Manipulating the DOM with Refs

Getting a ref to the node

To access a DOM node managed by React, first, import the useRef Hook:

import { useRef } from 'react';

Then, use it to declare a ref inside your component:

const myRef = useRef(null);

Finally, pass your ref as the ref attribute to the JSX tag for which you want to get the DOM node:

<div ref={myRef}>

The useRef Hook returns an object with a single property called current. Initially, myRef.current will be null. When React creates a DOM node for this <div>, React will put a reference to this node into myRef.current. You can then access this DOM node from your event handlers and use the built-in browser APIs defined on it.

// You can use any browser APIs, for example:
myRef.current.scrollIntoView();

Example: Focusing a text input

In this example, clicking the button will focus the input:

To implement this:

1. Declare inputRef with the useRef Hook.
2. Pass it as <input ref={inputRef}>. This tells React to put this <input>’s DOM node into inputRef.current.
3. In the handleClick function, read the input DOM node from inputRef.current and call focus() on it with inputRef.current.focus().
4. Pass the handleClick event handler to <button> with onClick.

While DOM manipulation is the most common use case for refs, the useRef Hook can be used for storing other things outside React, like timer IDs. Similarly to state, refs remain between renders. Refs are like state variables that don’t trigger re-renders when you set them. Read about refs in Referencing Values with Refs.

Example: Scrolling to an element

You can have more than a single ref in a component. In this example, there is a carousel of three images. Each button centers an image by calling the browser scrollIntoView() method on the corresponding DOM node:

<ul>
  {items.map((item) => {
    // Doesn't work!
    const ref = useRef(null);
    return <li ref={ref} />;
  })}
</ul>

import { useRef, useState } from "react";

export default function CatFriends() {
  const itemsRef = useRef(null);
  const [catList, setCatList] = useState(setupCatList);

  function scrollToCat(cat) {
    const map = getMap();
    const node = map.get(cat);
    node.scrollIntoView({
      behavior: "smooth",
      block: "nearest",
      inline: "center",
    });
  }

  function getMap() {
    if (!itemsRef.current) {
      // Initialize the Map on first usage.
      itemsRef.current = new Map();
    }
    return itemsRef.current;
  }

  return (
    <>
      <nav>
        <button onClick={() => scrollToCat(catList[0])}>Neo</button>
        <button onClick={() => scrollToCat(catList[5])}>Millie</button>
        <button onClick={() => scrollToCat(catList[9])}>Bella</button>
      </nav>
      <div>
        <ul>
          {catList.map((cat) => (
            <li
              key={cat}
              ref={(node) => {
                const map = getMap();
                map.set(cat, node);

                return () => {
                  map.delete(cat);
                };
              }}
            >
              <img src={cat} />
            </li>
          ))}
        </ul>
      </div>
    </>
  );
}

function setupCatList() {
  const catList = [];
  for (let i = 0; i < 10; i++) {
    catList.push("https://loremflickr.com/320/240/cat?lock=" + i);
  }

  return catList;
}

<li
  key={cat.id}
  ref={node => {
    const map = getMap();
    // Add to the Map
    map.set(cat, node);

    return () => {
      // Remove from the Map
      map.delete(cat);
    };
  }}
>

Accessing another component’s DOM nodes

When you put a ref on a built-in component that outputs a browser element like <input />, React will set that ref’s current property to the corresponding DOM node (such as the actual <input /> in the browser).

However, if you try to put a ref on your own component, like <MyInput />, by default you will get null. Here is an example demonstrating it. Notice how clicking the button does not focus the input:

This happens because by default React does not let a component access the DOM nodes of other components. Not even for its own children! This is intentional. Refs are an escape hatch that should be used sparingly. Manually manipulating another component’s DOM nodes makes your code even more fragile.

Instead, components that want to expose their DOM nodes have to opt in to that behavior. A component can specify that it “forwards” its ref to one of its children by passing the ref prop down.

function MyInput({ ref, ...props }) {
  return <input {...props} ref={ref} />;
}

This is how it works:

1. <MyInput ref={inputRef} /> tells React to put the corresponding DOM node into inputRef.current. However, it’s up to the MyInput component to opt into that—by default, it doesn’t.
2. MyInput itself passes the ref prop it received to the <input> inside of it.

Now clicking the button to focus the input works:

In design systems, it is a common pattern for low-level components like buttons, inputs, and so on, to forward their refs to their DOM nodes. On the other hand, high-level components like forms, lists, or page sections usually won’t expose their DOM nodes to avoid accidental dependencies on the DOM structure.

import { useRef, useImperativeHandle } from 'react';

function MyInput({ ref, ...props }) {
  const realInputRef = useRef(null);
  useImperativeHandle(ref, () => ({
    // Only expose focus and nothing else
    focus() {
      realInputRef.current.focus();
    },
  }));
  return <input {...props} ref={realInputRef} />;
}

export default function Form() {
  const inputRef = useRef(null);

  function handleClick() {
    inputRef.current.focus();
  }

  return (
    <>
      <MyInput ref={inputRef} />
      <button onClick={handleClick}>
        Focus the input
      </button>
    </>
  );
}

When React attaches the refs

In React, every update is split in two phases:

• During render, React calls your components to figure out what should be on the screen.
• During commit, React applies changes to the DOM.

In general, you don’t want to access refs during rendering. That goes for refs holding DOM nodes as well. During the first render, the DOM nodes have not yet been created, so ref.current will be null. And during the rendering of updates, the DOM nodes haven’t been updated yet. So it’s too early to read them.

React sets ref.current during the commit. Before updating the DOM, React sets the affected ref.current values to null. After updating the DOM, React immediately sets them to the corresponding DOM nodes.

Usually, you will access refs from event handlers. If you want to do something with a ref, but there is no particular event to do it in, you might need an Effect. We will discuss Effects on the next pages.

import { useState, useRef } from 'react';

export default function TodoList() {
  const listRef = useRef(null);
  const [text, setText] = useState('');
  const [todos, setTodos] = useState(
    initialTodos
  );

  function handleAdd() {
    const newTodo = { id: nextId++, text: text };
    setText('');
    setTodos([ ...todos, newTodo]);
    listRef.current.lastChild.scrollIntoView({
      behavior: 'smooth',
      block: 'nearest'
    });
  }

  return (
    <>
      <button onClick={handleAdd}>
        Add
      </button>
      <input
        value={text}
        onChange={e => setText(e.target.value)}
      />
      <ul ref={listRef}>
        {todos.map(todo => (
          <li key={todo.id}>{todo.text}</li>
        ))}
      </ul>
    </>
  );
}

let nextId = 0;
let initialTodos = [];
for (let i = 0; i < 20; i++) {
  initialTodos.push({
    id: nextId++,
    text: 'Todo #' + (i + 1)
  });
}

setTodos([ ...todos, newTodo]);
listRef.current.lastChild.scrollIntoView();

flushSync(() => {
  setTodos([ ...todos, newTodo]);
});
listRef.current.lastChild.scrollIntoView();

import { useState, useRef } from 'react';
import { flushSync } from 'react-dom';

export default function TodoList() {
  const listRef = useRef(null);
  const [text, setText] = useState('');
  const [todos, setTodos] = useState(
    initialTodos
  );

  function handleAdd() {
    const newTodo = { id: nextId++, text: text };
    flushSync(() => {
      setText('');
      setTodos([ ...todos, newTodo]);      
    });
    listRef.current.lastChild.scrollIntoView({
      behavior: 'smooth',
      block: 'nearest'
    });
  }

  return (
    <>
      <button onClick={handleAdd}>
        Add
      </button>
      <input
        value={text}
        onChange={e => setText(e.target.value)}
      />
      <ul ref={listRef}>
        {todos.map(todo => (
          <li key={todo.id}>{todo.text}</li>
        ))}
      </ul>
    </>
  );
}

let nextId = 0;
let initialTodos = [];
for (let i = 0; i < 20; i++) {
  initialTodos.push({
    id: nextId++,
    text: 'Todo #' + (i + 1)
  });
}

Best practices for DOM manipulation with refs

Refs are an escape hatch. You should only use them when you have to “step outside React”. Common examples of this include managing focus, scroll position, or calling browser APIs that React does not expose.

If you stick to non-destructive actions like focusing and scrolling, you shouldn’t encounter any problems. However, if you try to modify the DOM manually, you can risk conflicting with the changes React is making.

To illustrate this problem, this example includes a welcome message and two buttons. The first button toggles its presence using conditional rendering and state, as you would usually do in React. The second button uses the remove() DOM API to forcefully remove it from the DOM outside of React’s control.

Try pressing “Toggle with setState” a few times. The message should disappear and appear again. Then press “Remove from the DOM”. This will forcefully remove it. Finally, press “Toggle with setState”:

After you’ve manually removed the DOM element, trying to use setState to show it again will lead to a crash. This is because you’ve changed the DOM, and React doesn’t know how to continue managing it correctly.

Avoid changing DOM nodes managed by React. Modifying, adding children to, or removing children from elements that are managed by React can lead to inconsistent visual results or crashes like above.

However, this doesn’t mean that you can’t do it at all. It requires caution. You can safely modify parts of the DOM that React has no reason to update. For example, if some <div> is always empty in the JSX, React won’t have a reason to touch its children list. Therefore, it is safe to manually add or remove elements there.