react-native-onyx

Awesome persistent storage solution wrapped in a Pub/Sub library.

Features

• Onyx stores and retrieves data from persistent storage
• Data is stored as key/value pairs, where the value can be anything from a single piece of data to a complex object
• Collections of data are usually not stored as a single key (e.g. an array with multiple objects), but as individual keys+ID (e.g. report_1234, report_4567, etc.). Store collections as individual keys when a component will bind directly to one of those keys. For example: reports are stored as individual keys because SidebarLink.js binds to the individual report keys for each link. However, report actions are stored as an array of objects because nothing binds directly to a single report action.
• Onyx allows other code to subscribe to changes in data, and then publishes change events whenever data is changed
• Anything needing to read Onyx data needs to:
  1. Know what key the data is stored in (for web, you can find this by looking in the JS console > Application > local storage)
  2. Subscribe to changes of the data for a particular key or set of keys. React function components use the useOnyx() hook (recommended), both class and function components can use withOnyx() HOC (deprecated, not-recommended) and non-React libs use Onyx.connect().
  3. Get initialized with the current value of that key from persistent storage (Onyx does this by calling setState() or triggering the callback with the values currently on disk as part of the connection process)
• Subscribing to Onyx keys is done using a constant defined in ONYXKEYS. Each Onyx key represents either a collection of items or a specific entry in storage. For example, since all reports are stored as individual keys like report_1234, if code needs to know about all the reports (e.g. display a list of them in the nav menu), then it would subscribe to the key ONYXKEYS.COLLECTION.REPORT.

Getting Started

Installation

Onyx is published to npm

npm install react-native-onyx --save

Initialization

To initialize Onyx we call Onyx.init() with a configuration object.

import Onyx from 'react-native-onyx';

const ONYXKEYS = {
    SESSION: 'session',
};

const config = {
    keys: ONYXKEYS,
};

Onyx.init(config);

Usage in non react-native projects

Onyx can be used in non react-native projects, by leveraging the browser field in package.json Bundlers like Webpack respect that field and import code from the specified path We import Onyx the same way shown above - import Onyx from 'react-native-onyx'

Setting data

To store some data we can use the Onyx.set() method.

API.Authenticate(params)
    .then((response) => {
        Onyx.set(ONYXKEYS.SESSION, {token: response.token});
    });

The data will then be cached and stored via AsyncStorage.

Merging data

We can also use Onyx.merge() to merge new Object or Array data in with existing data.

For Array the default behavior is to replace it fully, effectively making it equivalent to set:

Onyx.merge(ONYXKEYS.EMPLOYEE_LIST, ['Joe']); // -> ['Joe']
Onyx.merge(ONYXKEYS.EMPLOYEE_LIST, ['Jack']); // -> ['Jack']

For Object values the default behavior uses lodash/merge under the hood to do a deep extend of the object.

Onyx.merge(ONYXKEYS.POLICY, {id: 1}); // -> {id: 1}
Onyx.merge(ONYXKEYS.POLICY, {name: 'My Workspace'}); // -> {id: 1, name: 'My Workspace'}

Arrays inside objects will be replaced fully, same as arrays not inside objects:

Onyx.merge(ONYXKEYS.POLICY, {employeeList: ['Joe', 'Jack']}); // -> {employeeList: ['Joe', 'Jack']}
Onyx.merge(ONYXKEYS.POLICY, {employeeList: ['Jack']}); // -> {employeeList: ['Jack']}

Should I use merge() or set() or both?

• Use merge() when creating a new object
• Use merge() to merge partial data into an existing object
• Use merge() when storing simple values (String, Boolean, Number)
• Use set() when you need to delete an Onyx key completely from storage
• Use set() when you need to completely reset an object or array of data

Consecutive calls to Onyx.merge() with the same key are batched in a stack and processed in the order that they were called. This helps avoid race conditions where one merge possibly finishes before another. However, it's important to note that calls to Onyx.set() are not batched together with calls to Onyx.merge(). For this reason, it is usually preferable to use one or the other, but not both. Onyx is a work-in-progress so always test code to make sure assumptions are correct!

Should I store things as an array or an object?

You should avoid arrays as much as possible. They do not work well with merge() because it can't update a single element in an array, it must always set the entire array each time. This forces you to use set() a lot, and as seen above, merge() is more performant and better to use in almost any situation. If you are working with an array of objects, then you should be using an Onyx collection because it's optimized for working with arrays of objects.

Subscribing to data changes

To set up a basic subscription for a given key use the Onyx.connect() method.

let session;
const connectionID = Onyx.connect({
    key: ONYXKEYS.SESSION,
    callback: (val) => session = val || {},
});

To teardown the subscription call Onyx.disconnect() with the connectionID returned from Onyx.connect(). It's recommended to clean up subscriptions anytime you are connecting from within a function to prevent memory leaks.

Onyx.disconnect(connectionID);

We can also access values inside React function components via the useOnyx() hook (recommended) or class and function components via the withOnyx() higher order component (deprecated, not-recommended). When the data changes the component will re-render.

import React from 'react';
import {useOnyx} from 'react-native-onyx';

const App = () => {
    const [session] = useOnyx('session');
    
    return (
        <View>
            {session.token ? <Text>Logged in</Text> : <Text>Logged out</Text>}
        </View>
    );
};

export default App;

The useOnyx() hook won't delay the rendering of the component using it while the key/entity is being fetched and passed to the component. However, you can simulate this behavior by checking if the status of the hook's result metadata is loading. When status is loading it means that the Onyx data is being loaded into cache and thus is not immediately available, while loaded means that the data is already loaded and available to be consumed.

const [reports, reportsResult] = useOnyx(ONYXKEYS.COLLECTION.REPORT);
const [session, sessionResult] = useOnyx(ONYXKEYS.SESSION);

if (reportsResult.status === 'loading' || sessionResult.status === 'loading') {
    return <Placeholder />; // or `null` if you don't want to render anything.
}

// rest of the component's code.

Warning

Deprecated Note

Please note that the withOnyx() Higher Order Component (HOC) is now considered deprecated. Use useOnyx() hook instead.

import React from 'react';
import {withOnyx} from 'react-native-onyx';

const App = ({session}) => (
    <View>
        {session.token ? <Text>Logged in</Text> : <Text>Logged out</Text> }
    </View>
);

export default withOnyx({
    session: {
        key: ONYXKEYS.SESSION,
    },
})(App);

Differently from useOnyx(), withOnyx() will delay the rendering of the wrapped component until all keys/entities have been fetched and passed to the component, this can be convenient for simple cases. This however, can really delay your application if many entities are connected to the same component, you can pass an initialValue to each key to allow Onyx to eagerly render your component with this value.

export default withOnyx({
    session: {
        key: ONYXKEYS.SESSION,
        initialValue: {}
    },
})(App);

Additionally, if your component has many keys/entities when your component will mount but will receive many updates as data is fetched from DB and passed down to it, as every key that gets fetched will trigger a setState on the withOnyx HOC. This might cause re-renders on the initial mounting, preventing the component from mounting/rendering in reasonable time, making your app feel slow and even delaying animations.

You can workaround this by passing an additional object with the shouldDelayUpdates property set to true. Onyx will then put all the updates in a queue until you decide when then should be applied, the component will receive a function markReadyForHydration. A good place to call this function is on the onLayout method, which gets triggered after your component has been rendered.

const App = ({session, markReadyForHydration}) => (
    <View onLayout={() => markReadyForHydration()}>
        {session.token ? <Text>Logged in</Text> : <Text>Logged out</Text> }
    </View>
);

// Second argument to funciton is `shouldDelayUpdates`
export default withOnyx({
    session: {
        key: ONYXKEYS.SESSION,
        initialValue: {}
    },
}, true)(App);

Dependent Onyx Keys and useOnyx()

Some components need to subscribe to multiple Onyx keys at once and sometimes, one key might rely on the data from another key. This is similar to a JOIN in SQL.

Example: To get the policy of a report, the policy key depends on the report key.

const App = ({reportID}) => {
    const [report] = useOnyx(`${ONYXKEYS.COLLECTION.REPORT}${reportID}`);
    const [policy] = useOnyx(`${ONYXKEYS.COLLECTION.POLICY}${report.policyID}`);

    return (
        <View>
            {/* Render with policy data */}
        </View>
    );
};

export default App;

Detailed explanation of how this is handled and rendered with useOnyx():

1. The component mounts with a reportID={1234} prop.
2. The useOnyx hook evaluates the mapping and subscribes to the key reports_1234 using the reportID prop.
3. The useOnyx hook fetches the data for the key reports_1234 from Onyx and sets the state with the initial value (if provided).
4. Since report is not defined yet, report?.policyID defaults to undefined. The useOnyx hook subscribes to the key policies_undefined.
5. The useOnyx hook reads the data and updates the state of the component:
   • report={{reportID: 1234, policyID: 1, ...rest of the object...}}
   • policy={undefined} (since there is no policy with ID undefined)
6. The useOnyx hook again evaluates the key policies_1 after fetching the updated report object which has policyID: 1.
7. The useOnyx hook reads the data and updates the state with:
   • policy={{policyID: 1, ...rest of the object...}}
8. Now, all mappings have values that are defined (not undefined), and the component is rendered with all necessary data.

• It is VERY important to NOT use empty string default values like report.policyID || ''. This results in the key returned to useOnyx as policies_, which subscribes to the ENTIRE POLICY COLLECTION and is most assuredly not what you were intending. You can use a default of 0 (as long as you are reasonably sure that there is never a policyID=0). This allows Onyx to return undefined as the value of the policy key, which is handled by useOnyx appropriately.

Detailed explanation of how this is handled and rendered with withOnyx HOC:

1. The component mounts with a reportID={1234} prop
2. withOnyx evaluates the mapping
3. withOnyx connects to the key reports_1234 because of the prop passed to the component
4. withOnyx connects to the key policies_undefined because report doesn't exist in the props yet, so the policyID defaults to undefined. * (see note below)
5. Onyx reads the data and updates the state of withOnyx with:
   • report={{reportID: 1234, policyID: 1, ... the rest of the object ...}}
   • policy={undefined} (since there is no policy with ID undefined)
6. There is still an undefined key in the mapping, so Onyx reads the data again
7. This time withOnyx connects to the key policies_1 because the report object exists in the component's state and it has a policyID: 1
8. Onyx reads the data and updates the state of withOnyx with:
   • policy={{policyID: 1, ... the rest of the object ...}
9. Now all mappings have values that are defined (not undefined) and the component is rendered with all necessary data

• It is VERY important to NOT use empty string default values like report.policyID || ''. This results in the key returned to withOnyx as policies_ which subscribes to the ENTIRE POLICY COLLECTION and is most assuredly not what you were intending. You can use a default of 0 (as long as you are reasonably sure that there is never a policyID=0). This allows Onyx to return undefined as the value of the policy key, which is handled by withOnyx appropriately.

DO NOT use more than one withOnyx component at a time. It adds overhead and prevents some optimizations like batched rendering from working to its full potential.

It's also beneficial to use a selector with the mapping in case you need to grab a single item in a collection (like a single report action).

Collections

Collections allow keys with similar value types to be subscribed together by subscribing to the collection key. To define one, it must be included in the ONYXKEYS.COLLECTION object and it must be suffixed with an underscore. Member keys should use a unique identifier or index after the collection key prefix (e.g. report_42).

const ONYXKEYS = {
    COLLECTION: {
        REPORT: 'report_',
    },
};

Setting Collection Values

To save a new collection key we can either do:

Onyx.merge(`${ONYXKEYS.COLLECTION.REPORT}${report1.reportID}`, report1);

or we can set many at once with mergeCollection() (see below for guidance on best practices):

Onyx.mergeCollection(ONYXKEYS.COLLECTION.REPORT, {
    [`${ONYXKEYS.COLLECTION.REPORT}${report1.reportID}`]: report1,
    [`${ONYXKEYS.COLLECTION.REPORT}${report2.reportID}`]: report2,
    [`${ONYXKEYS.COLLECTION.REPORT}${report3.reportID}`]: report3,
});

Subscribing to Collections

There are several ways to subscribe to these keys:

const MyComponent = () => {
    const [allReports] = useOnyx(ONYXKEYS.COLLECTION.REPORT);

    return (
        <View>
            {/* Render with allReports data */}
        </View>
    );
};

export default MyComponent;

This will add a prop to the component called allReports which is an object of collection member key/values. Changes to the individual member keys will modify the entire object and new props will be passed with each individual key update. The prop doesn't update on the initial rendering of the component until the entire collection has been read out of Onyx.

Onyx.connect({key: ONYXKEYS.COLLECTION.REPORT}, callback: (memberValue, memberKey) => {...});

This will fire the callback once per member key depending on how many collection member keys are currently stored. Changes to those keys after the initial callbacks fire will occur when each individual key is updated.

Onyx.connect({
    key: ONYXKEYS.COLLECTION.REPORT,
    waitForCollectionCallback: true,
    callback: (allReports) => {...},
});

This final option forces Onyx.connect() to behave more like useOnyx() and only update the callback once with the entire collection initially and later with an updated version of the collection when individual keys update.

Performance Considerations When Using Collections

Be cautious when using collections as things can get out of hand if you have a subscriber hooked up to a collection key that has large numbers of individual keys. If this is the case, it is critical to use mergeCollection() over merge().

Remember, mergeCollection() will notify a subscriber only once with the total collected values whereas each call to merge() would re-render a connected component each time it is called. Consider this example where reports is an array of reports that we want to index and save.

// Bad
_.each(reports, report => Onyx.merge(`${ONYXKEYS.COLLECTION.REPORT}${report.reportID}`, report)); // -> A component using useOnyx() will have it's state updated with each iteration

// Good
const values = {};
_.each(reports, report => values[`${ONYXKEYS.COLLECTION.REPORT}${report.reportID}`] = report);
Onyx.mergeCollection(ONYXKEYS.COLLECTION.REPORT, values); // -> A component using useOnyx() will only have its state updated once

Clean up

To clear all data from Onyx we can use Onyx.clear().

function signOut() {
    Onyx.clear();
}

Storage Providers

Onyx.get, Onyx.set, and the rest of the API accesses the underlying storage differently depending on the platform

Under the hood storage access calls are delegated to a StorageProvider Some platforms (like web and desktop) might use the same storage provider

If a platform needs to use a separate library (like using MMVK for react-native) it should be added in the following way:

1. Create a StorageProvider.js at lib/storage/providers Reference an existing StorageProvider for the interface that has to be implemented
2. Update the factory at lib/storage/index.js and lib/storage/index.native.js to return the newly created Provider for the desired Platform(s)

API Reference

Docs

Storage Eviction

Different platforms come with varying storage capacities and Onyx has a way to gracefully fail when those storage limits are encountered. When Onyx fails to set or modify a key the following steps are taken:

1. Onyx looks at a list of recently accessed keys (access is defined as subscribed to or modified) and locates the key that was least recently accessed
2. It then deletes this key and retries the original operation

By default, Onyx will not evict anything from storage and will presume all keys are "unsafe" to remove unless explicitly told otherwise.

To flag a key as safe for removal:

• Add the key to the safeEvictionKeys option in Onyx.init(options)
• Implement canEvict in the Onyx config for each component subscribing to a key
• The key will only be deleted when all subscribers return true for canEvict

e.g.

Onyx.init({
    safeEvictionKeys: [ONYXKEYS.COLLECTION.REPORT_ACTIONS],
});

const ReportActionsView = ({reportID, isActiveReport}) => {
    const [reportActions] = useOnyx(
        `${ONYXKEYS.COLLECTION.REPORT_ACTIONS}${reportID}_`,
        {canEvict: () => !isActiveReport}
    );

    return (
        <View>
            {/* Render with reportActions data */}
        </View>
    );
};

export default ReportActionsView;

Benchmarks

Provide the captureMetrics boolean flag to Onyx.init to capture call statistics

Onyx.init({
    keys: ONYXKEYS,
    safeEvictionKeys: [ONYXKEYS.COLLECTION.REPORT_ACTIONS],
    captureMetrics: Config.BENCHMARK_ONYX,
});

At any point you can get the collected statistics using Onyx.getMetrics(). This will return an object containing totalTime, averageTime and summaries. summaries is a collection of statistics for each method it contains data about:

• method name
• total, max, min, average times for this method calls
• calls - a list of individual calls with each having: start time; end time; call duration; call arguments
  • start/end times are relative to application launch time - 0.00 being exactly at launch

If you wish to reset the metrics and start over use Onyx.resetMetrics()

Finally, there's a Onyx.printMetrics() method which prints human statistics information on the dev console. You can use this method during debugging. For example add an Onyx.printMetrics() line somewhere in code or call it through the dev console. It supports 3 popular formats MD - human friendly markdown, CSV and JSON. The default is MD if you want to print another format call Onyx.printMetrics({ format: 'csv' }) or Onyx.printMetrics({ format: 'json' }).

Sample output of Onyx.printMetrics()

### Onyx Benchmark
  - Total: 1.5min
  - Last call finished at: 12.55sec

|     method      | total time spent |    max    |   min    |    avg    | time last call completed | calls made |
|-----------------|-----------------:|----------:|---------:|----------:|-------------------------:|-----------:|
| Onyx:getAllKeys |           1.2min |   2.16sec |  0.159ms | 782.230ms |                 12.55sec |         90 |
| Onyx:merge      |          4.73sec |   2.00sec | 74.412ms | 591.642ms |                 10.24sec |          8 |
| Onyx:set        |          3.90sec | 846.760ms | 43.663ms | 433.056ms |                  7.47sec |          9 |
| Onyx:get        |          8.87sec |   2.00sec |  0.063ms |  61.998ms |                 10.24sec |        143 |


|                           Onyx:set                            |
|---------------------------------------------------------------|
| start time | end time  | duration  |           args           |
|-----------:|----------:|----------:|--------------------------|
|  291.042ms | 553.079ms | 262.037ms | session, [object Object] |
|  293.719ms | 553.316ms | 259.597ms | account, [object Object] |
|  294.541ms | 553.651ms | 259.109ms | network, [object Object] |
|  365.378ms | 554.246ms | 188.867ms | iou, [object Object]     |
|    1.08sec |   2.20sec |   1.12sec | network, [object Object] |
|    1.08sec |   2.20sec |   1.12sec | iou, [object Object]     |
|    1.17sec |   2.20sec |   1.03sec | currentURL, /            |

Debug mode

It can be useful to log why Onyx is calling setState() on a particular React component so that we can understand which key changed, what changed about the value, and the connected component that ultimately rendered as a result. When used correctly this can help isolate problem areas and unnecessary renders in the code. To enable this feature, pass debugSetState: true to the config and grep JS console logs for [Onyx-Debug].

Redux DevTools Extension

If you want to debug updates made to the local storage on the web app, you can use Redux DevTools Extension, which provides an easy to use GUI. This extension provides the following features:

• Each update (merge/set/clear/etc) made to the local storage is logged with the data that was written or erased.
• Check the state of the local storage at a certain point in time and compare it with it's previous state.

Currently this tool is only available on Web.

Installing the extension

To use the extension, simply install it from your favorite web browser store:

• Google Chrome
• Microsoft Edge
• Mozilla Firefox

After installing the extension, Onyx will automatically connect to it and start logging any updates made to the local storage.

Usage

The extension interface is pretty simple, on the left sidebar you can see all the updates made to the local storage, in ascending order, and on the right pane you can see the whole the current state, payload of an action and the diff between the previous state and the current state after the action was triggered.

The action logs use this naming convention:

@@INIT - Initial action which is triggered when Onyx connects to the extension. It's payload consists of the initial state.

merge/<KEY> - Merge action which is triggered when Onyx.merge() is called.

mergecollection/<KEY> - Merge action which is triggered when Onyx.mergeCollection() is called.

set/<KEY> - Set action which is triggered when Onyx.set() is called.

multiset/<KEY> - Set action which is triggered when Onyx.multiSet() is called.

CLEAR - Clear action which is triggered when Onyx.clear() is called.

Development

React Native bundles source code using the metro bundler. Until React Native 0.73, metro does not follow symlinks, so we can't use npm link to link a local version of Onyx during development. Fortunately, we have set up a workflow that's easy to follow and enables you to edit the Onyx source directly in the Onyx repo, and have those changes hot-reload in a React Native project in realtime.

1. In one terminal tab, open the react-native-onyx directory and run npm run build:watch
2. In another terminal tab, open your React Native project and run npx link publish <path_to_onyx_directory_on_your_machine>
3. Then run your React Native project as normal!

Now you can make changes directly to the react-native-onyx source code and your React Native project should-hot reload with those changes in realtime.

Note: If you want to unlink react-native-onyx, simply run npm install from your React Native project directory again. That will reinstall react-native-onyx from npm.