Documentation
Resolvers

Resolvers

When using graphql-tools, you define your field resolvers separately from the schema. Since the schema already describes all of the fields, arguments, and result types, the only thing left is a collection of functions that are called to actually execute these fields.

Keep in mind that GraphQL resolvers can return promises. In fact, most resolvers that do real work - for example fetching data from a database or a REST API - will return a promise. If you’re not familiar with promises, here’s a brief overview.

Resolver Map

To respond to queries, a schema needs to have resolvers for all fields. Resolvers are per field functions that are given a parent object, arguments, and the execution context, and are responsible for returning a result for that field. Resolvers cannot be included in the GraphQL schema language, so they must be added separately. The collection of resolvers is called the “resolver map”.

The resolverMap object (IResolvers) should have a map of resolvers for each relevant GraphQL Object Type. The following is an example of a valid resolverMap object:

const authors = [
  { id: '1', name: 'Laurin' },
  { id: '2', name: 'Dotan' },
  { id: '3', name: 'Arda' }
]
 
const posts = [
  { id: '1', authorId: '1', title: 'Foo' },
  { id: '2', authorId: '1', title: 'Bar' },
  { id: '3', authorId: '2', title: 'Brrt' }
]
 
const resolverMap = {
  Query: {
    author(obj, args, context, info) {
      return authors.find(author => author.id === args.id) ?? null
    }
  },
  Author: {
    posts(author) {
      return posts.filter(post => post.authorId === author.id)
    }
  }
}
💡

Note: If you are using mocking, the preserveResolvers argument of addMocksToSchema must be set to true if you don’t want your resolvers to be overwritten by mock resolvers.

Note that you don’t have to put all of your resolvers in one object. Refer to the “merging resolvers” section to learn how to combine multiple resolver maps into one.

Resolver Function Signature

Every resolver in a GraphQL.js schema accepts four positional arguments:

function fieldName(obj, args, context, info) {
  /** resolver implementation */
}

These arguments have the following meanings and conventional names:

  1. obj: The object that contains the result returned from the resolver on the parent field, or, in the case of a top-level Query field, the rootValue passed from the server configuration. This argument enables the nested nature of GraphQL queries.
  2. args: An object with the arguments passed into the field in the query. For example, if the field was called with author(name: "Ada"), the args object would be: { "name": "Ada" }.
  3. context: This is an object shared by all resolvers in a particular query, and is used to contain the per-request state, including authentication information, DataLoader instances, and anything else that should be taken into account when resolving the query. If you’re using express-graphql, read about how to set the context in the setup documentation.
  4. info: This argument should only be used in advanced cases, but it contains information about the execution state of the query, including the field name, the path to the field from the root, and more. It’s only documented in the GraphQL.js source code.

Resolver Result Format

Resolvers in GraphQL can return different kinds of results which are treated differently:

  1. null or undefined - this indicates the object could not be found. If your schema says that field is nullable, then the result will have a null value at that position. If the field is non-null, the result will “bubble up” to the nearest nullable field and that result will be set to null. This is to ensure that the API consumer never gets a null value when they were expecting a result.
  2. An array - this is only valid if the schema indicates that the result of a field should be a list. The sub-selection of the query will run once for every item in this array.
  3. A promise - resolvers often do asynchronous actions like fetching from a database or backend API, so they can return promises. This can be combined with arrays, so a resolver can return:
    • A promise that resolves an array
    • An array of promises
  4. A scalar or object value - a resolver can also return any other kind of value, which doesn’t have any special meaning but is simply passed down into any nested resolvers, as described in the next section.

Resolver obj Argument

The first argument to every resolver, obj, can be a bit confusing at first, but it makes sense when you consider what a GraphQL query looks like:

query {
  getAuthor(id: 5) {
    name
    posts {
      title
      author {
        name # this will be the same as the name above
      }
    }
  }
}

You can think of every GraphQL query as a tree of function calls, as explained in detail in the GraphQL explained blog post. So the obj contains the result of the parent resolver, in this case:

  1. obj in Query.getAuthor will be whatever the server configuration passed for rootValue.
  2. obj in Author.name and Author.posts will be the result from getAuthor, likely an Author object from the backend.
  3. obj in Post.title and Post.author will be one item from the posts result array.
  4. obj in Author.name is the result from the above Post.author call.

It’s just every resolver function being called in a nested way according to the layout of the query.

Default Resolver

You don’t need to specify resolvers for every type in your schema. If you don’t specify a resolver, GraphQL.js falls back to a default one, which does the following:

  1. Returns a property from obj with the relevant field name, or
  2. Calls a function on obj with the relevant field name and passes the query arguments into that function

So, in the example query above, the name and title fields wouldn’t need a resolver if the Post and Author objects retrieved from the backend already had those fields.

Class Method Resolvers

When returning an object or a class instance from a parent resolver, we are implicitly relying on the GraphQL.js default resolver logic described above to execute the child resolvers. As such, class method resolvers have the following interface:

class Type {
  fieldName(args, context, info) {
    /* resolver implementation */
  }
}

Unions and Interfaces

Unions and interfaces are great when you have fields that are in common between two types.

When you have a field in your schema that returns a union or interface type, you will need to specify an extra __resolveType field in your resolver map, which tells the GraphQL executor which type the result is, out of the available options.

For example, if you have a Vehicle interface type with members Airplane and Car:

You could specify the schema like so:

interface Vehicle {
  maxSpeed: Int
}
 
type Airplane implements Vehicle {
  maxSpeed: Int
  wingspan: Int
}
 
type Car implements Vehicle {
  maxSpeed: Int
  licensePlate: String
}
const resolverMap = {
  Vehicle: {
    __resolveType(obj, context, info) {
      if (obj.wingspan) {
        return 'Airplane'
      }
 
      if (obj.licensePlate) {
        return 'Car'
      }
 
      return null
    }
  }
}
💡

Note: Returning the type name as a string from __resolveType is only supported starting with GraphQL.js 0.7.2. In previous versions, you had to get a reference using info.schema.getType('Car').

API

In addition to using a resolver map with makeExecutableSchema, you can use it with any GraphQL.js schema by importing the following function from graphql-tools:

addResolversToSchema({ schema, resolvers, resolverValidationOptions?, inheritResolversFromInterfaces? })

addResolversToSchema takes an options object of IAddResolveFunctionsToSchemaOptions and returns a new schema with resolvers attached to the relevant types.

import { addResolversToSchema } from '@graphql-tools/schema'
 
const resolvers = {
  RootQuery: {
    author(obj, { name }, context) {
      console.log('RootQuery called with context', context, 'to find', name)
      return Author.find({ name })
    }
  }
}
 
const schemaWithResolvers = addResolversToSchema({ schema, resolvers })

The IAddResolveFunctionsToSchemaOptions object consists of several of the underlying properties used to configure makeExecutableSchema and described there.

export interface IAddResolveFunctionsToSchemaOptions {
  schema: GraphQLSchema
  resolvers: IResolvers
  resolverValidationOptions?: IResolverValidationOptions
  inheritResolversFromInterfaces?: boolean
  updateResolversInPlace?: boolean
}

Additionally, the updateResolversInPlace property, when set to true, changes addResolversToSchema behavior to modify the original schema in place without recreating it. By default, a new schema will be returned without modification of the original schema.

addSchemaLevelResolver(schema, rootResolveFunction)

Some operations, such as authentication, need to be done only once per query. Logically, these operations belong in a schema level resolver field resolver, but unfortunately, GraphQL-JS does not let you define one. addSchemaLevelResolver solves this by returning a new schema with the addition of a root resolve function.

💡

You can check Resolvers Composition to compose resolvers with an authentication layer, and some checking operations etc.