Contributing fields to an entity

Overview

Even though we've defined the Listing entity in our reviews subgraph and contributed fields to it, we haven't told it how to resolve these fields.

In this lesson, we will:

Create a reference resolver
Populate the reviews and overallRating fields with data
Learn about the @DgsEntityFetcher annotation

The reference resolver

When we query for a particular listing and its review data, we need a way to tell the reviews subgraph which listing we're talking about. Then the reviews subgraph will understand what data to provide!

To do exactly this, the router passes along the entity representation we talked about in the last lesson: the minimum information required for the reviews subgraph to put together some idea of which listing it's fetching data for.

Example Listing entity representation

{
  "__typename": "Listing",
  "id": "listing-3"
}

But where in the reviews subgraph does this entity representation actually go?

This is just the piece we're missing: the reference resolver. The reference resolver is the special method that can receive an entity representation from the router, create a new instance of a Listing from that representation data, and return it.

Let's see what this looks like in code, and walk through the process step-by-step.

Adding `resolveListingReference`

Let's open up our datafetcher class in the reviews subgraph, under com.example.reviews/datafetchers/ReviewsDataFetcher.

Right away, let's take care of adding some imports at the top of the file. We'll use these soon.

import java.util.Map;
import com.example.reviews.generated.types.Listing;

Note: Not seeing the Listing type in your generated folder? Try restarting the server!

Make some space down in the body of the class for our new method. To be extra clear about its job, we'll give it the name resolveListingReference. (But you can call this method whatever you want!)

datafetchers/ReviewsDataFetcher

public void resolveListingReference() {
  // TODO
}

This method will receive the entity representation, which looks something like the following snippet: just the __typename and the field we set as the entity's primary key, id.

{__typename=Listing, id=listing-1}

Let's give our method a parameter to hold this representation. It's considered a Map<String, Object> type, and we'll call it entityRepresentation. Let's print out entityRepresentation.toString() so we can see what this method receives.

public void resolveListingReference(Map<String, Object> entityRepresentation) {
    System.out.println(entityRepresentation.toString());
}

The `@DgsEntityFetcher` annotation

In order for our DGS server to know that this is the method that should receive an entity from the router, we need to mark it with a specific annotation: @DgsEntityFetcher.

This annotation has a property called name, which we'll use to define the name of the entity that it resolves: Listing!

Let's apply this annotation to our method.

@DgsEntityFetcher(name = "Listing")
public void resolveListingReference(Map<String, Object> entityRepresentation) {
    System.out.println(entityRepresentation.toString());
}

Now it's extra clear that when the router brings a Listing entity representation to our reviews subgraph, this is the method that we'll use to resolve which listing we need to provide data for!

Go back to your terminal and restart the reviews server. Our rover dev process should still be running on port 4000, so let's return there and try out our query again.

query GetListingAndReviews {
  listing(id: "listing-1") {
    title
    description
    numOfBeds
    amenities {
      name
      category
    }
    overallRating
    reviews {
      id
      text
    }
  }
}

When we run the query, we still won't get any review-related data back; but when we check the terminal of our reviews subgraph, we'll see that our entity representation has arrived and is printed out!

{__typename=Listing, id=listing-1}

Our reviews subgraph is successfully receiving the listing representation from the router. Now, we just need to make sure our method returns a new Listing instance that other datafetcher methods can access and attach data to.

Delete the print statement, and replace it by instantiate a new Listing instance. We can also update our method's return type to Listing.

@DgsEntityFetcher(name = "Listing")
public Listing resolveListingReference(Map<String, Object> entityRepresentation) {
  Listing listing = new Listing();
}

Next, let's pull the "id" value from our entityRepresentation. We can cast it as a String type with the following line.

String id = (String) entityRepresentation.get("id");

Lastly, we'll set this as the id on our listing, and return it.

listing.setId(id);
return listing;

Great! We can consider our reference to the listing "resolved". We're ready to provide those review-relevant fields—reviews and overallRating—with their own datafetcher methods.

Adding `Listing.reviews`

Let's tackle the reviews method first.

Inside the ReviewsDataFetcher class, and let's define a new method called reviews. This is a regular datafetcher, so we'll use the @DgsData annotation, setting our parentType as Listing.

@DgsData(parentType="Listing")
public void reviews() {}

In our GraphQL schema, the Listing.reviews field returns a type of [Review!]!. So let's update our method with the corresponding Java return type of Flux<ReviewDto>.

public Flux<ReviewDto> reviews() {}

This method should use the id from the entity representation to find and return all the relevant reviews in the database. Because this datafetcher resolves a field on an entity type, we know that the previous datafetcher in the chain was the resolveListingReference method we just defined. This means we can access its return value (the Listing instance) using the DgsDataFetchingEnvironment parameter.

Let's bring that into our method as a parameter called dfe.

public Flux<ReviewDto> reviews(DgsDataFetchingEnvironment dfe) {}

We can call the getSource method on dfe to get access to the Listing instance; then call the listing's getId method to access its id value.

public Flux<ReviewDto> reviews(DgsDataFetchingEnvironment dfe) {
  Listing listing = dfe.getSource();
  String id = listing.getId();
}

With our listing id in hand, we can look up all of the reviews in our database that are associated with that listing. We're using our ReviewController, which is already instantiated on our class, to access our in-memory data source. It provides a mapping for reviewsForListing, so let's call that method here and pass in our id.

return this.reviewController.reviewsForListing(id);

The `overallRating` method

Onto the overallRating datafetcher method! Let's set up the initial structure with the @DgsData annotation.

@DgsData(parentType="Listing")
public void overallRating() {
  // TODO
}

Try this one out on your own, and check out the ReviewController class for a helpful method you can use to retrieve this data. (Hint: Check out its return type for a helpful cue on what our method should return!)

When you're ready, compare your code to our finished method below.

@DgsData(parentType="Listing")
public Mono<Float> overallRating(DgsDataFetchingEnvironment dfe) {
    Listing listing = dfe.getSource();
    String id = listing.getId();
    return this.reviewController.averageRatingForListing(id);
}

Note: Because we're coding in the reactive style, we've used the Flux type so far to describe the stream of data that we return. However, since we're returning just a single value—the average of a listing's ratings, returned as a Float—we can use the Mono type instead and pass it the type variable of Float.

Running our dream query

Time to recompile!

Task!

I've restarted my reviews subgraph.

With the rover dev process still running, let's try out our query at http://localhost:4000.

query GetListingAndReviews {
  listing(id: "listing-1") {
    title
    description
    numOfBeds
    amenities {
      name
      category
    }
    overallRating
    reviews {
      id
      text
    }
  }
}

Submit the query, and... we've got data! 🎉 We've associated reviews with a listing and made our dream query come to life!

{
  "data": {
    "listing": {
      "title": "Cave campsite in snowy MoundiiX",
      "description": "Enjoy this amazing cave campsite in snow MoundiiX, where you'll be one with the nature and wildlife in this wintery planet. All space survival amenities are available. We have complementary dehydrated wine upon your arrival. Check in between 34:00 and 72:00. The nearest village is 3AU away, so please plan accordingly. Recommended for extreme outdoor adventurers.",
      "numOfBeds": 2,
      "amenities": [
        {
          "name": "Towel",
          "category": "Accommodation Details"
        },
        {
          "name": "Oxygen",
          "category": "Space Survival"
        },
        {
          "name": "Prepackaged meals",
          "category": "Space Survival"
        },
        {
          "name": "SOS button",
          "category": "Space Survival"
        },
        {
          "name": "Meteor shower shield",
          "category": "Space Survival"
        },
        {
          "name": "First-aid kit",
          "category": "Space Survival"
        },
        {
          "name": "Water recycler",
          "category": "Space Survival"
        },
        {
          "name": "Panic button",
          "category": "Space Survival"
        },
        {
          "name": "Emergency life support systems",
          "category": "Space Survival"
        },
        {
          "name": "Universal translator",
          "category": "Space Survival"
        },
        {
          "name": "Aquatic breathing aid",
          "category": "Space Survival"
        },
        {
          "name": "Acid lake access",
          "category": "Outdoors"
        },
        {
          "name": "Time travel paradoxes",
          "category": "Outdoors"
        },
        {
          "name": "Meteor showers",
          "category": "Outdoors"
        },
        {
          "name": "Wildlife",
          "category": "Outdoors"
        }
      ],
      "overallRating": 3.75,
      "reviews": [
        {
          "id": "1",
          "text": "Wow, what an experience! I've never stayed in a cave before, so I was a little unprepared. Luckily, this listing had all the amenities I needed to feel safe and prepared for anything."
        },
        {
          "id": "2",
          "text": "100% enjoyed the wilderness experience. Do not book if you are not an adventurer and lover of the outdoors."
        },
        {
          "id": "3",
          "text": "I thought this was going to be a cozy cave, but I was sorely disappointed. The mattress was hard, I could feel stones digging into my back. And it was COLD. They need to be more clear about this on the description."
        },
        {
          "id": "6",
          "text": "Description was accurate. It was indeed a cave campsite in the snowy part of the planet. Exactly what I needed."
        }
      ]
    }
  }
}

Reviewing the query plan

Let's check out the query plan to see how this data came together. We'll see that first, the router will fetch data from listings, and then use that data to build out its request to reviews. The last step is flattening the response from both subgraphs into a single instance of a Listing type for the listing field we've queried!

http://localhost:4000

The operation in Sandbox, with the Query Plan Preview opened, showing a linear path to retrieve data from both subgraphs

Still seeing reviews: null? Try restarting your DGS server! The rover dev process running our router on port 4000 will automatically refresh.

Earlier, we stated that one of the requirements for a subgraph that contributes fields to an entity is that it also defines a reference resolver.

We took care of this in the reviews subgraph... but what about listings?

Well, we've gotten away with not writing a reference resolver in the listings subgraph for a simple reason: in its current state, our API never requires the router to send an entity representation to the listings subgraph.

Here's why. Whether we're building a query for featuredListings or a single listing, those entrypoints exist in the listings subgraph; so this subgraph already has an idea of which listing it's resolving data for.

Let's imagine instead that we update our reviews subgraph. Maybe we give our Review type a listing field that points back to the Listing it was written for.

A theoretical implementation of Review

type Review {
  # ... other Review fields
  listing: Listing!
}

Now let's say we use the Query entrypoint provided by our reviews subgraph, allReviews, to build the query below.

A query for all reviews, and their listings

query GetAllReviews {
  allReviews {
    id
    text
    listing {
      title
    }
  }
}

We can walk through exactly how this query should be resolved. Because the Query.allReviews field is provided by the reviews subgraph, that's where we'd start. For each Review type returned by the allReviews field, we'd then dig deeper to resolve its listing field. Here's where our router would need to take the Listing entity representation from the reviews subgraph, and pass it over to the listings subgraph, in order to resolve the title field!

Example Listing entity representation

{
  "__typename": "Listing",
  "id": "listing-3"
}

Now we'd run into trouble. The listings subgraph doesn't have a method that can receive this entity representation. We haven't added one!

Even though the scenario we just ran through doesn't actually apply to our API, we shouldn't rule out the possibility that one day our needs will change. In that case, our listings subgraph will need to know how to resolve a Listing entity representation that comes from somewhere else!

Practice

Where should an entity's reference resolver method be defined?

In every subgraph that is composed into the supergraph.In the router.In each subgraph that contributes fields to the entity.In the first subgraph that defines the entity.

Key takeaways

Any subgraph that contributes fields to an entity needs to define a reference resolver method for that entity. This method is called whenever the router needs to access fields of the entity from within another subgraph.
In DGS, we denote a method as a reference resolver using the @DgsEntityFetcher annotation, passing in the name of the type it resolves. This method receives an entity representation from the router.
An entity representation is an object that the router uses to represent a specific instance of an entity. It includes the entity's type and its key field(s).

Up next

Awesome! Our listings and reviews services are now collaborating on the same Listing entity. Each subgraph contributes its own fields, and the router launched by our local rover dev process packages up the response for us. Emphasis on the word local; to get our changes actually "live" (at least in the tutorial sense of the word), we need to tell GraphOS about them!

In the next lesson, we'll take a look at how we can land these changes safely and confidently using schema checks and launches.

Overview

The reference resolver

Adding resolveListingReference

The @DgsEntityFetcher annotation

Adding Listing.reviews

The overallRating method