# persist

This helper allows you to persist your store state, perform migrations, and subsequently rehydrate the store state when the store is recreated (e.g. on page refresh, new browser tab, etc).

By default it uses the browser's sessionStorage (opens new window), however, you can configure it to use localStorage (opens new window), or provide a custom storage engine.

# API

Below is the API of the persist helper function.

# Arguments

  • model (Object, required)

    The model that you wish to apply persistence to.

  • config (Object, optional)

    You can provide a second parameter to your persist instances, which represents a configuration for the instance. The configuration object supports the following properties:

    • allow (Array<string>, optional)

      A list of keys, representing the parts of the model that should be persisted. Any part of the model that is not represented in this list will not be persisted.

    • deny (Array<string>, optional)

      A list of keys, representing the parts of the model that should not be persisted. Any part of the model that is not represented in this list will be persisted.

    • mergeStrategy (string, optional, default=mergeDeep)

      The strategy that should be employed when rehydrating the persisted state over your store's initial state. It can be one of the following values:

      • mergeDeep
      • mergeShallow
      • overwrite

      Please see the docs below for a full insight and understanding of the various options and their respective implications.

    • migrations (Object, optional)

      This config is used to transform persisted store state from one representation to another. This object is keyed by version numbers, with migration functions attached to each version. A migrationVersion is also required for this object, to specify which version to target.

      persist(
        {...},
        {
          migrations: {
            migrationVersion: 2,
            1: (state) => { ... },
            2: (state) => { ... },
          },
        }
      );
      
    • transformers (Array<Transformer>, optional)

      Transformers are used to apply operations to your data prior to it being persisted or hydrated.

      One use case for a transformer is to handle data that can't be parsed to a JSON string. For example a Map or Set. To handle these data types you could utilise a transformer that converts the Map/Set to/from an Array or Object.

      Transformers are applied left to right during data persistence, and are applied right to left during data rehydration.

      redux-persist (opens new window) already has a robust set of transformer packages (opens new window) that have been built for it. These can be used here.

Please be aware that Store instances contain additional APIs relating to persistence.

For example the store.persist.flush() API will immediately execute any queued persist operations. This can be useful in the context of actions that cause your application to unmount, like the user navigating away from your application.

See the Store docs for more information.

# Tutorial

This section will provide an in-depth walkthrough to persisting and rehydrating your store's state.

# Configuring your store to persist

When choosing to persist your state you firstly need to decide on the scope of your persistence - i.e. how much of your state do you wish to be persisted. You can persist the whole state, a partial slice of your state, or multiple slices of your state.

In the example below we will persist our entire state by wrapping our root model with the persist helper.

import { persist } from 'easy-peasy';

const store = createStore(
  persist({
    count: 1,
    inc: action((state) => {
      state.count += 1;
    }),
  }),
);

You can alternatively target specific parts of your state by wrapping the desired nested models.

const store = createStore(
  products: productsModel,
  basket: persist(basketModel),
  session: persist(sessionModel)
);

Or you can utilise the configuration to explicitly select which keys of a model will be persisted.

const store = createStore(
  persist(
    {
      products: productsModel,
      basket: basketModel,
      session: sessionModel,
    },
    {
      allow: ['basket', 'session'],
    },
  ),
);

Every time a state change occurs the persistence process will be queued to save the state to the storage (sessionStorage (opens new window) by default).

Note: the persisting process is an asynchronous one. Internally we manage a queue and make sure that multiple persistence operations are not fired off concurrently.

# Rehydrating your store

Every time your store is created, we will check for persisted data. If any is found we will use it to rehydrate your store accordingly. This process is asynchronous. Therefore it is best practice to ensure that the rehydration has completed prior to rendering the components within your application that will access the rehydrated state.

To aid with this we expose a useStoreRehydrated hook. This hook will return true when the rehydration process has completed, otherwise it will return false.

Using this hook you could for example show a loading indicator in place of the components that will depend on the rehydrated state.

This allows for a partial application render, providing the user with some perceived performance, as you could render the skeleton of the application in the interim.

import { useStoreRehydrated } from 'easy-peasy';

const store = createStore(persist(model));

function App() {
  const isRehydrated = useStoreRehydrated();
  return (
    <div>
      <Header />
      {isRehydrated ? <Main /> : <div>Loading...</div>}
      <Footer />
    </div>
  );
}

ReactDOM.render(
  <StoreProvider store={store}>
    <App />
  </StoreProvider>,
  document.getElementById('app'),
);

In the example above, the <Main /> content will not render until our store has been successfully updated with the rehydration state.

Alternatively you could create a simple component to wrap your entire application, ensuring that state rehydration has completed prior to rendering it.

import { useStoreRehydrated } from 'easy-peasy';

function WaitForStateRehydration({ children }) {
  const isRehydrated = useStoreRehydrated();
  return isRehydrated ? children : null;
}

ReactDOM.render(
  <StoreProvider store={store}>
    <WaitForStateRehydration>
      <App />
    </WaitForStateRehydration>
  </StoreProvider>,
  document.getElementById('app'),
);

Ultimately, as shown, the useStoreRehydrated hook provides a lot of flexibility.

# Managing model updates

It's entirely reasonable that your store model will evolve over time. However, there is an inherent risk with this when utilizing the persist API.

If your application has previously been deployed to production then users may have persisted state based on a previous version of your store model. The user's persisted state may not align with that of your new store model, which could result in errors when a component tries to consume/update the store.

By default the persist API utilizes the mergeDeep strategy (you can read more about merge strategies further below). The mergeDeep strategy attempts to perform an optimistic merge of the persisted state against the store model. Where it finds that the persisted state is missing keys that are present in the store model, it will ensure to use the respective state from the store model. It will also verify the types of data at each key. If there is a misalignment (ignoring null or undefined) then it will opt for using the data from the store model instead as this generally indicates that the respective state has been refactored.

Whilst the mergeDeep strategy is fairly robust and should be able to cater for a wide variety of model updates, it can't provide a 100% guarantee that a valid state structure will be resolved.

Therefore, when dealing with production applications, we recommend that you consider removing this risk by using one of the two options described below:

# Migrations

Similar to redux-persist (opens new window), the persist API provides a mechanism for migrating persisted state across store updates via the migrations configuration object.

Example

Imagine a store model that has a property called session, and a recent requirements change necessitates that session be renamed to userSession for specificity reasons. Without a migration, if session was previously deployed to users and persisted, when the userSession change is released their application will break due to the mismatch between session and userSession as retrieved from local storage.

In order to mitigate we can add a state migration:

persist(
  {
    userSession: true,
  },
  {
    migrations: {
      migrationVersion: 1, // 👈 set the latest migration version

      1: (state) => {
        state.userSession = state.session; // 👈 update new prop with old value from local storage
        delete state.session; // and then delete, as it is no longer used
      },
    },
  },
);

If this property changes in the future, we can add another migration:

persist(
  {
    domainSession: true, // 👈 model has changed
  },
  {
    migrations: {
      migrationVersion: 2, // 👈 update to the latest version

      1: (state) => {
        state.userSession = state.session;
        delete state.session;
      },

      2: (state) => {
        state.domainSession = state.userSession;
        delete state.userSession;
      },
    },
  },
);

Well-written migrations should obviate the need for merge strategies and render them no-ops. Note, however, that merge strategies are still applied and unexpected behavior may occur during rehydration as a result of non-exhaustive migrations.

# Forced updates via version

If migrations are insufficient (which can often be the case after a major state refactor has taken place), the persist API also provides a means to "force update" the store via version. You can do so by utilizing the version configuration property that is available on the store config.

const store = createStore(
  persist({
    products: productsModel,
    basket: basketModel,
    session: sessionModel,
  }),
  {
    version: 1, 👈
  },
);

This version number is a convenient mechanism by which to mark the version of your store model.

Easy Peasy will be able to compare the version number for the user's persisted state against that of the current store model. If the versions do not align the persisted state will be ignored as it is for a previous version of the store model.

Simply update the version any time you perform a significant update to your store model.

const store = createStore(
   persist({
     products: productsModel,
     basket: basketModel,
     session: sessionModel,
   }),
   {
-     version: 1,
+     version: 2,
   },
);

Whilst this can have a negative effect on user experience, in that their persisted state will be lost, overall it provides stronger guarantees and stability for your users.

Please note that this is entirely optional. If you feel confident that your model changes are simple enough to be resolved by the mergeDeep strategy then there is no need to increment this version number.

# Advanced Tutorial

Below we will cover some of the more advanced aspects of the persist api.

# Merge Strategies

When configuring persistence against your model you can choose from 3 different merge strategies. Each of them have their own unique merits. We invite you to read the docs for each below so that you can choose the strategy that will work best for your needs.

# mergeDeep

This is the default strategy.

The mergeDeep strategy attempts to perform an optimistic merge of the persisted state against the store model.

The data from the persistence will be merged deeply, recursing through all objects and then performing a merge for each item within the object.

It will not merge arrays and other structures such as Map/Set. If it finds any of these structures it will use the value defined within the persisted state, else the value from the store model.

It will also verify the types of data at each key. If there is a misalignment (ignoring null or undefined) then it will opt for using the data from the store model instead as this generally indicates that the respective state has been refactored.

Where it finds that the persisted state is missing keys that are present in the store model, it will ensure to use the respective state from the store model.

We can demonstrate the above behaviour via the following example.

Given a store with the following definition;

import { persist, createStore } from 'easy-peasy';

const store = createStore(
  persist({
    animal: 'dolphin',
    address: {
      city: 'london',
      postCode: 'e3 1pq',
    },
    fruits: ['apple'],
    id: 1,
    name: null,
    counter: 20,
  }),
);

And the following persisted state;

{
  "address": {
    "city": "cape town"
  },
  "flagged": true,
  "fruits": ["banana"],
  "id": "one",
  "name": "Wonder Woman",
  "counter": null
}

The resulting state will be;

{
   "animal": "dolphin",

   "address": {
-    "city": "london",
+    "city": "cape town",

     "postCode": "e3 1pq"
   }

+  "flagged": true,

-  "fruits": ["apple"],
+  "fruits": ["banana"],

   "id": 1,

-  "name": null,
+  "name": "Wonder Woman",

-  "counter": 20
+  "counter": null
}

We can break down the reasoning behind each state item like so;

  • animal - The original value from the store model was maintained as their was no value within the persisted state
  • address.city - The persisted state contained a different value for this property, and hence this value was used.
  • address.postCode - The original value from the store model was maintained as their was no value within the persisted state
  • flagged - The store model didn't contain this property, however, as our persisted state did we copied it across.
  • fruits - The persisted state contained a different value for this property, and hence this value was used.
  • id - The persisted state contained a different data type for this property, so we assumed the model may have changed and therefore used the store model value.
  • name - The persisted state contained a different value for this property, and hence this value was used. Remember null and undefined don't break the type comparison. We consider this a nullable value.
  • counter - The persisted state contained a different value for this property, and hence this value was used. Remember null and undefined don't break the type comparison. We consider this a nullable value.

# mergeShallow

The mergeShallow strategy will compare and merge the properties at the root of the model it was bound against.

It will not merge arrays and other structures such as Map/Set. If it finds any of these structures it will use the value defined within the persisted state, else the value from the store model.

It will also verify the types of data at each key. If there is a misalignment (ignoring null or undefined) then it will opt for using the data from the store model instead as this generally indicates that the respective state has been refactored.

Where it finds that the persisted state is missing keys that are present in the store model, it will ensure to use the respective state from the store model.

We can demonstrate the above behaviour via the following example.

i.e.

Given a store with the following definition;

import { persist, createStore } from 'easy-peasy';

const store = createStore(
  persist(
    {
      animal: 'dolphin',
      address: {
        city: 'london',
        postCode: 'e3 1pq',
      },
      fruits: ['apple'],
      id: 1,
      name: null,
      counter: 20,
    },
    {
      mergeStrategy: 'mergeShallow',
    },
  ),
);

And the following persisted state;

{
  "address": {
    "city": "cape town"
  },
  "flagged": true,
  "fruits": ["banana"],
  "id": "one",
  "name": "Wonder Woman",
  "counter": null
}

The resulting state will be;

{
   "animal": "dolphin",

-  "address": {
-    "city": "london",
-    "postCode": "e3 1pq"
-  },
+  "address": {
+    "city": "cape town"
+  },

+  "flagged": true,

-  "fruits": ["apple"],
+  "fruits": ["banana"],

   "id": 1,

-  "name": null,
+  "name": "Wonder Woman",

-  "counter": 20
+  "counter": null
}

We can break down the reasoning behind each state item like so;

  • animal - The original value from the store model was maintained as their was no value within the persisted state
  • address - As we are doing a mergeShallow the address from the persisted state is used, replacing the address from the store model completely. In the process the address.postCode property from our store model is lost.
  • flagged - The store model didn't contain this property, however, as our persisted state did we copied it across.
  • fruits - The persisted state contained a different value for this property, and hence this value was used.
  • id - The persisted state contained a different data type for this property, so we assumed the model may have changed and therefore used the store model value.
  • name - The persisted state contained a different value for this property, and hence this value was used. Remember null and undefined don't break the type comparison. We consider this a nullable value.
  • counter - The persisted state contained a different value for this property, and hence this value was used. Remember null and undefined don't break the type comparison. We consider this a nullable value.

The behaviour of mergeShallow may be okay for your use-case, however, if you update nested models then it is entirely possible that persisted state may not match the required structure based on the evolved store model. This could cause errors within your application if your components assumed the existing of a particular state structure.

We therefore suggest using this strategy very carefully, and encourage you to use the mergeDeep strategy instead.

The mergeShallow strategy is perhaps more useful if you would like to define your persistence on the "leaf" models of your store, like so;

import { persist, createStore } from 'easy-peasy';

const model = {
  address: persist(
    {
      city: 'london',
      postCode: 'e3 1pq',
    },
    { mergeStrategy: 'mergeShallow' },
  ),
  todos: persist(
    {
      items: [],
    },
    { mergeStrategy: 'mergeShallow' },
  ),
};

# overwrite

Utilizing this strategy will cause the state from your store model to be completely overwritten by the persisted state.

i.e.

Given a store with the following definition;

import { persist, createStore } from 'easy-peasy';

const store = createStore(
  persist(
    {
      fruit: 'pear',
    },
    { mergeStrategy: 'overwrite' },
  ),
);

And the following persisted state;

{
  "city": "cape town"
}

The resulting state will be:

{
  "city": "cape town"
}

This is perhaps the most risky strategy as we intentionally replace our stores initial state with that of the persisted state. If the store model has diverged you could open yourself up to errors/bugs. Please take extra care and consideration when using this strategy.

# Ensuring persistence completes prior to application unmount

Persistence operations are asynchronous. Therefore if you perform a state update prior to unmounting your application it is possible that the state change may not be persisted before your application unmounts.

Your store instances contain an API which allows you to complete the queued persist operations, specifically the store.persist.flush() API.

When this API is executed any queued persist operations will be immediately executed, and a Promise will be returned. The resolution of the returned Promise indicates that the persist has completed, after which you can safely continue to unmount the application.

Below are some example utility functions that make use of this API.

import store from './store';

export const refreshPage = async () => {
  // Firstly ensure that any outstanding persist operations are complete.
  // Note that this is an asynchronous operation so we will await on it.
  await store.persist.flush();

  // we can now safely reload the page
  window.document.reload();
};

export const redirectTo = async (href) => {
  // Firstly ensure that any outstanding persist operations are complete.
  // Note that this is an asynchronous operation so we will await on it.
  await store.persist.flush();

  // We can now safely redirect the browser
  window.location.href = href;
};

# Deleting persisted data

Should you wish to remove all persisted data you can utilise the store instance's API to do so.

const store = createStore(model);

store.persist.clear().then(() => {
  console.log('Persisted state has been removed');
});

Note that a Promise was returned, which when resolved indicates that the data removal has completed.

# Rehydrating dynamic models

The persist API will work with dynamic models, i.e. models that were added to the store via the store.addModel API after the store was created.

Every time a dynamic model is added to your store Easy Peasy will attempt to rehydrate any persisted state for that model.

store.addModel('products', productsModel);

To ensure that the rehydration has completed you can use the resolveRehydration helper that is returned by the store.addModel API.

const { resolveRehydration } = store.addModel('products', productsModel);
//            👆
// Deconstruct the returned object to get a handle on resolveRehydration

// 👇 execute the helper and wait on the returned promise
resolveRehydration().then(() => {
  console.log('Rehydration is complete');
});

# Persisting multiple stores

If you utilise multiple stores, each with their own persistence configuration, you will need to ensure that each store is configured to have a unique name. The store name for each instance of your stores is used within the persistence cache keys created by Easy Peasy, therefore failure to do this can result in the stores overwriting each others persisted data.

const storeOne = createStore(persist(model), {
  name: 'StoreOne', // 👈
});

const storeTwo = createStore(persist(model), {
  name: 'StoreTwo', // 👈
});

# Custom storage engines

You can create a custom storage engine by defining an object that satisfies the following interface:

  • getItem(key) => any | Promise<any> | void

    This function will receive the cache key and should return the associated state from the persistence if it exists. It can alternatively return a Promise that resolves the state, or undefined if no persisted state was found.

  • setItem(key, data) => void | Promise<void>

    This function will receive the cache key as well as the state to persist. It should then store the respective data. It can optionally return a Promise to indicate when the state has been successfully persisted.

  • removeItem(key) => void | Promise<void>

    This function will receive the cache key and should remove any persisted state that is associated with it. It can optionally return a Promise to indicate when the persisted state has been successfully removed.

Once defined you can provide your custom storage engine to the configuration for your persist instance.

import myCustomStorageEngine from './my-custom-storage-engine';

const storeModel = persist(model, {
  storage: myCustomStorageEngine,
});

# Custom data transformers

Transforms allow you to customize the state object that gets persisted and rehydrated. They allow you to for instance convert store data from a complex object, such as a Map, into a structure that is JSON serialisable (and back again).

Easy Peasy outputs a createTransformer function, which has been directly copied from redux-persist (opens new window) in order to maximum compatiblity with it's ecosystem.

We recommend you read the Redux Persist (opens new window) documentation on these for a full understanding.