HOC
Deprecated React Apollo HOC API
Note: Official support for React Apollo higher order components ended in March 2020. This library is still included in the
@apollo/client
package, but it no longer receives feature updates or bug fixes.
Installation
The HOC library is included in the core @apollo/client
package:
1npm install @apollo/client
You then import the library's symbols from @apollo/client/react/hoc
.
graphql(query, [config])(component)
1import { graphql } from '@apollo/client/react/hoc';
The graphql()
function is the core of Apollo's HOC API. Use this function to create higher-order components that can execute queries and update reactively based on the data in your Apollo store.
The graphql()
function returns a function that "enhances" any component with reactive GraphQL capabilities. This follows the React higher-order component pattern that's also used by react-redux
’s connect
function.
The graphql()
function can only provide access to your GraphQL data if there is an <ApolloProvider/>
component higher up in your tree to provide an ApolloClient
instance that's used to fetch your data.
The behavior of your component enhanced with the graphql()
function will be different depending on if your GraphQL operation is a query, a mutation, or a subscription. See the appropriate API documentation for more information about the functionality and available options for each type.
Examples
You can use the graphql()
function like this:
1function TodoApp({ data: { todos } }) {
2 return (
3 <ul>
4 {todos.map(({ id, text }) => (
5 <li key={id}>{text}</li>
6 ))}
7 </ul>
8 );
9}
10
11export default graphql(gql`
12 query TodoAppQuery {
13 todos {
14 id
15 text
16 }
17 }
18`)(TodoApp);
You can also define an intermediate function and hook up your component with the graphql()
function like this:
1// Create our enhancer function.
2const withTodoAppQuery = graphql(gql`query TodoAppQuery { ... }`);
3
4// Enhance our component.
5const TodoAppWithData = withTodoAppQuery(TodoApp);
6
7// Export the enhanced component.
8export default TodoAppWithData;
The config
object
Before we look into the specific behaviors of each operation, let's look at the config
object. The config
object is the second argument you pass into the graphql()
function, after your GraphQL document. The config is optional and allows you to add some custom behavior to your higher order component.
1export default graphql(
2 gql`query MyQuery { ... }`,
3 config, // <- The `config` object.
4)(MyComponent);
Lets go through all of the properties that may live on your config
object.
config.options
config.options
is an object or a function that allows you to define the specific behavior your component should use in handling your GraphQL data.
The specific options available for configuration depend on the operation you pass as the first argument to graphql()
. There are options specific to queries and mutations.
You can define config.options
as a plain object, or you can compute your options from a function that takes the component’s props as an argument.
Example
1export default graphql(gql`query MyQuery { ... }`, {
2 options: {
3 // Options go here.
4 },
5})(MyComponent);
1export default graphql(gql`query MyQuery { ... }`, {
2 options: props => ({
3 // Options are computed from `props` here.
4 }),
5})(MyComponent);
config.props
The config.props
property allows you to define a map function that takes the props
(and optionally lastProps
) added by the graphql()
function (props.data
for queries and props.mutate
for mutations) and allows you to compute a new props
(and optionally lastProps
) object that will be provided to the component that graphql()
is wrapping.
The function you define behaves almost exactly like mapProps
from Recompose providing the same benefits without the need for another library.
config.props
is most useful when you want to abstract away complex function calls into a simple prop that you can pass down to your component.
Another benefit of config.props
is that it also allows you to decouple your pure UI components from your GraphQL and Apollo concerns. You can write your pure UI components in one file and then keep the logic required for them to interact with the store in a completely different place in your project. You can accomplish this by your pure UI components only asking for the props needed to render and config.props
can contain the logic to provide exactly the props your pure component needs from the data provided by your GraphQL API.
Example
This example uses props.data.fetchMore
.
1export default graphql(gql`query MyQuery { ... }`, {
2 props: ({ data: { fetchMore } }) => ({
3 onLoadMore: () => {
4 fetchMore({ ... });
5 },
6 }),
7})(MyComponent);
8
9function MyComponent({ onLoadMore }) {
10 return (
11 <button onClick={onLoadMore}>
12 Load More!
13 </button>
14 );
15}
To access props that are not added by the graphql()
function, use the ownProps
keyword. For example:
1export default graphql(gql`query MyQuery { ... }`, {
2 props: ({ data: { liveImage }, ownProps: { loadingImage } }) => ({
3 image: liveImage || loadingImage,
4 }),
5})(MyComponent);
To access lastProps
, use the second argument of config.props
. For example:
1export default graphql(gql`query MyQuery { ... }`, {
2 props: ({ data: { liveImage } }, lastProps) => ({
3 image: liveImage,
4 lastImage: lastProps.data.liveImage,
5 }),
6})(MyComponent);
config.skip
If config.skip
is true
, then all of the React Apollo code is skipped entirely. Your component behaves as if the graphql()
function isn't there at all.
You can also pass a function to config.skip
. If you do, the function takes your component's props and should return a boolean. If the function returns true
, then the skip behavior goes into effect.
config.skip
is especially useful if you want to use a different query based on some prop. You can see this in an example below.
Example
1export default graphql(gql`query MyQuery { ... }`, {
2 skip: props => !!props.skip,
3})(MyComponent);
The following example uses the compose
function to use multiple graphql()
enhancers at once.
1export default compose(
2 graphql(gql`query MyQuery1 { ... }`, { skip: props => !props.useQuery1 }),
3 graphql(gql`query MyQuery2 { ... }`, { skip: props => props.useQuery1 }),
4)(MyComponent);
5
6function MyComponent({ data }) {
7 // The data may be from `MyQuery1` or `MyQuery2` depending on the value
8 // of the prop `useQuery1`.
9 console.log(data);
10}
config.name
This property allows you to configure the name of the prop that gets passed down to your component. By default, if the GraphQL document you pass into graphql()
is a query, then your prop is named data
. If you pass a mutation, then your prop will be named mutate
. These default names collide when you use multiple queries or mutations with the same component. To avoid collisions, use config.name
to specify a different name.
Example
This example uses the compose
function to use multiple graphql()
HOCs together.
1export default compose(
2 graphql(gql`mutation CreateTodoMutation (...) { ... }`, { name: 'createTodo' }),
3 graphql(gql`mutation UpdateTodoMutation (...) { ... }`, { name: 'updateTodo' }),
4 graphql(gql`mutation DeleteTodoMutation (...) { ... }`, { name: 'deleteTodo' }),
5)(MyComponent);
6
7function MyComponent(props) {
8 // Instead of the default prop name, `mutate`,
9 // we have three different prop names.
10 console.log(props.createTodo);
11 console.log(props.updateTodo);
12 console.log(props.deleteTodo);
13
14 return null;
15}
config.withRef
By setting config.withRef
to true
, you can get the instance of your wrapped component from your higher-order GraphQL component using a getWrappedInstance
method available on the instance of your higher-order GraphQL component.
You might want to set this to true
when you want to call functions or access properties that are defined on your wrapped component’s class instance.
Example
This example uses the React ref
feature.
1class MyComponent extends Component {
2 saySomething() {
3 console.log('Hello, world!');
4 }
5
6 render() {
7 // ...
8 }
9}
10
11const MyGraphQLComponent = graphql(gql`query MyQuery { ... }`, { withRef: true })(
12 MyComponent,
13);
14
15class MyContainerComponent extends Component {
16 render() {
17 return (
18 <MyGraphQLComponent
19 ref={component => {
20 const wrappedInstance = component.getWrappedInstance();
21 assert(wrappedInstance instanceof MyComponent);
22 // We can call methods on the component class instance.
23 wrappedInstance.saySomething();
24 }}
25 />
26 );
27 }
28}
config.alias
Use this property to configure the name of your higher order component wrapper. For example, if you set config.alias
to 'withCurrentUser'
, your wrapper component display name becomes withCurrentUser(${WrappedComponent.displayName})
instead of Apollo(${WrappedComponent.displayName})
.
The default display name for React Apollo components is Apollo(${WrappedComponent.displayName})
. This pattern is used by most React libraries that make use of higher order components. However, this might get confusing when you are using more than one higher order component and you look at the React Devtools.
Example
This example uses the compose
function to use multiple graphql()
HOCs together.
1export default compose(
2 graphql(gql`query MyQuery { ... }`, { alias: 'withCurrentUser' }),
3 graphql(gql`query MyQuery { ... }`, { alias: 'withList' }),
4)(MyComponent);
graphql() options for queries
props.data
The higher-order component created with graphql()
feeds a data
prop into your component. Like so:
1render() {
2 const { data } = this.props; // <- The `data` prop.
3}
The data
prop contains the data fetched from your query in addition to some other useful information and functions to control the lifecycle of your GraphQL-connected component. So for example, if we had a query that looked like:
1query ViewerAndTodos {
2 viewer {
3 name
4 }
5 todos {
6 text
7 }
8}
Your data
prop would contain that data:
1render() {
2 const { data } = this.props;
3
4 console.log(data.viewer); // <- The data returned by your query for `viewer`.
5 console.log(data.todos); // <- The data returned by your query for `todos`.
6}
The data
prop has some other useful properties which can be accessed directly from data
. For example, data.loading
or data.error
. These properties are documented below.
Make sure to always check data.loading
and data.error
in your components before rendering. Properties like data.todos
which contain your app’s data may be undefined while your component is performing its initial fetch. Checking data.loading
and data.error
helps you avoid any issues with undefined data. Such checks may look like:
1render() {
2 const { data: { loading, error, todos } } = this.props;
3 if (loading) {
4 return <p>Loading...</p>;
5 }
6 if (error) {
7 return <p>Error!</p>;
8 }
9 return (
10 <ul>
11 {todos.map(({ id, text }) => (
12 <li key={id}>{text}</li>
13 ))}
14 </ul>
15 );
16}
data.loading
A boolean representing whether or not a query request is currently in flight for this component. This means that a query request has been sent using your network interface, and we have not yet gotten a response back. Use this property to render a loading component.
However, just because data.loading
is true it does not mean that you won’t have data. For instance, if you already have data.todos
, but you want to get the latest todos from your API data.loading
might be true, but you will still have the todos from your previous request.
There are multiple different network states that your query may be in. If you want to see what the network state of your component is in more detail then refer to data.networkStatus
.
Example:
1function MyComponent({ data: { loading } }) {
2 if (loading) {
3 return <div>Loading...</div>;
4 } else {
5 // ...
6 }
7}
8
9export default graphql(gql`query MyQuery { ... }`)(MyComponent);
data.error
If an error occurred then this property will be an instance of ApolloError
. If you do not handle this error you will get a warning in your console that says something like: "Unhandled (in react-apollo) Error: ..."
.
Example:
1function MyComponent({ data: { error } }) {
2 if (error) {
3 return <div>Error!</div>;
4 } else {
5 // ...
6 }
7}
8
9export default graphql(gql`query MyComponentQuery { ... }`)(MyComponent);
data.networkStatus
data.networkStatus
is useful if you want to display a different loading indicator (or no indicator at all) depending on your network status as it provides a more detailed view into the state of a network request on your component than data.loading
does. data.networkStatus
is an enum with different number values between 1 and 8. These number values each represent a different network state.
loading
: The query has never been run before and the request is now pending. A query will still have this network status even if a result was returned from the cache, but a query was dispatched anyway.setVariables
: If a query’s variables change and a network request was fired then the network status will besetVariables
until the result of that query comes back. React users will see this whenoptions.variables
changes on their queries.fetchMore
: Indicates thatfetchMore
was called on this query and that the network request created is currently in flight.refetch
: It means thatrefetch
was called on a query and the refetch request is currently in flight.Unused.
poll
: Indicates that a polling query is currently in flight. So for example if you are polling a query every 10 seconds then the network status will switch topoll
every 10 seconds whenever a poll request has been sent but not resolved.ready
: No request is in flight for this query, and no errors happened. Everything is OK.error
: No request is in flight for this query, but one or more errors were detected.
If the network status is less then 7 then it is equivalent to data.loading
being true. In fact you could replace all of your data.loading
checks with data.networkStatus < 7
and you would not see a difference. It is recommended that you use data.loading
, however.
Example:
1function MyComponent({ data: { networkStatus } }) {
2 if (networkStatus === 6) {
3 return <div>Polling!</div>;
4 } else if (networkStatus < 7) {
5 return <div>Loading...</div>;
6 } else {
7 // ...
8 }
9}
10
11export default graphql(gql`query MyComponentQuery { ... }`)(MyComponent);
data.variables
The variables that Apollo used to fetch data from your GraphQL endpoint. This property is helpful if you want to render some information based on the variables that were used to make a request against your server.
Example:
1function MyComponent({ data: { variables } }) {
2 return (
3 <div>
4 Query executed with the following variables:
5 <code>{JSON.stringify(variables)}</code>
6 </div>
7 );
8}
9
10export default graphql(gql`query MyComponentQuery { ... }`)(MyComponent);
data.refetch(variables)
Forces your component to refetch the query you defined in the graphql()
function. This method is helpful when you want to reload the data in your component, or retry a fetch after an error.
data.refetch
returns a promise that resolves with the new data fetched from your API once the query has finished executing. The promise will reject if the query failed.
The data.refetch
function takes a single variables
object argument. The variables
argument will replace variables
used with either the query
option or the query from your graphql()
HOC (depending on whether or not you specified a query
) option to refetch the query you defined in the graphql()
function.
Example:
1function MyComponent({ data: { refetch } }) {
2 return <button onClick={() => refetch()}>Reload</button>;
3}
4
5export default graphql(gql`query MyComponentQuery { ... }`)(MyComponent);
data.fetchMore(options)
The data.fetchMore
function allows you to do pagination with your query component. To learn more about pagination with data.fetchMore
, be sure to read the pagination documentation.
data.fetchMore
returns a promise that resolves once the query executed to fetch more data has resolved.
The data.fetchMore
function takes a single options
object argument. The options
argument may take the following properties:
[query]
: This is an optional GraphQL document created with thegql
GraphQL tag. If you specify aquery
then that query will be fetched when you calldata.fetchMore
. If you do not specify aquery
, then the query from yourgraphql()
HOC will be used.[variables]
: The optional variables you may provide that will be used with either thequery
option or the query from yourgraphql()
HOC (depending on whether or not you specified aquery
).updateQuery(previousResult, { fetchMoreResult, variables })
: This is the required function you define that will actually update your paginated list. The first argument,previousResult
, will be the previous data returned by the query you defined in yourgraphql()
function. The second argument is an object with two properties,fetchMoreResult
andvariables
.fetchMoreResult
is the data returned by the new fetch that used thequery
andvariables
options fromdata.fetchMore
.variables
are the variables that were used when fetching more data. Using these arguments you should return a new data object with the same shape as the GraphQL query you defined in yourgraphql()
function. See an example of this below, and also make sure to read the pagination documentation.
Example:
1data.fetchMore({
2 updateQuery: (previousResult, { fetchMoreResult, variables }) => {
3 return {
4 ...previousResult,
5 // Add the new feed data to the end of the old feed data.
6 feed: [...previousResult.feed, ...fetchMoreResult.feed],
7 };
8 },
9});
data.subscribeToMore(options)
This function will set up a subscription, triggering updates whenever the server sends a subscription publication. This requires subscriptions to be set up on the server to properly work. Check out the subscriptions guide for more information on getting this set up.
This function returns an unsubscribe
function handler which can be used to unsubscribe later.
A common practice is to wrap the subscribeToMore
call within getDerivedStateFromProps
and perform the subscription after the original query has completed. To ensure the subscription isn't created multiple times, you can add it to component state. See the example for more details.
[document]
: Document is a required property that accepts a GraphQL subscription created with thegql
template string tag. It should contain a single GraphQL subscription operation with the data that will be returned.[variables]
: The optional variables you may provide that will be used with thedocument
option.[updateQuery]
: An optional function that runs every time the server sends an update. This modifies the results of the HOC query. The first argument,previousResult
, will be the previous data returned by the query you defined in yourgraphql()
function. The second argument is an object with two properties.subscriptionData
is result of the subscription.variables
is the variables object used with the subscription query. Using these arguments you should return a new data object with the same shape as the GraphQL query you defined in yourgraphql()
function. This is similar to thefetchMore
callback.[onError]
: An optional error callback.
In order to update the query's store with the result of the subscription, you must specify either the updateQuery
option in subscribeToMore
or the reducer
option in your graphql()
function.
Example:
1class SubscriptionComponent extends Component {
2 state = {
3 subscriptionParam: null,
4 unsubscribe: null,
5 };
6
7 static getDerivedStateFromProps(nextProps, prevState) {
8 if (!nextProps.data.loading) {
9 // Check for existing subscription
10 if (prevState.unsubscribe) {
11 // Only unsubscribe/update state if subscription variable has changed
12 if (prevState.subscriptionParam === nextProps.subscriptionParam) {
13 return null;
14 }
15 prevState.unsubscribe();
16 }
17
18 return {
19 // Subscribe
20 unsubscribe: nextProps.data.subscribeToMore({
21 document: gql`subscription MySubscription {...}`,
22 variables: {
23 param: nextProps.subscriptionParam,
24 },
25 updateQuery: (previousResult, { subscriptionData, variables }) => {
26 // Perform updates on previousResult with subscriptionData
27 return updatedResult;
28 },
29 }),
30 // Store subscriptionParam in state for next update
31 subscriptionParam: nextProps.subscriptionParam,
32 };
33 }
34
35 return null;
36 }
37
38 render() {
39 ...
40 }
41}
data.startPolling(interval)
This function will set up an interval and send a fetch request every time that interval ellapses. The function takes only one integer argument which allows you to configure how often you want your query to be executed in milliseconds. In other words, the interval
argument represents the milliseconds between polls.
Polling is a good way to keep the data in your UI fresh. By refetching your data every 5,000 milliseconds (or 5 seconds, for example) you may effectively emulate realtime data without needing to build up a realtime backend.
If you call data.startPolling
when your query is already polling then the current polling process will be cancelled and a new process will be started with the interval you specified.
You may also use options.pollInterval
to start polling immediately after your component mounts. It is recommend that you use options.pollInterval
if you don’t need to arbitrarily start and stop polling.
If you set your interval
to 0 then that means no polling instead of executing a request every JavaScript event loop tick.
Example:
1class MyComponent extends Component {
2 componentDidMount() {
3 // In this specific case you may want to use `options.pollInterval` instead.
4 this.props.data.startPolling(1000);
5 }
6
7 render() {
8 // ...
9 }
10}
11
12export default graphql(gql`query MyComponentQuery { ... }`)(MyComponent);
data.stopPolling()
By calling this function you will stop any current polling process. Your query will not start polling again until you call data.startPolling
.
Example:
1class MyComponent extends Component {
2 render() {
3 return (
4 <div>
5 <button
6 onClick={() => {
7 this.props.data.startPolling(1000);
8 }}
9 >
10 Start Polling
11 </button>
12 <button
13 onClick={() => {
14 this.props.data.stopPolling();
15 }}
16 >
17 Stop Polling
18 </button>
19 </div>
20 );
21 }
22}
23
24export default graphql(gql`query MyComponentQuery { ... }`)(MyComponent);
data.updateQuery(updaterFn)
This function allows you to update the data for your query outside of the context of any mutation, subscription, or fetch. This function only takes a single argument which will be another function. The argument function has the following signature:
1(previousResult, { variables }) => nextResult
The first argument will be the data for your query that currently exists in the store, and you are expected to return a new data object with the same shape. That new data object will be written to the store and any components tracking that data will be updated reactively.
The second argument is an object with a single property, variables
. The variables
property allows you to see what variables were used when reading the previousResult
from the store.
This method will not update anything on the server. It will only update data in your client cache and if you reload your JavaScript environment then your update will disappear.
Example:
1data.updateQuery(previousResult => ({
2 ...previousResult,
3 count: previousResult.count + 1,
4}));
config.options
An object or function that returns an object of options that are used to configure how the query is fetched and updated.
If config.options
is a function then it will take the component’s props as its first argument.
The options available for use in this object depend on the operation type you pass in as the first argument to graphql()
. The references below will document which options are available when your operation is a query. To see what other options are available for different operations, see the generic documentation for config.options
.
Example:
1export default graphql(gql`query MyQuery { ... }`, {
2 options: {
3 // Options go here.
4 },
5})(MyComponent);
1export default graphql(gql`query MyQuery { ... }`, {
2 options: props => ({
3 // Options are computed from `props` here.
4 }),
5})(MyComponent);
options.variables
The variables that will be used when executing the query operation. These variables should correspond with the variables that your query definition accepts. If you define config.options
as a function then you may compute your variables from your props.
Example:
1export default graphql(
2 gql`
3 query MyQuery ($width: Int!, $height: Int!) {
4 ...
5 }
6`,
7 {
8 options: props => ({
9 variables: {
10 width: props.size,
11 height: props.size,
12 },
13 }),
14 },
15)(MyComponent);
options.fetchPolicy
The fetch policy is an option that allows you to specify how you want your component to interact with the Apollo Client cache. By default, your component will try to read from the cache first, and if the full data for your query is in the cache then Apollo simply returns the data from the cache. If the full data for your query is not in the cache then Apollo will execute your request using your network interface. By changing this option you can change this behavior.
For a list of supported fetch policies, see Setting a fetch policy.
Example:
1export default graphql(gql`query MyQuery { ... }`, {
2 options: { fetchPolicy: 'cache-and-network' },
3})(MyComponent);
options.errorPolicy
The error policy is an option which allows you to specify how you want your component to handle errors that can happen when fetching data from GraphQL. There are two types of errors that can happen during your request; a runtime error on the client or server which results in no data, or some GraphQL errors which may be delivered alongside actual data. In order to control how your UI interacts with these errors, you can use the error policy to tell Apollo when you want to know about GraphQL Errors or not!
Valid errorPolicy
values are:
none
: This is the default value where we treat GraphQL errors as runtime errors. Apollo will discard any data that came back with the request and render your component with anerror
prop.ignore
: Much likenone
, this causes Apollo to ignore any data from your server, but it also won't update your UI aside from setting the loading state back to false.all
: Selecting all means you want to be notified any time there are any GraphQL errors. It will render your component with any data from the request and any errors with their information. It is particularly helpful for server side rendering so your UI always shows something
Example:
1export default graphql(gql`query MyQuery { ... }`, {
2 options: { errorPolicy: 'all' },
3})(MyComponent);
options.pollInterval
The interval in milliseconds at which you want to start polling. Whenever that number of milliseconds elapses your query will be executed using the network interface and another execution will be scheduled using the configured number of milliseconds.
This option will start polling your query immediately when the component mounts. If you want to start and stop polling dynamically then you may use data.startPolling
and data.stopPolling
.
If you set options.pollInterval
to 0 then that means no polling instead of executing a request every JavaScript event loop tick.
Example:
1export default graphql(gql`query MyQuery { ... }`, {
2 options: { pollInterval: 5000 },
3})(MyComponent);
options.notifyOnNetworkStatusChange
Whether or not updates to the network status or network error should trigger re-rendering of your component.
The default value is false
.
Example:
1export default graphql(gql`query MyQuery { ... }`, {
2 options: { notifyOnNetworkStatusChange: true },
3})(MyComponent);
options.context
With the flexibility and power of Apollo Link being part of Apollo Client, you may want to send information from your operation straight to a link in your network chain! This can be used to do things like set headers
on HTTP requests from props, control which endpoint you send a query to, and so much more depending on what links your app is using. Everything under the context
object gets passed directly to your network chain. For more information about using context, check out the HttpLink
context docs
partialRefetch
If true
, perform a query refetch
if the query result is marked as being partial, and the returned data is reset to an empty Object by the Apollo Client QueryManager
(due to a cache miss).
The default value is false
for backwards-compatibility's sake, but should be changed to true for most use-cases.
Example:
1export default graphql(gql`query MyQuery { ... }`, {
2 options: { partialRefetch: true },
3})(MyComponent);
graphql() options for mutations
props.mutate
The higher order component created when you pass a mutation to graphql()
will provide your component with a single prop named mutate
. Unlike the data
prop which you get when you pass a query to graphql()
, mutate
is a function.
The mutate
function will actually execute your mutation using the network interface therefore mutating your data. The mutate
function will also then update your cache in ways you define.
To learn more about how mutations work, be sure to check out the mutations usage documentation.
The mutate
function accepts the same options that config.options
for mutations accepts, so make sure to read through the documentation for that to know what you can pass into the mutate
function.
The reason the mutate
function accepts the same options is that it will use the options from config.options
by default. When you pass an object into the mutate
function you are just overriding what is already in config.options
.
Example:
1function MyComponent({ mutate }) {
2 return (
3 <button
4 onClick={() => {
5 mutate({
6 variables: { foo: 42 },
7 });
8 }}
9 >
10 Mutate
11 </button>
12 );
13}
14
15export default graphql(gql`mutation MyMutation { ... }`)(MyComponent);
config.options
An object or function that returns an object of options that are used to configure how the query is fetched and updated.
If config.options
is a function then it will take the component’s props as its first argument.
The options available for use in this object depend on the operation type you pass in as the first argument to graphql()
. The references below will document which options are available when your operation is a mutation. To see what other options are available for different operations, see the generic documentation for config.options
.
The properties accepted in this options object may also be accepted by the props.mutate
function. Any options passed into the mutate
function will take precedence over the options defined in the config
object.
Example:
1export default graphql(gql`mutation MyMutation { ... }`, {
2 options: {
3 // Options go here.
4 },
5})(MyComponent);
1export default graphql(gql`mutation MyMutation { ... }`, {
2 options: props => ({
3 // Options are computed from `props` here.
4 }),
5})(MyComponent);
1function MyComponent({ mutate }) {
2 return (
3 <button
4 onClick={() => {
5 mutate({
6 // Options are component from `props` and component state here.
7 });
8 }}
9 >
10 Mutate
11 </button>
12 );
13}
14
15export default graphql(gql`mutation MyMutation { ... }`)(MyComponent);
options.variables
The variables which will be used to execute the mutation operation. These variables should correspond to the variables that your mutation definition accepts. If you define config.options
as a function, or you pass variables into the props.mutate
function then you may compute your variables from props and component state.
Example:
1export default graphql(
2 gql`
3 mutation MyMutation ($foo: String!, $bar: String!) {
4 ...
5 }
6`,
7 {
8 options: props => ({
9 variables: {
10 foo: props.foo,
11 bar: props.bar,
12 },
13 }),
14 },
15)(MyComponent);
options.optimisticResponse
Often when you mutate data it is fairly easy to predict what the response of the mutation will be before asking your server. The optimistic response option allows you to make your mutations feel faster by simulating the result of your mutation in your UI before the mutation actually finishes.
To learn more about the benefits of optimistic data and how to use it be sure to read the recipe on Optimistic UI.
This optimistic response will be used with options.update
and options.updateQueries
to apply an update to your cache which will be rolled back before applying the update from the actual response.
Example:
1function MyComponent({ newText, mutate }) {
2 return (
3 <button
4 onClick={() => {
5 mutate({
6 variables: {
7 text: newText,
8 },
9 // The optimistic response has all of the fields that are included in
10 // the GraphQL mutation document below.
11 optimisticResponse: {
12 createTodo: {
13 id: -1, // A temporary id. The server decides the real id.
14 text: newText,
15 completed: false,
16 },
17 },
18 });
19 }}
20 >
21 Add Todo
22 </button>
23 );
24}
25
26export default graphql(gql`
27 mutation CreateTodo ($text: String!) {
28 createTodo(text: $text) {
29 id
30 text
31 completed
32 }
33 }
34`)(MyComponent);
options.update
This option allows you to update your store based on your mutation’s result. By default Apollo Client will update all of the overlapping nodes in your store. Anything that shares the same id as returned by the dataIdFromObject
you defined will be updated with the new fields from your mutation results. However, sometimes this alone is not sufficient. Sometimes you may want to update your cache in a way that is dependent on the data currently in your cache. For these updates you may use an options.update
function.
options.update
takes two arguments. The first is an instance of a DataProxy
object which has some methods which will allow you to interact with the data in your store. The second is the response from your mutation - either the optimistic response, or the actual response returned by your server (see the mutation result described in the mutation render prop section for more details).
In order to change the data in your store call methods on your DataProxy
instance like writeQuery
and writeFragment
. This will update your cache and reactively re-render any of your GraphQL components which are querying affected data.
To read the data from the store that you are changing, make sure to use methods on your DataProxy
like readQuery
and readFragment
.
For more information on updating your cache after a mutation with the options.update
function make sure to read the Apollo Client technical documentation on the subject.
Example:
1const query = gql`query GetAllTodos { todos { ... } }`;
2
3export default graphql(
4 gql`
5 mutation CreateTodo ($text: String!) {
6 createTodo(text: $text) { ... }
7 }
8`,
9 {
10 options: {
11 update: (proxy, { data: { createTodo } }) => {
12 const data = proxy.readQuery({ query });
13 data.todos.push(createTodo);
14 proxy.writeQuery({ query, data });
15 },
16 },
17 },
18)(MyComponent);
options.refetchQueries
Sometimes when you make a mutation you also want to update the data in your queries so that your users may see an up-to-date user interface. There are more fine-grained ways to update the data in your cache which include options.updateQueries
, and options.update
. However, you can update the data in your cache more reliably at the cost of efficiency by using options.refetchQueries
.
options.refetchQueries
will execute one or more queries using your network interface and will then normalize the results of those queries into your cache. Allowing you to potentially refetch queries you had fetched before, or fetch brand new queries.
options.refetchQueries
is either an array of strings or objects, or a function which takes the result of the mutation and returns an array of strings or objects.
If options.refetchQueries
is an array of strings then Apollo Client will look for any queries with the same names as the provided strings and will refetch those queries with their current variables. So for example if you have a GraphQL query component with a query named Comments
(the query may look like: query Comments { ... }
), and you pass an array of strings containing Comments
to options.refetchQueries
then the Comments
query will be re-executed and when it resolves the latest data will be reflected in your UI.
If options.refetchQueries
is an array of objects then the objects must have two properties:
query
: Query is a required property that accepts a GraphQL query created with thegql
template string tag. It should contain a single GraphQL query operation that will be executed once the mutation has completed.[variables]
: Is an optional object of variables that is required whenquery
accepts some variables.
If an array of objects with this shape is specified then Apollo Client will refetch these queries with their variables.
Example:
1export default graphql(gql`mutation MyMutation { ... }`, {
2 options: {
3 refetchQueries: ['CommentList', 'PostList'],
4 },
5})(MyComponent);
1import { COMMENT_LIST_QUERY } from '../components/CommentList';
2
3export default graphql(gql`mutation MyMutation { ... }`, {
4 options: props => ({
5 refetchQueries: [
6 {
7 query: COMMENT_LIST_QUERY,
8 },
9 {
10 query: gql`
11 query GetPostById ($id: ID!) {
12 post(id: $id) {
13 commentCount
14 }
15 }
16 `,
17 variables: {
18 id: props.postID,
19 },
20 },
21 ],
22 }),
23})(MyComponent);
1export default graphql(gql`mutation MyMutation { ... }`, {
2 options: {
3 refetchQueries: mutationResult => ['CommentList', 'PostList'],
4 },
5})(MyComponent);
Please note that refetched queries are handled asynchronously, and by default are not necessarily completed before the mutation has completed. If you want to make sure refetched queries are completed before the mutation is considered done (or resolved), set options.awaitRefetchQueries
to true
.
options.awaitRefetchQueries
Queries refetched using options.refetchQueries
are handled asynchronously, which means by default they are not necessarily completed before the mutation has completed. Setting options.awaitRefetchQueries
to true
will make sure refetched queries are completed before the mutation is considered done (or resolved). options.awaitRefetchQueries
is false
by default.
options.updateQueries
Note: We recommend using options.update
instead of updateQueries
. updateQueries
will be removed in the next version of Apollo Client
This option allows you to update your store based on your mutation’s result. By default Apollo Client will update all of the overlapping nodes in your store. Anything that shares the same id as returned by the dataIdFromObject
you defined will be updated with the new fields from your mutation results. However, sometimes this alone is not sufficient. Sometimes you may want to update your cache in a way that is dependent on the data currently in your cache. For these updates you may use an options.updateQueries
function.
options.updateQueries
takes an object where query names are the keys and reducer functions are the values. If you are familiar with Redux, defining your options.updateQueries
reducers is very similar to defining your Redux reducers. The object looks something like this:
1{
2 Comments: (previousData, { mutationResult, queryVariables }) => nextData,
3}
Make sure that the key of your options.updateQueries
object corresponds to an actual query that you have made somewhere else in your app. The query name will be the name you put after specifying the query
operation type. So for example in the following query:
1query Comments {
2 entry(id: 5) {
3 comments {
4 ...
5 }
6 }
7}
The query name would be Comments
. If you have not executed a GraphQL query with the name of Comments
before somewhere in your application, then the reducer function will never be run by Apollo and the key/value pair in options.updateQueries
will be ignored.
The first argument to the function you provide as the value for your object will be the previous data for your query. So if your key is Comments
then the first argument will be the last data object that was returned for your Comments
query, or the current object that is being rendered by any component using the Comments
query.
The second argument to your function value will be an object with three properties:
mutationResult
: ThemutationResult
property will represent the result of your mutation after hitting the server. If you provided anoptions.optimisticResponse
thenmutationResult
may be that object.queryVariables
: The last set of variables that the query was executed with. This is helpful because when you specify the query name it will only update the data in the store for your current variable set.queryName
: This is the name of the query you are updating. It is the same name as the key you provided tooptions.updateQueries
.
The return value of your options.updateQueries
functions must have the same shape as your first previousData
argument. However, you must not mutate the previousData
object. Instead you must create a new object with your changes. Just like in a Redux reducer.
Example:
1export default graphql(
2 gql`
3 mutation SubmitComment ($text: String!) {
4 submitComment(text: $text) { ... }
5 }
6`,
7 {
8 options: {
9 updateQueries: {
10 Comments: (previousData, { mutationResult }) => {
11 const newComment = mutationResult.data.submitComment;
12 // Note how we return a new copy of `previousData` instead of mutating
13 // it. This is just like a Redux reducer!
14 return {
15 ...previousData,
16 entry: {
17 ...previousData.entry,
18 comments: [newComment, ...previousData.entry.comments],
19 },
20 };
21 },
22 },
23 },
24 },
25)(MyComponent);
withApollo(component)
1import { withApollo } from '@apollo/client/react/hoc';
An enhancer that provides direct access to your ApolloClient
instance. This is useful if you want to do custom logic with Apollo, such as executing one-off queries. By calling this function with the component you want to enhance, withApollo()
creates a new component that passes an instance of ApolloClient
as a client
prop.
Most of the time you want to use graphql()
instead of withApollo()
. graphql()
provides helpful features for working with your GraphQL data. You should only use withApollo()
if you want the GraphQL client without any of these features.
This will only be able to provide access to your client if there is an <ApolloProvider/>
component higher up in your tree to actually provide the client.
Example:
1function MyComponent({ client }) {
2 console.log(client);
3}
4
5export default withApollo(MyComponent);