0.1.2 • Published 4 years ago

react-state-tree v0.1.2

Weekly downloads
2
License
MIT
Repository
github
Last release
4 years ago

react-state-tree

react-state-tree is a drop-in replacement for useState that persists your state into a redux-like state tree stored in a top-level provider component.

If you use useStateTree in an app with no StateTreeProvider, it will fall back to useState. So it is safe to use in custom hooks and components.

Tutorial

To start using react-state-tree, wrap your app with a StateTreeProvider:

import { StateTreeProvider } from "react-state-tree";

function App() {
  return (
    <StateTreeProvider> {/* added StateTreeProvider */}
      <Counter />
    </StateTreeProvider>
  );
}

Now, use useStateTree instead of useState every time you want to persist state to the tree:

import { useStateTree } from "react-state-tree";

function Counter({ id }) {
  const [count, setCount] = useStateTree(0); // replaced useState with useStateTree

  const handleClick = () => setCount(count + 1);
  return <button onClick={handleClick}>{count}</button>;
}

Your app state is now stored within the StateTreeProvider; try adding an onUpdate callback to your StateTreeProvider:

function App() {
  return (
    <StateTreeProvider onUpdate={(state) => console.log(state)}> {/* added onUpdate prop */}
      <Counter />
    </StateTreeProvider>
  );
}

You can now look in the console to see your state:

// Sample log output
{ $0: 1 }
{ $0: 2 }
{ $0: 3 }

It's kind of ugly, though. Let's customize that key by adding a second argument to the useStateTree call:

function Counter({ id }) {
  const [count, setCount] = useStateTree(0, "count"); // added "count" as the second argument

  const handleClick = () => setCount(count + 1);
  return <button onClick={handleClick}>{count}</button>;
}

Now if we look at the log output:

{ count: 1 }
{ count: 2 }
{ count: 3 }

A lot nicer. But what if we want multiple counters in our app?

function App() {
  return (
    <StateTreeProvider onUpdate={(state) => console.log(state)}>
      <Counter />
      <Counter /> {/* added another Counter */}
    </StateTreeProvider>
  );
}

With our custom key, they'll both try to write to count, which won't work...

If we remove the custom key, it'll work, but it'll be ugly again:

{ $0: 1, $1: 0 }
{ $0: 1, $1: 1 }
{ $0: 1, $1: 2 }

How can we keep our nice custom count key but use a different part of the state tree for each counter?

One way is to wrap each counter with the StateKey component:

import { StateTreeProvider, StateKey } from "react-state-tree";

function App() {
  return (
    <StateTreeProvider onUpdate={(state) => console.log(state)}>
      <StateKey value="first"> {/* added StateKey around Counter */}
        <Counter />
      </StateKey>
      <StateKey value="second"> {/* added StateKey around Counter */}
        <Counter />
      </StateKey>
    </StateTreeProvider>
  );
}

And now we can put our custom key back:

function Counter({ id }) {
  const [count, setCount] = useStateTree(0, "count"); // put count back

  const handleClick = () => setCount(count + 1);
  return <button onClick={handleClick}>{count}</button>;
}

And now our state looks like this:

{ first: { count: 1 }, second: {count: 0 } }
{ first: { count: 1 }, second: {count: 1 } }
{ first: { count: 1 }, second: {count: 2 } }

Neat! But what if we want them in an array instead of under the first and second keys?

To do that, we use StateKeyList instead of StateKey:

import { StateTreeProvider, StateKey } from "react-state-tree";

function App() {
  return (
    <StateTreeProvider onUpdate={(state) => console.log(state)}>
      <StateKeyList value="counters"> {/* removed StateKeys and added StateKeyList instead */}
        <Counter />
        <Counter />
      </StateKeyList>
    </StateTreeProvider>
  );
}

Now our state looks like this:

{
  counters: [
    { count: 1 },
    { count: 0 }
  ]
}
{
  counters: [
    { count: 1 },
    { count: 1 }
  ]
}
{
  counters: [
    { count: 1 },
    { count: 2 }
  ]
}

As you can see, the shape of your state follows the shape of your React tree in your app.

Now that we have all your app's state in one tree, let's try persisting it to localStorage:

import { StateTreeProvider, StateKey } from "react-state-tree";

function App() {
  return (
    <StateTreeProvider
      // Added initialState prop and changed onUpdate prop
      initialState={localStorage.state ? JSON.parse(localStorage.state) : {}}
      onUpdate={(state) => localStorage.state = JSON.stringify(state)}
    >
      {/* ... omitted for brevity ... */}
    </StateTreeProvider>
  );
}

Now click your counters a few times, then refresh the page. The state persists across page refreshes!

API Documentation

The react-state-tree module has several named exports; each is documented here.

StateTreeProvider

StateTreeProvider is a React component that should be wrapped around your app- or at least, the part of your app where you want a state tree.

import { StateTreeProvider } from "react-state-tree";

function App() {
  return (
    <StateTreeProvider>
      {/* rest of your app goes here... */}
    </StateTreeProvider>
  );
}

StateTreeProvider accepts two optional props (in addition to children):

  • initialValue - The initial value for the state tree. Defaults to {}.
  • onUpdate - A function to call whenever the state tree changes. It will be called with a single argument: the state tree.

useStateTree

useStateTree is a React hook that behaves the same as useState, but the state is stored on the state tree instead of in component state.

Its signature is the same as useState:

import { useStateTree } from "react-state-tree";

const [count, setCount] = useStateTree(0);

But it also accepts an optional second argument, the key to persist the state under:

import { useStateTree } from "react-state-tree";

const [count, setCount] = useStateTree(0, "count");

If you don't pass the second argument, it will us an autogenerated key by default (unless you use nextStateTreeKey; see the docs for that function below for more details).

StateKey

This component wraps a portion of your app and changes where all useStateTree calls in child components link to on the state tree.

When you don't have a StateKey in your app, all useStateTree calls will link to the top object in the state tree:

import {useStateKey, StateTreeProvider} from "react-state-tree";

// Given this component:
function Counter() {
  const [count, setCount] = useStateTree(0, "count");

  return (
    <button
      onClick={() => setCount(count + 1)}
    >
      {count}
    </button>
  );
}

// And this app:
function App() {
  return (
    <StateTreeProvider>
      <Counter />
    </StateTreeProvider>
  )
}

// The state tree will look like this:
// { count: 0 }

But if you wrap a portion of your app with a StateKey, then all of the state for child components of that StateKey are linked to a sub-tree of your state tree:

import { useStateKey, StateTreeProvider, StateKey } from "react-state-tree";

// If you wrap the same Counter from above with a StateKey:
function App() {
  return (
    <StateTreeProvider>
      <StateKey value="counter">
        <Counter />
      </StateKey>
    </StateTreeProvider>
  );
}

// Then all of that Counter's state is moved underneath the "counter" key in your state tree:
// { counter: { count: 0 } }

The value prop passed to the StateKey determines where child component state gets linked. In the above example, the value prop was "counter", so the state was moved to { counter: ... } in the state tree.

StateKeys compose with parent StateKeys, allowing you to create large, deep state trees.

StateListKey

StateListKey works the same as StateKey, but it creates an Array instead of an object.

Suppose we had an app with two counters in it:

import {useStateKey, StateTreeProvider} from "react-state-tree";

// Given this component:
function Counter({ stateKey }) {
  const [count, setCount] = useStateTree(0, stateKey);

  return (
    <button
      onClick={() => setCount(count + 1)}
    >
      {count}
    </button>
  );
}

// And this app:
function App() {
  return (
    <StateTreeProvider>
      <Counter stateKey="first" />
      <Counter stateKey="second" />
    </StateTreeProvider>
  )
}

// The state tree will look like this:
// { first: 0, second: 0 }

If we wrap the counters in a StateListKey, then they will be linked to an array in the state tree instead:

import {useStateKey, StateTreeProvider, StateListKey} from "react-state-tree";

function App() {
  return (
    <StateTreeProvider>
      <StateListKey value="counters">
        <Counter stateKey="first" />
        <Counter stateKey="second" />
      </StateListKey>
    </StateTreeProvider>
  )
}

// Each Counter's state is moved into an object that lives in an array at the "counters" key in your state tree:
// {
//   counters: [
//     { first: 0 },
//     { second: 0 }
//   ]
// }

// Since they're in separate objects now, it probably makes sense to remove the `stateKey` prop from `Counter` and always put the state at `count` instead:
function Counter() {
  const [count, setCount] = useStateTree(0, "count");

  return (
    <button
      onClick={() => setCount(count + 1)}
    >
      {count}
    </button>
  );
}

// With that change in place, here's how the state tree will look:
// {
//   counters: [
//     { count: 0 },
//     { count: 0 }
//   ]
// }

The value prop passed to the StateListKey determines where the array for child component states will get linked. In the above example, the value prop was "counters", so the state was linked to { counters: [...] } in the state tree.

nextStateTreeKey

nextStateTreeKey is an escape-hatch that lets you specify the state key for the next useStateTree call.

Suppose you had a custom hook that created some state:

function usePosition(leftKey, rightKey) {
  const [position, setPosition] = useStateTree({ x: 0, y: 0 });

  // ... and a bunch of custom logic that
  // calls `setPosition` based on user input,
  // using `leftKey` and `rightKey`.

  return position;
}

And you were using that hook multiple times in a component:

function MyComponent() {
  const player1Pos = usePosition("ArrowLeft", "ArrowRight");
  const player2Pos = usePosition("a", "d");

  // ... rest of the component omitted
}

If you want to specify a state key for each position, you can't do so unless you modify usePosition to accept a state key:

function usePosition(leftKey, rightKey, stateKey) {
  const [position, setPosition] = useStateTree({ x: 0, y: 0 }, stateKey);

  // ...
}

function MyComponent() {
  const player1Pos = usePosition("ArrowLeft", "ArrowRight", "player1Position");
  const player2Pos = usePosition("a", "d", "player2Position");

  // ...
}

This works fine, but if you have a lot of nested hooks, it can be tedious to thread the custom key all the way down to where the useStateTree call is.

As a workaround, you can use the nextStateTreeKey function to specify the key for the next useStateTree call:

function usePosition(leftKey, rightKey) {
  const [position, setPosition] = useStateTree({ x: 0, y: 0 });

  // ...
}

function MyComponent() {
  nextStateTreeKey("player1Position");
  const player1Pos = usePosition("ArrowLeft", "ArrowRight");

  nextStateTreeKey("player2Position");
  const player2Pos = usePosition("a", "d");

  // ...
}

However, use this sparingly; it creates an implicit contract that the next custom hook will use useStateTree at some point, which can be dangerous. If usePosition was changed later to not use useStateTree, then nextStateTreeKey could affect some other unrelated useStateTree call elsewhere in the app (whichever useStateTree call happens next, as long as it doesn't have a key specified).

Note that nextStateTreeKey has no effect on useStateTree calls that have a state key specified as their second argument; nextStateTreeKey only affects useStateTree calls without a second argument. If you call nextStateTreeKey before a useStateTree call, but that useStateTree call had a second argument specified, then the key passed into nextStateTreeKey will be unused, because the second argument to useStateTree will be used instead.

useEntireStateTree

useEntireStateTree is an escape hatch that gives you the entire state tree, and the ability to change it. It's useful if you want to create a component that shows the state tree (for debugging), or you want to implement a redux-style dispatch pattern on top of react-state-tree.

When useEntireStateTree is called, it returns an object with two keys: stateTree and replaceStateTree.

  • stateTree is the entire state tree.
  • replaceStateTree is a function that can be called with a new state tree object to replace the state tree. Think of it as setState for the whole state tree.

License

MIT