
InitialState
);
useEffect(() => {
(async () => {
setMemoized(await clientFactory());
})();
// eslint-disable-next-line react-hooks/exhaustive-deps
}, deps);
return typeof window === "undefined" || memoized === InitialState
? serverFactory?.()
: memoized;
}
Usage Example:
I am using it to dynamically import libaries that are not compatible with SSR in next.js, since its own dynamic import is only compatible with components.
const renderer = useClientSideMemo(
async () =>
(await import("#/components/table/renderers/HighlightTextRenderer"))
.HighlightTextRendererAlias,
[],
() => "text"
);
As you can see I even implemented a fallback factory callback, so you may provide a result when initially rendering on the server aswell. In all other aspects this hook should behave similiar to reacts useMemo hook. Open to feedback.
For such cases, Next.js has Dynamic Import.
A module that includes a library that only works in the browser, it's suggested to use Dynamic Import. Refer
Date: 06/08/2021
Check if the window object exists or not and then follow the code along with it.
function getSelectedAddress() {
if (typeof window === 'undefined') return;
// Some other logic
}
For Next.js version 12.1.0, I find that we can use process.title to determine whether we are in browser or in node side. Hope it helps!
export default function Projects(props) {
console.log({ 'process?.title': process?.title });
return (
<div></div>
);
}
1. From the terminal, I receive { 'process?.title': 'node' }
2. From Chrome devtool, I revice { 'process?.title': 'browser' }
I had this same issue when refreshing the page (caused by an import that didn't work well with SSR).
What fixed it for me was going to pages where this was occurring and forcing the import to be dynamic:
import dynamic from 'next/dynamic';
const SomeComponent = dynamic(()=>{return import('../Components/SomeComponent')}, {ssr: false});
//import SomeComponent from '../Components/SomeComponent'
Commenting out the original import and importing the component dynamically forces the client-side rendering of the component.
The dynamic import is covered in Nextjs's documentation here:
https://nextjs.org/docs/advanced-features/dynamic-import
I got to this solution by watching the youtube video here:
https://www.youtube.com/watch?v=DA0ie1RPP6g
You can define a state var and use the window event handle to handle changes like so.
const [height, setHeight] = useState();
useEffect(() => {
if (!height) setHeight(window.innerHeight - 140);
window.addEventListener("resize", () => {
setHeight(window.innerHeight - 140);
});
}, []);
You can try the below code snippet for use-cases such as - to get current pathname (CurrentUrl Path)
import { useRouter } from "next/router";
const navigator = useRouter()
console.log(navigator.pathname);
For anyone who somehow cannot use hook (for example, function component):
Use setTimeout(() => yourFunctionWithWindow()); will allow it get the window instance. Guess it just need a little more time to load.
I want to leave this approach that I found interesting for future researchers. It's using a custom hook useEventListener that can be used in so many others needs.
Note that you will need to apply a little change in the originally posted one, like I suggest here.
So it will finish like this:
import { useRef, useEffect } from 'react'
export const useEventListener = (eventName, handler, element) => {
const savedHandler = useRef()
useEffect(() => {
savedHandler.current = handler
}, [handler])
useEffect(() => {
element = !element ? window : element
const isSupported = element && element.addEventListener
if (!isSupported) return
const eventListener = (event) => savedHandler.current(event)
element.addEventListener(eventName, eventListener)
return () => {
element.removeEventListener(eventName, eventListener)
}
}, [eventName, element])
}
If it is NextJS app and inside _document.js, use below:
<script dangerouslySetInnerHTML={{
__html: `
var innerHeight = window.innerHeight;
`
}} />
My problem is, I'm not being able to use the 'setInterval' attached to a button and use another one to stop it with 'clearInterval'.
My thought process:
Since clearInterval() would need a target, I stored the setInterval() inside a variable. However, I noticed that setInterval() starts running without being triggered by the eventListener attached to it. I was expecting this behaviour of firing on its own only if I had put setInterval() out in the open without being stored on a variable and on the global scope.
So then I tried to encapsulate it inside a function, only this time the clearInterval() seemed to 'lose' the target.
I had a look on some situtations and couldn't find an explanation that would satisfy my case. Right now it looks like I can't have both ways (a btn to 'fire' the timer and a btn to shut it off).
I don't know if this is a matter of scope,event handling or targeting. Had a look on HTMLStandard but couldn't make much use of it.
Thank you
<script>
let repeater = function () {
setInterval(() => {
console.log('loading');
}, 1000);
};
const startButton = document.querySelector('#firstB');
const stopButton = document.querySelector('#secondB');
startButton.addEventListener('click', repeater);
stopButton.addEventListener('click', () => {
clearInterval(repeater);
console.log('finished');
});
</script>
const startButton = document.querySelector("#firstB");
const stopButton = document.querySelector("#secondB");
let repeater;
startButton.addEventListener("click", function () {
repeater = setInterval(() => {
console.log("loading");
}, 1000);
});
stopButton.addEventListener("click", () => {
clearInterval(repeater);
console.log("finished");
});
This should work.
I'm trying to safely remove a DOM node from a component made whit StencilJS.
I've put the removing code in a public method - It's what I need.
But, depending on which moment this method is called, I have a problem. If it is called too early, it don't have the DOM node reference yet - it's undefined.
The code below shows the component code (using StencilJS) and the HTML page.
Calling alert.dismiss() in page script is problematic. Calling the same method clicking the button works fine.
There is a safe way to do this remove()? Do StencilJS provide some resource, something I should test or I should wait?
import {
Component,
Element,
h,
Method
} from '#stencil/core';
#Component({
tag: 'my-alert',
scoped: true
})
export class Alert {
// Reference to dismiss button
dismissButton: HTMLButtonElement;
/**
* StencilJS lifecycle methods
*/
componentDidLoad() {
// Dismiss button click handler
this.dismissButton.addEventListener('click', () => this.dismiss());
}
// If this method is called from "click" event (handler above), everything is ok.
// If it is called from a script executed on the page, this.dismissButton may be undefined.
#Method()
async dismiss() {
// Remove button from DOM
// ** But this.dismissButton is undefined before `render` **
this.dismissButton.remove();
}
render() {
return (
<div>
<slot/>
<button ref={el => this.dismissButton = el as HTMLButtonElement} >
Dismiss
</button>
</div>
);
}
}
<!DOCTYPE html>
<html lang="pt-br">
<head>
<title>App</title>
</head>
<body>
<my-alert>Can be dismissed.</my-alert>
<script type="module">
import { defineCustomElements } from './node_modules/my-alert/alert.js';
defineCustomElements();
(async () => {
await customElements.whenDefined('my-alert');
let alert = document.querySelector('my-alert');
// ** Throw an error, because `this.dismissButton`
// is undefined at this moment.
await alert.dismiss();
})();
</script>
</body>
</html>
There are multiple ways to delete DOM nodes in Stencil.
The simplest is to just call remove() on the element, like any other element:
document.querySelector('my-alert').remove();
Another would be to have a parent container that manages the my-alert message(s). This is especially useful for things like notifications.
#Component({...})
class MyAlertManager {
#Prop({ mutable: true }) alerts = ['alert 1'];
removeAlert(alert: string) {
const index = this.alerts.indexOf(alert);
this.alerts = [
...this.alerts.slice(0, index),
...this.alerts.slice(index + 1, 0),
];
}
render() {
return (
<Host>
{this.alerts.map(alert => <my-alert text={alert} />)}
</Host>
);
}
}
There are other options and which one to choose will depend on the exact use case.
Update
In your specific case I would just render the dismiss button conditionally:
export class Alert {
#State() shouldRenderDismissButton = true;
#Method()
async dismiss() {
this.shouldRenderDismissButton = false;
}
render() {
return (
<div>
<slot/>
{this.shouldRenderDismissButton && <button onClick={() => this.dismiss()}>
Dismiss
</button>
</div>
);
}
}
Generally I would not recommend manually manipulating the DOM in Stencil components directly since that could lead to problems with the next renders since the virtual DOM is out of sync with the real DOM.
And if you really need to wait for the component to render you can use a Promise:
class Alert {
loadPromiseResolve;
loadPromise = new Promise(resolve => this.loadPromiseResolve = resolve);
#Method()
async dismiss() {
// Wait for load
await this.loadPromise;
// Remove button from DOM
this.dismissButton.remove();
}
componentDidLoad() {
this.loadPromiseResolve();
}
}
I previously asked a question about waiting for the next render which would make this a bit cleaner but I don't think it's easily possible at the moment. I might create a feature request for this in the future.
How can I detect when Mapbox or Leaflet enters or exits fullscreen mode?
I found this answer where someone said this:
Documentation says:
map.on('fullscreenchange', function () {
if (map.isFullscreen()) {
console.log('entered fullscreen');
} else {
console.log('exited fullscreen');
}
});
If doesnt work, use this instead:
map.on('enterFullscreen', function(){
});
map.on('exitFullscreen', function(){
});
I tried that, as well as a few variations of the event type parameter. No dice.
Also, the documentation doesn't mention an event for this.
Note that I am using Mapbox GL JS.
I know this is a late response but to anyone in the future this is how I approached it (for mapbox GL JS (without leaflet).
map.on("resize", () => {
if (document.fullscreenElement) // do something
});
You can give the map wrapper div a name and exclusively also check if the map is what triggered the fullscreen event
map.on("resize", () => {
if (document.fullscreenElement?.attributes.name.value === "mapWrapper") // do something
});
And if you are using React you can use a state to hold this information.
const [isFullScreen, setIsFullScreen] = useState();
...
map.on("resize", () => {
setIsFullScreen(
document.fullscreenElement?.attributes.name.value === "mapWrapper"
);
});
...
if (isFullScreen) //do something
This is actually really simple. You don't need anything from Leaflet or Mapbox. Just use an event listener on the document object.
let fullScreenChange;
if ('onfullscreenchange' in window.document) {
fullScreenChange = 'fullscreenchange';
} else if ('onmozfullscreenchange' in window.document) {
fullScreenChange = 'mozfullscreenchange';
} else if ('onwebkitfullscreenchange' in window.document) {
fullScreenChange = 'webkitfullscreenchange';
} else if ('onmsfullscreenchange' in window.document) {
fullScreenChange = 'MSFullscreenChange';
}
function onFullscreenChange() {
// Your stuff.
}
window.document.addEventListener(fullScreenChange, onFullscreenChange);