We instantiate an `ApolloServer` instance and pass our schema to it. We then subscribe to it with a [listener](/Programming_Languages/Node/Modules/Core/Node_JS_events_module.md#extending-the-eventemitter-class).
When we access the local URL we are able to access the Apollo server using the Explorer GUI. This automatically loads our schema and is basically a more fancy version of GraphiQL:
It makes it easy to read descriptions of the dataypes and to construct queries by clicking to insert fields.
### Adding some mock data
We are not connected to a database yet but we can create a mock that will enable us to run test queries.
We do this just by updating the Apollo Server options. We can either use generic dummy data or provide our own mock.
#### Generic mock
```js
const server = new ApolloServer({ typeDefs, mocks: true });
A resolver is a [function](/Trash/Creating_a_GraphQL_server.md#resolvers) that populates data for a given query. It should have **the same name as the field for the query**. So far we have one query in our schema: `tracksForHome` which returns the tracks to be listed on the home page. We must therefore also name our resolver for this query `tracksForHome`.
It can fetch data from a single data source or multiple data sources (other servers, databases, REST APIs) and present this as a single integrated resource to the client, matching the shape requested.
The `resolvers` object's keys will correspond to the schema's types and fields. You distinguish resolves which directly correspond to a query in the schema from other resolver types by wraping them in `Query {}`.
Each resolver function has the same standard parameters that you can invoke when implementing the resolution: `resolverFunction(parent, args, context, info)`.
- an object comprising arguments provided for the given field by the client. For instance if the client requests a field with an accompanying `id` argument, `id` can be parsed via the `args` object
- shared state between different resolvers that contains essential connection parameters such as authentication, a database connection, or a `RESTDataSource` (see below). This will be typically instantiated via a class which is then invoked within the `ApolloServer` instance under the `dataSources` key.
A resolver can return data from multiple sources. One of the most common sources is a RESTful endpoint. Apollo provides a specific class for handling REST endpoints in your resolvers: `RESTDataSource`.
REST APIs fall victim to the "n + 1" problem: say you want to get an array of one resource type, then for each element returned you need to send another request using one of its properties to fetch a related resource.
This is implicit in the case of the `Track` type in the schema. Each `Track` has an `author` key but the `Author` type isn't embedded in `Track` it has to be fetched using an `id`. In a REST API, this would require a request to a separate end point for each `Track` returned, increasing the time complexity of the request.
Here is an example of `RESTDataSource` being used. It is just a class that can be extended and which provides inbuilt methods for running fetches against a REST API:
As our GraphQL server is sourcing data from a REST API, we can now integrate the `RESTDataSource` class with our resolver.
First thing, we need to instantiate an instance of our `TrackApi` class, otherwise we won't be able to use any of its methods in the resolver.
We will create an instance of this class and pass it into `ApolloServer` object we established at the beginning. We will pass it to the `dataSources` key. **This will allow us to access it from within the `context` parameter in our resolver function**
We can also get rid of the `mocks` object since we don't need it any more. We will replace it with our `resolvers` constant:
```diff
const server = new ApolloServer({
typeDefs,
- mocks,
+ resolvers,
+ dataSources: () => {
+ return {
+ trackApi: new TrackApi()
+ }
}
})
```
Now we can complete our resolver:
```js
const resolvers = {
Query: {
tracksForHome: (_, __, {dataSources}) => {},
return dataSources.trackApi.getTracksForHome()
},
};
```
So we destructure the `dataSources` object from the parent Apollo Server instance (in the place of the `context` parameter) which gives us access to our `trackApi` class. This resolver will now make the API request and return the tracks.
The `tracksForHome` query returns `Track` objects and these have a required `author` key that returns an `Author` type. So we are also going to need a resolver that can return the author data that will be populated along with `Track`.
We already have this functionality in our class: `getAuthor` so we just need to integrate it:
- Just as we nest the `tracksForHome` resolver under `Query`, we must nest `author` under `Track` to match the structure of the schema. This resolver doesn't respond to a query that is exposed to the client so it shouldn't go under `Query`.
* We invoke the `context` again when we destructure `dataSources` from the `ApolloServer` instance.
* This time we utilise the `args` parameter in the resolver since an `id` will be provided as a client-side [argument](/Databases/GraphQL/Apollo/Using_arguments_with_Apollo_Client.md) to return a specific author.
## The `useMutation` hook
We invoke the `useMutation` hook to issue mutations from the client-side.
As with queries and [query constants](/Databases/GraphQL/Apollo/Apollo_Client.md#query-constants)
