Catch the highlights of GraphQLConf 2023! Click for recordings. Or check out our recap blog post.
Docs
Approaches
Schema Extensions

Schema Extensions

Schema extensions add gateway-level type definitions and resolvers into a combined API, which is useful for establishing connections between types that exist in separate subschemas.

When considering these capabilities, be sure to compare them with the newer automated features available through type merging. While type merging frequently eliminates the need for schema extensions, it does not preclude their use.

This approach uses Schema Delegation approach to delegate queries to subschemas.

Schema Delegation

Schema delegation is a way to automatically forward a query (or a part of a query) from a parent schema to another schema (called a subschema) that can execute the query. Delegation is useful when the parent schema shares a significant part of its data model with the subschema. For example:

  • A GraphQL gateway that connects multiple existing endpoints, each with its schema, could be implemented as a parent schema that delegates portions of queries to the relevant subschemas.
  • Any local schema can directly wrap remote schemas and optionally extend them with additional fields. As long as schema delegation is unidirectional, no gateway is necessary. Simple examples are schemas that wrap other autogenerated schemas (e.g. Postgraphile, Hasura, Prisma) to add custom functionality.

Delegation is performed by one function, delegateToSchema, called from within a resolver function of the parent schema. The delegateToSchema function sends the query subtree received by the parent resolver to the subschema that knows how to execute it. Fields for the merged types use the defaultMergedResolver resolver to extract the correct data from the query response.

Motivational Example

Let's consider two schemas, a subschema and a parent schema that reuses parts of a subschema. While the parent schema reuses the definitions of the subschema, we want to keep the implementations separate, so that the subschema can be tested independently, or even used as a remote service.

# Subschema
type Repository {
  id: ID!
  url: String
  issues: [Issue]
  userId: ID!
}
 
type Issue {
  id: ID!
  text: String!
  repository: Repository!
}
 
type Query {
  repositoryById(id: ID!): Repository
  repositoriesByUserId(id: ID!): [Repository]
}
 
# Parent schema
type Repository {
  id: ID!
  url: String
  issues: [Issue]
  userId: ID!
  user: User
}
 
type Issue {
  id: ID!
  text: String!
  repository: Repository!
}
 
type User {
  id: ID!
  username: String
  repositories: [Repository]
}
 
type Query {
  userById(id: ID!): User
}

Suppose we want the parent schema to delegate retrieval of repositories to the subschema, in order to execute queries such as this one:

query {
  userById(id: "1") {
    id
    username
    repositories {
      id
      url
      user {
        username
        id
      }
      issues {
        text
      }
    }
  }
}

The resolver function for the repositories field of the User type would be responsible for the delegation, in this case. While it's possible to call a remote GraphQL endpoint or resolve the data manually, this would require us to transform the query manually or always fetch all possible fields, which could lead to overfetching. Delegation automatically extracts the appropriate query to send to the subschema:

# To the subschema
query ($id: ID!) {
  repositoriesByUserId(id: $id) {
    id
    url
    issues {
      text
    }
  }
}

The delegation also removes the fields that don't exist on the subschema, such as user. This field would be retrieved from the parent schema using normal GraphQL resolvers.

Each field on the Repository and Issue types should use the defaultMergedResolver to properly extract data from the delegated response. Although in the simplest case, the default resolver can be used for the merged types, defaultMergedResolver resolves aliases, converts custom scalars and enums to their internal representations, and maps errors.

Basic Example

Going back to the posts and users service example:

import { addMocksToSchema } from '@graphql-tools/mock'
import { makeExecutableSchema } from '@graphql-tools/schema'
 
let postSchema = makeExecutableSchema({
  typeDefs: /* GraphQL */ `
    type Post {
      id: ID!
      text: String
      userId: ID!
    }
 
    type Query {
      postById(id: ID!): Post
      postsByUserId(userId: ID!): [Post!]!
    }
  `
})
 
let userSchema = makeExecutableSchema({
  typeDefs: /* GraphQL */ `
    type User {
      id: ID!
      email: String
    }
 
    type Query {
      userById(id: ID!): User
    }
  `
})
 
// just mock the schemas for now to make them return dummy data
postSchema = addMocksToSchema({ schema: postSchema })
userSchema = addMocksToSchema({ schema: userSchema })
 
// setup subschema config objects
export const postsSubschema = { schema: postSchema }
export const usersSubschema = { schema: userSchema }

postsSubschema and usersSubschema are subschema config objects described in Configuring Subschemas. Typically local subschemas are already executable (via embedded resolvers) and thus do not need to supply an executor.

We may want to navigate from a particular user to their posts, or from a post to its user. This is possible within our service architecture by connecting an existing key of each object to a corresponding root query:

  • Post.userId -> userById(id) gets a Post's user.
  • User.id -> postsByUserId(userId) gets a User's posts.

To formalize this navigation within our gateway schema, we can extend each type with a new field that will translate its respective key into an actual object association:

import { stitchSchemas } from '@graphql-tools/stitch'
 
export const schema = stitchSchemas({
  subschemas: [postsSubschema, usersSubschema],
  typeDefs: /* GraphQL */ `
    extend type Post {
      user: User!
    }
    extend type User {
      posts: [Post!]!
    }
  `
})

The typeDefs option provides type extensions (using the extend keyword) that add additional fields into the combined gateway schema and therefore may cross-reference types from any subschema.

However, these extensions alone won't do anything until they have corresponding resolvers. A complete example would look like this:

import { delegateToSchema } from '@graphql-tools/delegate'
import { stitchSchemas } from '@graphql-tools/stitch'
 
export const schema = stitchSchemas({
  subschemas: [postsSubschema, usersSubschema],
  typeDefs: /* GraphQL */ `
    extend type Post {
      user: User!
    }
    extend type User {
      posts: [Post!]!
    }
  `,
  resolvers: {
    User: {
      posts: {
        selectionSet: `{ id }`,
        resolve(user, args, context, info) {
          return delegateToSchema({
            schema: postsSubschema,
            operation: 'query',
            fieldName: 'postsByUserId',
            args: { userId: user.id },
            context,
            info
          })
        }
      }
    },
    Post: {
      user: {
        selectionSet: `{ userId }`,
        resolve(post, args, context, info) {
          return delegateToSchema({
            schema: usersSubschema,
            operation: 'query',
            fieldName: 'userById',
            args: { id: post.userId },
            context,
            info
          })
        }
      }
    }
  }
})

When resolving User.posts and Post.user, we delegate each key reference to its corresponding root query. Note that the structure of stitching resolvers has a selectionSet property and a resolve method.

selectionSet

Post: {
  user: {
    selectionSet: `{ userId }`,
    // ... resolve
  }
}

The selectionSet specifies the key field(s) needed from an object to query for its associations. For example, Post.user will require that a Post provide its userId. Rather than relying on incoming queries to manually request this key for the association, the selection set will automatically be included in subschema requests to guarantee that these fields are fetched. Dynamic selection sets are also possible by providing a function that receives a GraphQL FieldNode (the gateway field) and returns a SelectionSetNode.

💡

Note: As of version 7 of GraphQL-Tools, fragment hints are removed in favor of selectionSet hints, read more in the migration guide.

resolve

Post: {
  user: {
    // ... selectionSet
    resolve(post, args, context, info) {
      return delegateToSchema({
        schema: usersSubschema,
        operation: 'query',
        fieldName: 'userById',
        args: { id: post.userId },
        context,
        info
      })
    }
  }
}

Resolvers use the delegateToSchema function to forward parts of queries (or even whole new queries) to any other schema—inside or outside of the stitched schema. When delegating to a stitched subschema, always provide the complete subschema config object as the schema option.

By default, delegateToSchema assumes that the delegated operation will return the same GraphQL type as the resolved field (ex: a User field would delegate to a User query). If this is not the case, then you should manually provide a returnType option citing the expected GraphQL return type, and transform the result accordingly in the resolver.

Batch Delegation (Array Batching)

The drawback of performing individual delegateToSchema calls is that they can be fairly inefficient. Say we request Post.user from an array of ten posts—that would delegate ten individual userById queries while resolving each user! To improve this, we can instead delegate in batches, where many instances of a field resolver are consolidated into one delegation.

To setup batching, the first thing we'll need is a new query in the users' service that allows fetching many users at once:

usersByIds(ids: [ID!]!): [User]!

With this many-users query available, we can now delegate the Post.user field in batches across many records:

import { batchDelegateToSchema } from '@graphql-tools/batch-delegate'
import { stitchSchemas } from '@graphql-tools/stitch'
 
const schema = stitchSchemas({
  subschemas: [postsSubschema, usersSubschema],
  typeDefs: /* GraphQL */ `
    extend type Post {
      user: User!
    }
  `,
  resolvers: {
    Post: {
      user: {
        selectionSet: `{ userId }`,
        resolve(post, _args, context, info) {
          return batchDelegateToSchema({
            schema: usersSubschema,
            operation: 'query',
            fieldName: 'usersByIds',
            key: post.userId,
            argsFromKeys: ids => ({ ids }),
            context,
            info
          })
        }
      }
    }
  }
})

Internally, batchDelegateToSchema wraps a single delegateToSchema call in a DataLoader scoped by context, field, arguments, and query selection. It assumes that the delegated operation will return an array of objects matching the gateway field's named GraphQL type (ex: a User field delegates to a [User] query). If this is not the case, then you should manually provide a returnType option citing the expected GraphQL return type. Since it is a thin wrapper around DataLoader, it also makes the following assumptions on the results:

💡
  • The Array of values must be the same length as the Array of keys.

  • Each index in the Array of values must correspond to the same index in the Array of keys.

If the query you're delegating to don't conform to these expectations, you can provide a custom valuesFromResults function to transform it appropriately.

Batch delegation is generally preferable over plain delegation because it eliminates the redundancy of requesting the same field across an array of parent objects. Even so, delegation costs can add up because there is still one subschema request made per batched field—for remote services, this may create many network requests sent to the same service. Consider enabling an additional layer of batching by enabling batch execution with batch: true in subschema configuration;

const someSubschema = {
  schema: someNonExecutableSchema,
  executor: someExecutor,
  batch: true
}

Passing Gateway Arguments

Exhaustive accessors like User.posts do not scale well (...what happens when a user has tens of thousands of posts?), so the gateway should probably accept scoping arguments and pass them through to the underlying subschemas. Let's add a pageNumber argument to the User.posts schema extension:

extend type User {
  posts(pageNumber: Int = 1): [Post]!
}

This argument only exists in the gateway schema and won't do anything until passed through to subschemas. How we pass this input through depends on which subservice owns the association data...

Via Delegation

First, let's say that the Posts service defines this association. The first thing we'll need is a corresponding argument in the posts query; and while we're at it, let's also support batching:

postPagesByUserIds(userIds: [ID!]!, pageNumber: Int=1): [[Post!]!]!

This postPagesByUserIds query is a very primitive example of pagination, and simply returns an array of posts for each user ID. Now we just need to pass the resolver's page number argument through to batchDelegateToSchema, and manually specify a returnType that matches the pagination format:

import { batchDelegateToSchema } from '@graphql-tools/batch-delegate'
import { GraphQLList } from 'graphql'
 
/// ...
User: {
  posts: {
    selectionSet: `{ id }`,
    resolve(user, args, context, info) {
      return batchDelegateToSchema({
        schema: postsSubschema,
        operation: 'query',
        fieldName: 'postPagesByUserIds',
        key: user.id,
        argsFromKeys: userIds => ({ userIds, pageNumber: args.pageNumber }),
        returnType: new GraphQLList(new GraphQLList(postsSubschema.schema.getType('Post'))),
        context,
        info
      })
    }
  }
}

Via selectionSet

Alternatively, let's say that users and posts have a many-to-many relationship and the users service owns the association data. That might give us a User.postIds field to stitch from:

User.postIds(pageNumber: Int=1): [ID]!

In this configuration, resolver arguments will need to pass through with the initial selectionSet. The forwardArgsToSelectionSet helper handles this:

import { forwardArgsToSelectionSet } from '@graphql-tools/stitch'
import { batchDelegateToSchema } from '@graphql-tools/batch-delegate'
 
//...
User: {
  posts: {
    selectionSet: forwardArgsToSelectionSet('{ postIds }'),
    resolve(user, args, context, info) {
      return batchDelegateToSchema({
        schema: postsSubschema,
        operation: 'query',
        fieldName: 'postsByIds',
        key: user.postIds,
        argsFromKeys: ids => ({ ids }),
        context,
        info
      })
    }
  }
}

By default, forwardArgsToSelectionSet will pass through all arguments from the gateway field to all root fields in the selection set. For complex selections that request multiple fields, you may provide an additional mapping of selection names with their respective arguments:

forwardArgsToSelectionSet('{ id postIds }', { postIds: ['pageNumber'] })

Extending Transformed Schemas

Transformed schemas are nuanced because they involve two versions of the same schema: the original schema, and the transformed gateway schema. When extending a transformed schema, we extend the gateway schema but delegate to the original schema. For example:

import { delegateToSchema } from '@graphql-tools/delegate'
import { addMocksToSchema } from '@graphql-tools/mock'
import { makeExecutableSchema } from '@graphql-tools/schema'
import { stitchSchemas } from '@graphql-tools/stitch'
import { FilterRootFields, RenameTypes } from '@graphql-tools/wrap'
 
const postSchema = makeExecutableSchema({
  typeDefs: /* GraphQL */ `
    type Post {
      id: ID!
      text: String
      userId: ID!
    }
    type Query {
      postById(id: ID!): Post
      postsByUserId(userId: ID!): [Post]!
    }
  `
})
 
const postsSubschema = {
  schema: addMocksToSchema({ schema: postSchema }),
  transforms: [
    // remove the "postsByUserId" root field
    new FilterRootFields((op, field) => field !== 'postsByUserId'),
    // prefix all type names with "Post_"
    new RenameTypes(name => `Post_${name}`)
  ]
}
 
const userSchema = makeExecutableSchema({
  typeDefs: /* GraphQL */ `
    type User {
      id: ID!
      email: String
    }
 
    type Query {
      userById(id: ID!): User
    }
  `
})
 
const usersSubschema = {
  schema: addMocksToSchema({ schema: userSchema })
}
 
const stitchedSchema = stitchSchemas({
  subschemas: [postsSubschema, usersSubschema],
  typeDefs: /* GraphQL */ `
    extend type User {
      posts: [Post_Post!]!
    }
    extend type Post_Post {
      user: User!
    }
  `,
  resolvers: {
    User: {
      posts: {
        selectionSet: `{ id }`,
        resolve(user, args, context, info) {
          return delegateToSchema({
            schema: postsSubschema,
            operation: 'query',
            fieldName: 'postsByUserId',
            args: { userId: user.id },
            context,
            info
          })
        }
      }
    },
    Post_Post: {
      user: {
        selectionSet: `{ userId }`,
        resolve(post, args, context, info) {
          return delegateToSchema({
            schema: usersSubschema,
            operation: 'query',
            fieldName: 'userById',
            args: { id: post.userId },
            context,
            info
          })
        }
      }
    }
  }
})

A few key points to note here:

💡
  • All schema extensions and their resolvers exist in the gateway schema and therefore refer to the transformed type name Post_Post.

  • Delegations refer to the original subschema and therefore may reference fields such as postsByUserId that have been removed from the gateway schema.

API

delegateToSchema

The delegateToSchema method should be called with the following named options:

delegateToSchema(options: {
  schema: GraphQLSchema
  operation: 'query' | 'mutation' | 'subscription'
  fieldName: string
  args?: Record<string, any>
  context: Record<string, any>
  info: GraphQLResolveInfo
  transforms?: Array<Transform>
}): Promise<any>

schema: GraphQLSchema

A subschema to delegate to.

operation: 'query' | 'mutation' | 'subscription'

The operation type to use during the delegation.

fieldName: string

A root field in a subschema from which the query should start.

args: Record<string, any>

Additional arguments to be passed to the field. Arguments passed to the field that is being resolved will be preserved if the subschema expects them, so you don't have to pass existing arguments explicitly, though you could use the additional arguments to override the existing ones. For example:

# Subschema
type Booking {
  id: ID!
}
 
type Query {
  bookingsByUser(userId: ID!, limit: Int): [Booking]
}
 
# Schema
type User {
  id: ID!
  bookings(limit: Int): [Booking]
}
 
type Booking {
  id: ID!
}

If we delegate User.bookings to Query.bookingsByUser, we want to preserve the limit argument and add a userId argument by using the User.id. So the resolver would look like the following:

import { delegateToSchema } from '@graphql-tools/delegate'
 
const resolvers = {
  User: {
    bookings(parent, args, context, info) {
      return delegateToSchema({
        schema: subschema,
        operation: 'query',
        fieldName: 'bookingsByUser',
        args: {
          userId: parent.id
        },
        context,
        info
      })
    }
    // ...
  }
  // ...
}

context: Record<string, any>

GraphQL's context that is going to be passed to the subschema execution or subscription call.

info: GraphQLResolveInfo

GraphQL resolves info of the current resolver. Provides access to the subquery that starts at the current resolver.

transforms: Transform[]

Any additional operation transforms to apply to the query and results. Transforms are specified similarly to the transforms used in conjunction within the subschemas, but only the operational components of transforms will be used by delegateToSchema, i.e. any specified transformRequest and transformResult functions. The following transforms are automatically applied during schema delegation to translate between source and target types and fields: