owl/doc/reference/hooks.md

# 🦉 Hooks 🦉

## Content

- [Overview](#overview)
- [Example: Mouse Position](#example-mouse-position)
- [Example: Autofocus](#example-autofocus)
- [Reference](#reference)
  - [One Rule](#one-rule)
  - [`useState`](#usestate)
  - [`onMounted`](#onmounted)
  - [`onWillUnmount`](#onwillunmount)
  - [`onWillPatch`](#onwillpatch)
  - [`onPatched`](#onpatched)
  - [`onWillStart`](#onwillstart)
  - [`onWillUpdateProps`](#onwillupdateprops)
  - [`useContext`](#usecontext)
  - [`useRef`](#useref)
  - [`useSubEnv`](#usesubenv)
  - [`useExternalListener`](#useexternallistener)
  - [`useStore`](#usestore)
  - [`useDispatch`](#usedispatch)
  - [`useGetters`](#usegetters)
  - [Making customized hooks](#making-customized-hooks)

## Overview

Hooks were popularised by React as a way to solve the following issues:

- help reusing stateful logic between components
- help organizing code by feature in complex components
- use state in functional components, without writing a class.

Owl hooks serve the same purpose, except that they work for class components
(note: React hooks do not work on class components, and maybe because of that,
there seems to be the misconception that hooks are in opposition to class. This
is clearly not true, as shown by Owl hooks).

Hooks work beautifully with Owl components: they solve the problems mentioned
above, and in particular, they are the perfect way to make your component
reactive.

## Example: mouse position

Here is the classical example of a non trivial hook to track the mouse position.

```js
const { useState, onMounted, onWillUnmount } = owl.hooks;

// We define here a custom behaviour: this hook tracks the state of the mouse
// position
function useMouse() {
  const position = useState({ x: 0, y: 0 });

  function update(e) {
    position.x = e.clientX;
    position.y = e.clientY;
  }
  onMounted(() => {
    window.addEventListener("mousemove", update);
  });
  onWillUnmount(() => {
    window.removeEventListener("mousemove", update);
  });

  return position;
}

// Main root component
class App extends owl.Component {
  static template = xml`
    <div t-name="App">
      <div>Mouse: <t t-esc="mouse.x"/>, <t t-esc="mouse.y"/></div>
    </div>`;

  // this hooks is bound to the 'mouse' property.
  mouse = useMouse();
}
```

Note that we use the prefix `use` for hooks, just like in React. This is just
a convention.

## Example: autofocus

Hooks can be combined to create the desired effect. For example, the following
hook combines the `useRef` hook with the `onPatched` and `onMounted` functions
to create an easy way to focus an input whenever it appears in the DOM:

```js
function useAutofocus(name) {
  let ref = useRef(name);
  let isInDom = false;
  function updateFocus() {
    if (!isInDom && ref.el) {
      isInDom = true;
      ref.el.focus();
    } else if (isInDom && !ref.el) {
      isInDom = false;
    }
  }
  onPatched(updateFocus);
  onMounted(updateFocus);
}
```

This hook takes the name of a valid `t-ref` directive, which should be present
in the template. It then checks whenever the component is mounted or patched if
the reference is not valid, and in this case, it will focus the node element.
This hook can be used like this:

```js
class SomeComponent extends Component {
  static template = xml`
    <div>
        <input />
        <input t-ref="myinput"/>
    </div>`;

  constructor(...args) {
    super(...args);
    useAutofocus("myinput");
  }
}
```

## Reference

### One rule

There is only one rule: every hook for a component has to be called in the
constructor (or in class fields):

```js
// ok
class SomeComponent extends Component {
  state = useState({ value: 0 });
}

// also ok
class SomeComponent extends Component {
  constructor(...args) {
    super(...args);
    this.state = useState({ value: 0 });
  }
}

// not ok: this is executed after the constructor is called
class SomeComponent extends Component {
  async willStart() {
    this.state = useState({ value: 0 });
  }
}
```

As you can see, the `useState` hook does not need to be given a reference to
the component. This is possible because there is a way to get a reference to the
current component: the `Component.current` static property is the reference to the
component instance that is currently being created.

Hooks need to be called in the constructor to ensure that this reference is
properly set. This is also a good thing for performance reasons (Owl can use
this to optimize its implementation), and for a clean architecture (this makes
it easier for developers to understand what is really happening in a component).

### `useState`

The `useState` hook is certainly the most important hook for Owl components:
this is what allows a component to be reactive, to react to state change.

The `useState` hook has to be given an object or an array, and will return
an observed version of it (using a `Proxy`).

```javascript
const { useState } = owl.hooks;

class Counter extends owl.Component {
  static template = xml`
    <button t-on-click="increment">
        Click Me! [<t t-esc="state.value"/>]
    </button>`;

  state = useState({ value: 0 });

  increment() {
    this.state.value++;
  }
}
```

It is important to remember that `useState` only works with objects or arrays. It
is necessary, since Owl needs to react to a change in state.

### `onMounted`

`onMounted` is not a user hook, but is a building block designed to help make useful
abstractions. `onMounted` registers a callback, which will be called when the component
is mounted (see example on top of this page).

### `onWillUnmount`

`onWillUnmount` is not a user hook, but is a building block designed to help make useful
abstractions. `onWillUnmount` registers a callback, which will be called when the component
is unmounted (see example on top of this page).

### `onWillPatch`

`onWillPatch` is not a user hook, but is a building block designed to help make useful
abstractions. `onWillPatch` registers a callback, which will be called just
before the component patched.

### `onPatched`

`onPatched` is not a user hook, but is a building block designed to help make useful
abstractions. `onPatched` registers a callback, which will be called just
after the component patched.

### `onWillStart`

`onWillStart` is an asynchronous hook. This means that the function registered
in the hook will be run just before the component is first rendered and can return a
promise, to express the fact that it is an asynchronous operation.

Note that if there are more than one `onWillStart` registered callback, then they
will all be run in parallel.

It can be used to load some initial data. For example, the following hook will
automatically load some data from the server, and return an object that will
be ready whenever the component is rendered:

```js
function useLoader() {
  const component = Component.current;
  const record = useState({});
  onWillStart(async () => {
    const recordId = component.props.id;
    Object.assign(record, await fetchSomeRecord(recordId));
  });
  return record;
}
```

Note that this example does not update the record value whenever props are
updated. For that situation, we need to use the `onWillUpdateProps` hook.

### `onWillUpdateProps`

Just like `onWillStart`, `onWillUpdateProps` is an asynchronous hook. It is
designed to be run whenever the component props are updated. This could be
useful to perform some asynchronous task such as fetching updated data.

```js
function useLoader() {
  const component = Component.current;
  const record = useState({});

  async function updateRecord(id) {
    Object.assign(record, await fetchSomeRecord(id));
  }

  onWillStart(() => updateRecord(component.props.id));
  onWillUpdateProps((nextProps) => updateRecord(nextProps.id));

  return record;
}
```

Note that if there are more than one `onWillUpdateProps` registered callback,
then they will all be run in parallel.

### `useContext`

See [`useContext`](context.md#usecontext) for reference documentation.

### `useRef`

The `useRef` hook is useful when we need a way to interact with some inside part
of a component, rendered by Owl. It can work either on a DOM node, or on a component,
tagged by the `t-ref` directive:

```xml
<div>
    <div t-ref="someDiv"/>
    <SubComponent t-ref="someComponent"/>
</div>
```

In this example, the component will be able to access the `div` and the component
`SubComponent` using the `useRef` hook:

```js
class Parent extends Component {
  subRef = useRef("someComponent");
  divRef = useRef("someDiv");

  someMethod() {
    // here, if component is mounted, refs are active:
    // - this.divRef.el is the div HTMLElement
    // - this.subRef.comp is the instance of the sub component
    // - this.subRef.el is the root HTML node of the sub component (i.e. this.subRef.comp.el)
  }
}
```

As shown by the example above, html elements are accessed by using the `el`
key, and components references are accessed with `comp`.

Notes:

- if used on a component, the reference will be set in the `refs`
  variable between `willPatch` and `patched`,
- on a component, accessing `ref.el` will get the root node of the component.

The `t-ref` directive also accepts dynamic values with string interpolation
(like the [`t-attf-`](qweb_templating_language.md#dynamic-attributes) and
`t-component` directives). For example,

```xml
<div t-ref="component_{{someCondition ? '1' : '2'}}"/>
```

Here, the references need to be set like this:

```js
this.ref1 = useRef("component_1");
this.ref2 = useRef("component_2");
```

References are only guaranteed to be active while the parent component is mounted.
If this is not the case, accessing `el` or `comp` on it will return `null`.

### `useSubEnv`

The environment is sometimes useful to share some common information between
all components. But sometimes, we want to _scope_ that knowledge to a subtree.

For example, if we have a form view component, maybe we would like to make some
`model` object available to all sub components, but not to the whole application.
This is where the `useSubEnv` hook may be useful: it lets a component add some
information to the environment in a way that only the component and its children
can access it:

```js
class FormComponent extends Component {
  constructor(...args) {
    super(...args);
    const model = makeModel();
    useSubEnv({ model });
  }
}
```

The `useSubEnv` takes one argument: an object which contains some key/value that
will be added to the parent environment. Note that it will extend, not replace
the parent environment. And of course, the parent environment will not be
affected.

### `useExternalListener`

The `useExternalListener` hook helps solve a very common problem: adding and removing
a listener on some target whenever a component is mounted/unmounted. For example,
a dropdown menu (or its parent) may need to listen to a `click` event on `window`
to be closed:

```js
useExternalListener(window, "click", this.closeMenu);
```

### `useStore`

The `useStore` hook is the entry point for a component to connect to the store.
See the [store documentation](store.md) for more information.

### `useDispatch`

The `useDispatch` hook is the way for components to get a reference to the store
`dispatch` function. See the [store documentation](store.md) for more information.

### `useGetters`

The `useGetters` hook is the way for components to get a reference to the store
getters. See the [store documentation](store.md) for more information.

### Making customized hooks

Hooks are a wonderful way to organize the code of a complex component by feature
instead of by lifecycle methods. They are like mixins, except that they can be
easily composed together.

But, like every good things in life, hooks should be used with moderation. They are
not the solution to every problem.

- they may be overkill: if your component needs to perform some action specific
  to itself (so, the specific code does not need to be shared), there is nothing
  wrong with a simple class method:

  ```js
  // maybe overkill
  class A extends Component {
    constructor(...args) {
      super(...args);
      useMySpecificHook();
    }
  }

  // ok
  class B extends Component {
    constructor(...args) {
      super(...args);
      this.performSpecificTask();
    }
  }
  ```

  Note that the second solution is easier to extend in sub components.

- they may be harder to test: if a customized hook injects some external side
  effect dependency, then it is harder to test without doing some non obvious
  manipulation. For example, assume that we want to give a reference to a
  router in a `useRouter` hook. We could do this:

  ```js
  const router = new Router(...);

  function useRouter() {
      return router;
  }
  ```

  As you can see, this does not _hook_ into the internal of the component. It
  simply returns a global object, which is difficult to mock.

  A better way would be to do something like this: get the reference from the
  environment.

  ```js
  function useRouter() {
    return Component.current.env.router;
  }
  ```

  This means that we give control to the application developer to create the
  router, which is good, so they can set it up, subclass it, ... And then, to
  test our components, we can just add a mock router in the environment.

Note: the code above makes use of the `Component.current` property. This is the
way hooks are able to get a reference to the component currently being created.