# Directives

# @aggregate

"""
Returns an aggregate of a column in a given relationship or model.

Requires Laravel 8+.
"""
directive @aggregate(
  """
  The column to aggregate.
  """
  column: String!

  """
  The aggregate function to compute.
  """
  function: AggregateFunction!

  """
  The relationship with the column to aggregate.
  Mutually exclusive with `model` and `builder`.
  """
  relation: String

  """
  The model with the column to aggregate.
  Mutually exclusive with `relation` and `builder`.
  """
  model: String

  """
  Point to a function that provides a Query Builder instance.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  Mutually exclusive with `relation` and `model`.
  """
  builder: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

"""
Options for the `function` argument of `@aggregate`.
"""
enum AggregateFunction {
  """
  Return the average value.
  """
  AVG

  """
  Return the sum.
  """
  SUM

  """
  Return the minimum.
  """
  MIN

  """
  Return the maximum.
  """
  MAX
}

If all you need is counting, use @count.

To retrieve the aggregate of a column on a root field, reference a model:

type Query {
  totalDownloads: Int!
    @aggregate(model: "Song", column: "downloads", function: SUM)
}

To retrieve the aggregate of a column in related models, reference the relation:

type Album {
  rating: Float! @aggregate(relation: "songs", column: "rating", function: AVG)
}

You may combine filters and scopes:

type Query {
  mostListened(genre: String @eq): Int!
    @aggregate(
      model: "Song"
      column: "listen_count"
      function: MAX
      scope: ["published"]
    )
}

# @all

"""
Fetch all Eloquent models and return the collection as the result.
"""
directive @all(
  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  Mutually exclusive with `builder`.
  """
  model: String

  """
  Point to a function that provides a Query Builder instance.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  Mutually exclusive with `model`.
  """
  builder: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

This assumes your model has the same name as the type you are returning and is defined in the default model namespace App. You can change this configuration.

type Query {
  users: [User!]! @all
}

If you need to use a different model for a single field, you can pass a class name as the model argument.

type Query {
  posts: [Post!]! @all(model: "App\\Blog\\BlogEntry")
}

# @auth

"""
Return the currently authenticated user as the result of a query.
"""
directive @auth(
  """
  Specify which guard to use, e.g. "api".
  When not defined, the default from `lighthouse.php` is used.
  """
  guard: String
) on FIELD_DEFINITION

type Query {
  me: User @auth
}

If you need to use a guard besides the default to resolve the authenticated user, you can pass the guard name as the guard argument

type Query {
  me: User @auth(guard: "api")
}

# @belongsTo

"""
Resolves a field through the Eloquent `BelongsTo` relationship.
"""
directive @belongsTo(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

It assumes both the field and the relationship method to have the same name.

type Post {
  author: User @belongsTo
}

namespace App\Models;

use Illuminate\Database\Eloquent\Model;
use Illuminate\Database\Eloquent\Relations\BelongsTo;

class Post extends Model
{
    public function author(): BelongsTo
    {
        return $this->belongsTo(User::class);
    }
}

The directive accepts an optional relation argument if your relationship method has a different name than the field.

type Post {
  user: User @belongsTo(relation: "author")
}

# @belongsToMany

"""
Resolves a field through the Eloquent `BelongsToMany` relationship.
"""
directive @belongsToMany(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  Allows to resolve the relation as a paginated list.
  """
  type: BelongsToManyType

  """
  Allow clients to query paginated lists without specifying the amount of items.
  Overrules the `pagination.default_count` setting from `lighthouse.php`.
  Setting this to `null` means clients have to explicitly ask for the count.
  """
  defaultCount: Int

  """
  Limit the maximum amount of items that clients can request from paginated lists.
  Overrules the `pagination.max_count` setting from `lighthouse.php`.
  Setting this to `null` means the count is unrestricted.
  """
  maxCount: Int

  """
  Specify a custom type that implements the Edge interface
  to extend edge object.
  Only applies when using Relay style "connection" pagination.
  """
  edgeType: String
) on FIELD_DEFINITION

"""
Options for the `type` argument of `@belongsToMany`.
"""
enum BelongsToManyType {
  """
  Offset-based pagination, similar to the Laravel default.
  """
  PAGINATOR

  """
  Offset-based pagination like the Laravel "Simple Pagination", which does not count the total number of records.
  """
  SIMPLE

  """
  Cursor-based pagination, compatible with the Relay specification.
  """
  CONNECTION
}

# Basic Usage

The field and the relationship method are assumed to have the same name.

type User {
  roles: [Role!]! @belongsToMany
}

use Illuminate\Database\Eloquent\Model;
use Illuminate\Database\Eloquent\Relations\BelongsToMany;

class User extends Model
{
    public function roles(): BelongsToMany
    {
        return $this->belongsToMany(Role::class);
    }
}

# Rename Relation

The directive accepts an optional relation argument if your relationship method has a different name than the field.

type User {
  jobs: [Role!]! @belongsToMany(relation: "roles")
}

# Retrieving Intermediate Table Columns

You may want to allow accessing data that describes the relation between the models and is stored in the intermediate table - see retrieving intermediate table columns in Laravel (opens new window).

Just like in Laravel, you can access the pivot attribute on the models (or its alias). Even though this attribute is always present when querying the model through the relation, it may not be present when reaching the node through another path in the schema, so it is recommended to define the field as nullable (no !).

The following example assumes the intermediate table between User and Role defines a column meta.

use Illuminate\Database\Eloquent\Model;
use Illuminate\Database\Eloquent\Relations\BelongsToMany;

class User extends Model
{
    public function roles(): BelongsToMany
    {
        return $this->belongsToMany(Role::class)
            ->withPivot('meta');
    }
}

class Role extends Model
{
    public function users(): BelongsToMany
    {
        return $this->belongsToMany(User::class)
            ->withPivot('meta');
    }
}

type User {
  id: ID!
  roles: [Role!]! @belongsToMany
  pivot: RoleUserPivot
}

type Role {
  id: ID!
  users: [Users!]! @belongsToMany
  pivot: RoleUserPivot
}

type RoleUserPivot {
  meta: String
}

When using the type argument with pagination style CONNECTION, you may create your own edge type (opens new window) that contains the attributes of the intermediate table.

The custom edge type must contain at least the following two fields:

cursor: String!
node: <RelatedModel>! (in this case node: Role!)

It is expected to be named <RelatedModel>Edge (in this case RoleEdge). Assuming the intermediate table defines a column meta, the definition could look like this:

type User {
  roles: [Role!]! @belongsToMany(type: CONNECTION)
}

type RoleEdge {
  node: Role!
  cursor: String!
  meta: String
}

# @broadcast

"""
Broadcast the results of a mutation to subscribed clients.
"""
directive @broadcast(
  """
  Name of the subscription that should be retriggered as a result of this operation.
  """
  subscription: String!

  """
  Specify whether or not the job should be queued.
  This defaults to the global config option `lighthouse.subscriptions.queue_broadcasts`.
  """
  shouldQueue: Boolean
) repeatable on FIELD_DEFINITION

# @builder

"""
Manipulate the query builder with a method.
"""
directive @builder(
  """
  Reference a method that is passed the query builder.
  Consists of two parts: a class name and a method name, separated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  """
  method: String!

  """
  Pass a value to the method as the second argument after the query builder.
  Only used when the directive is added on a field.
  """
  value: BuilderValue
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION | FIELD_DEFINITION

"""
Any constant literal value: https://graphql.github.io/graphql-spec/draft/#sec-Input-Values
"""
scalar BuilderValue

You must point to a method which will receive the builder instance and can apply additional constraints to the query.

When used on an argument, the value is supplied as the second parameter to the method. When used on a field, the value argument inside the directive is applied as the second parameter to the method.

type Query {
    users(
        minimumHighscore: Int @builder(method: "App\MyClass@minimumHighscore")
    ): [User!]! @all
    highrankedUsers: [User!]! @all @builder(method: "App\MyClass@minimumHighscore", value: 1000)
}

use Illuminate\Database\Eloquent\Builder;

class MyClass
{
    public function minimumHighscore(Builder $builder, int $minimumHighscore): Builder
    {
        return $builder->whereHas('game', static function (Builder $builder) use ($minimumHighscore): void {
            $builder->where('score', '>', $minimumHighscore);
        });
    }
}

# @cache

"""
Cache the result of a resolver.
"""
directive @cache(
  """
  Set the duration it takes for the cache to expire in seconds.
  If not given, the result will be stored forever.
  """
  maxAge: Int

  """
  Limit access to cached data to the currently authenticated user.
  When the field is accessible by guest users, this will not have
  any effect, they will access a shared cache.
  """
  private: Boolean = false
) on FIELD_DEFINITION

You can find usage examples of this directive in the caching docs.

# @cacheControl

"""
Influences the HTTP `Cache-Control` headers of the response.
"""
directive @cacheControl(
  """
  The maximum amount of time the field's cached value is valid, in seconds.
  0 means the field is not cacheable.
  Mutually exclusive with `inheritMaxAge = true`.
  """
  maxAge: Int! = 0

  """
  Is the value specific to a single user?
  """
  scope: CacheControlScope! = PUBLIC

  """
  Should the field inherit the `maxAge` of its parent field instead of using the default `maxAge`?
  Mutually exclusive with `maxAge`.
  """
  inheritMaxAge: Boolean! = false
) on FIELD_DEFINITION | OBJECT | INTERFACE | UNION

"""
Options for the `scope` argument of `@cacheControl`.
"""
enum CacheControlScope {
  """
  The value is the same for each user.
  """
  PUBLIC

  """
  The value is specific to a single user.
  """
  PRIVATE
}

Find usage examples of this directive in the caching docs.

# @cacheKey

"""
Specify the field to use as a key when creating a cache.
"""
directive @cacheKey on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION | FIELD_DEFINITION

You can find usage examples of this directive in the caching docs.

# @can

"""
Check a Laravel Policy to ensure the current user is authorized to access a field.

When `injectArgs` and `args` are used together, the client given
arguments will be passed before the static args.
"""
directive @can(
  """
  The ability to check permissions for.
  """
  ability: String!

  """
  Check the policy against the model instances returned by the field resolver.
  Only use this if the field does not mutate data, it is run before checking.

  Mutually exclusive with `query` and `find`.
  """
  resolved: Boolean! = false

  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String

  """
  Pass along the client given input data as arguments to `Gate::check`.
  """
  injectArgs: Boolean! = false

  """
  Statically defined arguments that are passed to `Gate::check`.

  You may pass arbitrary GraphQL literals,
  e.g.: [1, 2, 3] or { foo: "bar" }
  """
  args: CanArgs

  """
  Query for specific model instances to check the policy against, using arguments
  with directives that add constraints to the query builder, such as `@eq`.

  Mutually exclusive with `resolved` and `find`.
  """
  query: Boolean! = false

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  If your policy checks against specific model instances, specify
  the name of the field argument that contains its primary key(s).

  You may pass the string in dot notation to use nested inputs.

  Mutually exclusive with `resolved` and `query`.
  """
  find: String
) repeatable on FIELD_DEFINITION

"""
Any constant literal value: https://graphql.github.io/graphql-spec/draft/#sec-Input-Values
"""
scalar CanArgs

The name of the returned Type Post is used as the Model class, however you may overwrite this by passing the model argument:

type Mutation {
  createBlogPost(input: PostInput!): BlogPost
    @can(ability: "create", model: "App\\Post")
}

Check the policy against the resolved model instances with the resolved argument:

type Query {
  fetchUserByEmail(email: String! @eq): User
    @can(ability: "view", resolved: true)
    @find
}

You can find usage examples of this directive in the authorization docs.

# @clearCache

"""
Clear a resolver cache by tags.
"""
directive @clearCache(
  """
  Name of the parent type of the field to clear.
  """
  type: String!

  """
  Source of the parent ID to clear.
  """
  idSource: ClearCacheIdSource

  """
  Name of the field to clear.
  """
  field: String
) repeatable on FIELD_DEFINITION

"""
Options for the `id` argument on `@clearCache`.

Exactly one of the fields must be given.
"""
input ClearCacheIdSource {
  """
  Path of an argument the client passes to the field `@clearCache` is applied to.
  """
  argument: String

  """
  Path of a field in the result returned from the field `@clearCache` is applied to.
  """
  field: String
}

You can find usage examples of this directive in the caching docs.

# @complexity

"""
Customize the calculation of a fields complexity score before execution.
"""
directive @complexity(
  """
  Reference a function to customize the complexity score calculation.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  """
  resolver: String
) on FIELD_DEFINITION

type Query {
  posts: [Post!]! @complexity
}

You can provide your own function to calculate complexity.

type Query {
  posts: [Post!]!
    @complexity(resolver: "App\\Security\\ComplexityAnalyzer@userPosts")
}

A custom complexity function may look like the following, refer to the complexity function signature.

namespace App\Security;

class ComplexityAnalyzer {

    public function userPosts(int $childrenComplexity, array $args): int
    {
        $postComplexity = $args['includeFullText'])
            ? 3
            : 2;

        return $childrenComplexity * $postComplexity;
    }

# @convertEmptyStringsToNull

"""
Replaces `""` with `null`.
"""
directive @convertEmptyStringsToNull on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION | FIELD_DEFINITION

Use this when there is no meaningful distinction between an empty string and null.

type Mutation {
  createPost(title: String @convertEmptyStringsToNull): Post!
}

Usage on a field applies the conversion recursively to all inputs.

type Mutation {
  createPost(input: CreatePostInput!): Post! @convertEmptyStringsToNull
}

Non-nullable arguments will not be converted when this directive is used on a field, but will be converted when it is used directly on the argument.

type Mutation {
  createPost(
    willBeConvertedBecauseExplicitlyMarked: String! @convertEmptyStringsToNull
    willNotBeConvertedToMaintainInvariants: String!
  ): Post! @convertEmptyStringsToNull
}

If you want this for all your fields, consider adding this directive to your global field middleware in lighthouse.php:

    'field_middleware' => [
        \Nuwave\Lighthouse\Schema\Directives\ConvertEmptyStringsToNullDirective::class,
        ...
    ],

# @count

"""
Returns the count of a given relationship or model.
"""
directive @count(
  """
  The relationship to count.
  Mutually exclusive with `model`.
  """
  relation: String

  """
  The model to count.
  Mutually exclusive with `relation`.
  """
  model: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  Count only rows where the given columns are non-null.
  `*` counts every row.
  """
  columns: [String!]! = ["*"]

  """
  Should exclude duplicated rows?
  """
  distinct: Boolean! = false
) on FIELD_DEFINITION

Specify the name of the model to count when using this directive on a root query:

type Query {
  categoryCount: Int! @count(model: "Category")
}

You can also count relations:

type User {
  id: ID!
  likeCount: Int! @count(relation: "likes")
}

# @create

"""
Create a new Eloquent model with the given arguments.
"""
directive @create(
  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String

  """
  Specify the name of the relation on the parent model.
  This is only needed when using this directive as a nested arg
  resolver and if the name of the relation is not the arg name.
  """
  relation: String
) on FIELD_DEFINITION | ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

Use it on a root mutation field that returns an instance of the Model.

type Mutation {
  createPost(title: String!): Post @create
}

If you are using a single input object as an argument, you must tell Lighthouse to spread out the nested values before applying it to the resolver.

type Mutation {
  createPost(input: CreatePostInput! @spread): Post @create
}

input CreatePostInput {
  title: String!
}

If the name of the Eloquent model does not match the return type of the field, or is located in a non-default namespace, set it with the model argument.

type Mutation {
  createPost(title: String!): Post @create(model: "Foo\\Bar\\MyPost")
}

This directive can also be used as a nested arg resolver.

# @delete

"""
Delete one or more models by their ID.
The field must have a single non-null argument that may be a list.
"""
directive @delete(
  """
  DEPRECATED use @globalId, will be removed in v6

  Set to `true` to use global ids for finding the model.
  If set to `false`, regular non-global ids are used.
  """
  globalId: Boolean = false

  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String

  """
  Specify the name of the relation on the parent model.
  This is only needed when using this directive as a nested arg
  resolver and if the name of the relation is not the arg name.
  """
  relation: String
) on FIELD_DEFINITION | ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

Use it on a root mutation field that returns an instance of the Model.

type Mutation {
  deletePost(id: ID!): Post @delete
}

In the upcoming v6, the @delete, @forceDelete and @restore directives no longer offer the globalId argument. Use @globalId on the argument instead.

type Mutation {
-   deleteUser(id: ID!): User! @delete(globalId: true)
+   deleteUser(id: ID! @globalId): User! @delete
}

You can also delete multiple models at once. Define a field that takes a list of IDs and returns a collection of the deleted models.

In contrast to Laravel mass updates, this does trigger model events.

type Mutation {
  deletePosts(id: [ID!]!): [Post!]! @delete
}

If the name of the Eloquent model does not match the return type of the field, or is located in a non-default namespace, set it with the model argument.

type Mutation {
  deletePost(id: ID!): Post @delete(model: "Bar\\Baz\\MyPost")
}

This directive can also be used as a nested arg resolver.

type Mutation {
  updateUser(id: Int, deleteTasks: [Int!]! @delete(relation: "tasks")): User
    @update
}

If the model relates to a single other model through a HasOne, MorphOne, BelongsTo or MorphTo relationship, you can just pass a Boolean instead of an ID, as there is only one possible model that can be deleted.

type Mutation {
  updateTask(id: Int, deleteUser: Boolean @delete(relation: "user")): Task
    @update
}

# @deprecated

"""
Marks an element of a GraphQL schema as no longer supported.
"""
directive @deprecated(
  """
  Explains why this element was deprecated, usually also including a
  suggestion for how to access supported similar data.
  Formatted in [Markdown](https://commonmark.org).
  """
  reason: String = "No longer supported"
) on FIELD_DEFINITION | ENUM_VALUE

You can mark fields as deprecated by adding the @deprecated directive. It is recommended to provide a reason for the deprecation, as well as a suggestion on how to move forward.

type Query {
  allUsers: [User!]! @deprecated(reason: "Use `users`")
  users: [User!]!
}

Deprecated elements are not included in introspection queries by default, but they can still be queried by clients.

# @drop

"""
Ignore the user given value, don't pass it to the resolver.
"""
directive @drop on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

This is useful when you want to deprecate a field, but avoid breaking changes for clients that still pass the value.

type User {
  email: String!
  foo: String @deprecated
}

type Mutation {
  createUser(email: String!, foo: String @drop): User @create
}

# @field

"""
Assign a resolver function to a field.
"""
directive @field(
  """
  A reference to the resolver function to be used.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  """
  resolver: String!

  """
  Supply additional data to the resolver.
  """
  args: [String!]
) on FIELD_DEFINITION

Pass a class and a method to the resolver argument and separate them with an @ symbol. If you pass only a class name, the method name defaults to __invoke.

type Mutation {
  createPost(title: String!): Post
    @field(resolver: "App\\GraphQL\\Mutations\\PostMutator@create")
}

If your field is defined on the root types Query or Mutation, you can take advantage of the default namespaces that are defined in the configuration. The following will look for a class in App\GraphQL\Queries by default.

type Query {
  usersTotal: Int @field(resolver: "Statistics@usersTotal")
}

Be aware that resolvers are not limited to root fields. A resolver can be used for basic tasks such as transforming the value of scalar fields, e.g. reformat a date.

type User {
  created_at: String!
    @field(resolver: "App\\GraphQL\\Types\\UserType@created_at")
}

# @find

"""
Find a model based on the arguments provided.
"""
directive @find(
  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

type Query {
  userById(id: ID! @eq): User @find
}

This throws when more than one result is returned. Use @first if the query constraints do not ensure uniqueness.

If your model does not sit in the default namespace, you can overwrite it.

type Query {
  userById(id: ID! @eq): User @find(model: "App\\Authentication\\User")
}

# @first

"""
Get the first query result from a collection of Eloquent models.
"""
directive @first(
  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

Other than @find, this will not throw an error if more than one item is in the collection.

type Query {
  userByFirstName(first_name: String! @eq): User @first
}

If your model does not sit in the default namespace, you can overwrite it.

type Query {
  userByFirstName(first_name: String! @eq): User
    @first(model: "App\\Authentication\\User")
}

# @forceDelete

"""
Permanently remove one or more soft deleted models by their ID.
The field must have a single non-null argument that may be a list.
"""
directive @forceDelete(
  """
  DEPRECATED use @globalId, will be removed in v6

  Set to `true` to use global ids for finding the model.
  If set to `false`, regular non-global ids are used.
  """
  globalId: Boolean = false

  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String
) on FIELD_DEFINITION

Use it on a root mutation field that returns an instance of the Model.

type Mutation {
  forceDeletePost(id: ID!): Post @forceDelete
}

Works very similar to the @delete directive.

# @enum

"""
Assign an internal value to an enum key.
When dealing with the Enum type in your code,
you will receive the defined value instead of the string key.
"""
directive @enum(
  """
  The internal value of the enum key.
  """
  value: EnumValue
) on ENUM_VALUE

"""
Any constant literal value: https://graphql.github.io/graphql-spec/draft/#sec-Input-Values
"""
scalar EnumValue

enum Role {
  ADMIN @enum(value: 1)
  EMPLOYEE @enum(value: 2)
}

You do not need this directive if the internal value of each enum key is an identical string. Read more about enum types

# @eq

"""
Add an equal conditional to a database query.
"""
directive @eq(
  """
  Specify the database column to compare.
  Required if the directive is:
  - used on an argument and the database column has a different name
  - used on a field
  """
  key: String

  """
  Provide a value to compare against.
  Only required when this directive is used on a field.
  """
  value: EqValue
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION | FIELD_DEFINITION

"""
Any constant literal value: https://graphql.github.io/graphql-spec/draft/#sec-Input-Values
"""
scalar EqValue

type User {
  posts(category: String @eq): [Post!]! @hasMany
}

If the name of the argument does not match the database column, pass the actual column name as the key.

type User {
  posts(category: String @eq(key: "cat")): [Post!]! @hasMany
}

You can also use this on a field to define a default filter:

type User {
  sportPosts: [Post!]! @hasMany @eq(key: "category", value: "sport")
}

# @event

"""
Dispatch an event after the resolution of a field.

The event constructor will be called with a single argument:
the resolved value of the field.
"""
directive @event(
  """
  Specify the fully qualified class name (FQCN) of the event to dispatch.
  """
  dispatch: String!
) repeatable on FIELD_DEFINITION

For example, you might want to have an event when new orders are placed in a shop:

type Mutation {
  placeOrder(items: [CartItems!]!): Order!
    @event(dispatch: "App\\Events\\PlacedOrder")
}

The event class must accept an Order in the constructor:

class PlacedOrder
{
    public function __construct(Order $order) { ... }
}

# @globalId

"""
Converts between IDs/types and global IDs.

When used upon a field, it encodes;
when used upon an argument, it decodes.
"""
directive @globalId(
  """
  Decoding a global id produces a tuple of `$type` and `$id`.
  This setting controls which of those is passed along.
  """
  decode: GlobalIdDecode = ARRAY
) on FIELD_DEFINITION | INPUT_FIELD_DEFINITION | ARGUMENT_DEFINITION

"""
Options for the `decode` argument of `@globalId`.
"""
enum GlobalIdDecode {
  """
  Return an array of `[$type, $id]`.
  """
  ARRAY

  """
  Return just `$type`.
  """
  TYPE

  """
  Return just `$id`.
  """
  ID
}

Instead of the original ID, the id field will now return a base64-encoded String that globally identifies the User and can be used for querying the node endpoint.

type User {
  id: ID! @globalId
  name: String
}

The field resolver will receive the decoded version of the passed id, split into type and ID.

type Mutation {
  deleteNode(id: ID @globalId): Node
}

You may rebind the \Nuwave\Lighthouse\Support\Contracts\GlobalId interface to add your own mechanism of encoding/decoding global ids.

# @guard

"""
Run authentication through one or more guards.

This is run per field and may allow unauthenticated
users to still receive partial results.

Used upon an object, it applies to all fields within.
"""
directive @guard(
  """
  Specify which guards to use, e.g. ["web"].
  When not defined, the default from `lighthouse.php` is used.
  """
  with: [String!]
) repeatable on FIELD_DEFINITION | OBJECT

Note that @guard does not log in users. To ensure the user is logged in, add the AttemptAuthenticate middleware to your lighthouse.php middleware config.

'middleware' => [
    ...

    // Logs in a user if they are authenticated. In contrast to Laravel's 'auth'
    // middleware, this delegates auth and permission checks to the field level.
    \Nuwave\Lighthouse\Support\Http\Middleware\AttemptAuthentication::class,
],

A useful pattern is to group fields in an extend type to apply @guard on all of them at once.

extend type Query @guard { ... }

The @guard directive will be prepended to other directives defined on the fields and thus executes before them.

extend type Query {
  user: User!
    @guard
    @can(ability: "adminOnly")
  ...
}

# @hash

"""
Use Laravel hashing to transform an argument value.

Useful for hashing passwords before inserting them into the database.
This uses the default hashing driver defined in `config/hashing.php`.
"""
directive @hash on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

The most common use case for this is when dealing with passwords:

type Mutation {
  createUser(name: String!, password: String! @hash): User!
}

# @hasMany

"""
Corresponds to [the Eloquent relationship HasMany](https://laravel.com/docs/eloquent-relationships#one-to-many).
"""
directive @hasMany(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  Allows to resolve the relation as a paginated list.
  Allowed values: `paginator`, `connection`.
  """
  type: HasManyType

  """
  Allow clients to query paginated lists without specifying the amount of items.
  Overrules the `pagination.default_count` setting from `lighthouse.php`.
  Setting this to `null` means clients have to explicitly ask for the count.
  """
  defaultCount: Int

  """
  Limit the maximum amount of items that clients can request from paginated lists.
  Overrules the `pagination.max_count` setting from `lighthouse.php`.
  Setting this to `null` means the count is unrestricted.
  """
  maxCount: Int

  """
  Specify a custom type that implements the Edge interface
  to extend edge object.
  Only applies when using Relay style "connection" pagination.
  """
  edgeType: String
) on FIELD_DEFINITION

"""
Options for the `type` argument of `@hasMany`.
"""
enum HasManyType {
  """
  Offset-based pagination, similar to the Laravel default.
  """
  PAGINATOR

  """
  Offset-based pagination like the Laravel "Simple Pagination", which does not count the total number of records.
  """
  SIMPLE

  """
  Cursor-based pagination, compatible with the Relay specification.
  """
  CONNECTION
}

type User {
  posts: [Post!]! @hasMany
}

You can return the related models paginated by setting the type.

type User {
  postsPaginated: [Post!]! @hasMany(type: PAGINATOR)
  postsSimplePaginated: [Post!]! @hasMany(type: SIMPLE)
  postsRelayConnection: [Post!]! @hasMany(type: CONNECTION)
}

If the name of the relationship on the Eloquent model is different than the field name, you can override it by setting relation.

type User {
  posts: [Post!]! @hasMany(relation: "articles")
}

# @hasManyThrough

"""
Corresponds to [the Eloquent relationship HasManyThrough](https://laravel.com/docs/eloquent-relationships#has-many-through).
"""
directive @hasManyThrough(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  Allows to resolve the relation as a paginated list.
  Allowed values: `paginator`, `connection`.
  """
  type: HasManyThroughType

  """
  Allow clients to query paginated lists without specifying the amount of items.
  Overrules the `pagination.default_count` setting from `lighthouse.php`.
  Setting this to `null` means clients have to explicitly ask for the count.
  """
  defaultCount: Int

  """
  Limit the maximum amount of items that clients can request from paginated lists.
  Overrules the `pagination.max_count` setting from `lighthouse.php`.
  Setting this to `null` means the count is unrestricted.
  """
  maxCount: Int

  """
  Specify a custom type that implements the Edge interface
  to extend edge object.
  Only applies when using Relay style "connection" pagination.
  """
  edgeType: String
) on FIELD_DEFINITION

"""
Options for the `type` argument of `@hasManyThrough`.
"""
enum HasManyThroughType {
  """
  Offset-based pagination, similar to the Laravel default.
  """
  PAGINATOR

  """
  Offset-based pagination like the Laravel "Simple Pagination", which does not count the total number of records.
  """
  SIMPLE

  """
  Cursor-based pagination, compatible with the Relay specification.
  """
  CONNECTION
}

Usage is the same as @hasMany.

# @hasOne

"""
Corresponds to [the Eloquent relationship HasOne](https://laravel.com/docs/eloquent-relationships#one-to-one).
"""
directive @hasOne(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

type User {
  phone: Phone @hasOne
}

If the name of the relationship on the Eloquent model is different than the field name, you can override it by setting relation.

type User {
  phone: Phone @hasOne(relation: "telephone")
}

# @in

"""
Use the client given list value to add an IN conditional to a database query.
"""
directive @in(
  """
  Specify the database column to compare.
  Only required if database column has a different name than the attribute in your schema.
  """
  key: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

type Query {
  posts(includeIds: [Int!] @in(key: "id")): [Post!]! @paginate
}

# @inject

"""
Inject a value from the context object into the arguments.
"""
directive @inject(
  """
  A path to the property of the context that will be injected.
  If the value is nested within the context, you may use dot notation
  to get it, e.g. "user.id".
  """
  context: String!

  """
  The target name of the argument into which the value is injected.
  You can use dot notation to set the value at arbitrary depth
  within the incoming argument.
  """
  name: String!
) repeatable on FIELD_DEFINITION

This is useful to ensure that the authenticated user's id is automatically used for creating new models and cannot be manipulated.

type Mutation {
  createPost(title: String!, content: String!): Post
    @create
    @inject(context: "user.id", name: "user_id")
}

If you are using an Input Object as an argument, you can use dot notation to set a nested argument.

type Mutation {
  createTask(input: CreateTaskInput!): Task
    @create
    @inject(context: "user.id", name: "input.user_id")
}

# @interface

"""
Use a custom resolver to determine the concrete type of an interface.
"""
directive @interface(
  """
  Reference to a custom type-resolver function.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  """
  resolveType: String!
) on INTERFACE

Make sure you read the basics about Interfaces before deciding to use this directive, you probably don't need it.

Set the resolveType argument to a function that returns the implementing Object Type.

interface Commentable
  @interface(resolveType: "App\\GraphQL\\Interfaces\\Commentable@resolveType") {
  id: ID!
}

The function receives the value of the parent field as its single argument and must return an Object Type. You can get the appropriate Object Type from Lighthouse's type registry.

namespace App\GraphQL\Interfaces;

use GraphQL\Type\Definition\Type;
use GraphQL\Type\Definition\ResolveInfo;
use Nuwave\Lighthouse\Schema\TypeRegistry;
use Nuwave\Lighthouse\Support\Contracts\GraphQLContext;

class Commentable
{
    /**
     * @var \Nuwave\Lighthouse\Schema\TypeRegistry
     */
    protected $typeRegistry;

    public function __construct(TypeRegistry $typeRegistry)
    {
        $this->typeRegistry = $typeRegistry;
    }

    /**
     * Decide which GraphQL type a resolved value has.
     *
     * @param  mixed  $rootValue  The value that was resolved by the field. Usually an Eloquent model.
     * @param  \Nuwave\Lighthouse\Support\Contracts\GraphQLContext  $context
     * @param  \GraphQL\Type\Definition\ResolveInfo  $resolveInfo
     * @return \GraphQL\Type\Definition\Type
     */
    public function resolveType($rootValue, GraphQLContext $context, ResolveInfo $resolveInfo): Type
    {
        // Default to getting a type with the same name as the passed in root value
        // TODO implement your own resolver logic - if the default is fine, just delete this class
        return $this->typeRegistry->get(class_basename($rootValue));
    }
}

# @lazyLoad

"""
Perform a [lazy eager load](https://laravel.com/docs/eloquent-relationships#lazy-eager-loading)
on the relations of a list of models.
"""
directive @lazyLoad(
  """
  The names of the relationship methods to load.
  """
  relations: [String!]!
) repeatable on FIELD_DEFINITION

This is often useful when loading relationships with the @hasMany directive.

type Post {
  comments: [Comment!]! @hasMany @lazyLoad(relations: ["replies"])
}

# @like

"""
Add a `LIKE` conditional to a database query.
"""
directive @like(
  """
  Specify the database column to compare.
  Required if the directive is:
  - used on an argument and the database column has a different name
  - used on a field
  """
  key: String

  """
  Fixate the positions of wildcards (`%`, `_`) in the LIKE comparison around the
  placeholder `{}`, e.g. `%{}`, `__{}` or `%{}%`.
  If specified, wildcard characters in the client-given input are escaped.
  If not specified, the client can pass wildcards unescaped.
  """
  template: String

  """
  Provide a value to compare against.
  Only used when the directive is added on a field.
  """
  value: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION | FIELD_DEFINITION

# @limit

"""
Allow clients to specify the maximum number of results to return.
"""
directive @limit on ARGUMENT_DEFINITION | FIELD_DEFINITION

Place this on any argument to a field that returns a list of results.

type Query {
  users(limit: Int @limit): [User!]!
}

Lighthouse will return at most the number of results that the client requested.

{
  users(limit: 5) {
    name
  }
}

{
  "data": {
    "users": [
      { "name": "Never" },
      { "name": "more" },
      { "name": "than" },
      { "name": "5" }
    ]
  }
}

# @method

"""
Resolve a field by calling a method on the parent object.

Use this if the data is not accessible through simple property access or if you
want to pass argument to the method.
"""
directive @method(
  """
  Specify the method of which to fetch the data from.
  Defaults to the name of the field if not given.
  """
  name: String
) on FIELD_DEFINITION

This can be useful on models or other classes that have getters:

type User {
  mySpecialData: String! @method(name: "getMySpecialData")
}

This will call the method User::purchasedItemsCount() with the client given arguments.

type User {
  purchasedItemsCount(year: Int!, includeReturns: Boolean): Int @method
}

Ensure the order of the argument definition matches the parameters of your method.

public function purchasedItemsCount(int $year, ?bool $includeReturns)

Lighthouse will always pass down the same number of arguments and default to null if the client passes nothing.

{
  user(id: 3) {
    purchasedItemsCount(year: 2017)
  }
}

The method will get called like this:

$user->purchasedItemsCount(2017, null)

# @model

"""
Map a model class to an object type.

This can be used when the name of the model differs from the name of the type.
"""
directive @model(
  """
  The class name of the corresponding model.
  """
  class: String!
) on OBJECT

Lighthouse will respect the overwritten model name in its directives.

type Post @model(class: "\\App\\BlogPost") {
  title: String!
}

# @morphMany

"""
Corresponds to [Eloquent's MorphMany-Relationship](https://laravel.com/docs/eloquent-relationships#one-to-many-polymorphic-relations).
"""
directive @morphMany(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  Allows to resolve the relation as a paginated list.
  """
  type: MorphManyType

  """
  Allow clients to query paginated lists without specifying the amount of items.
  Overrules the `pagination.default_count` setting from `lighthouse.php`.
  Setting this to `null` means clients have to explicitly ask for the count.
  """
  defaultCount: Int

  """
  Limit the maximum amount of items that clients can request from paginated lists.
  Overrules the `pagination.max_count` setting from `lighthouse.php`.
  Setting this to `null` means the count is unrestricted.
  """
  maxCount: Int

  """
  Specify a custom type that implements the Edge interface
  to extend edge object.
  Only applies when using Relay style "connection" pagination.
  """
  edgeType: String
) on FIELD_DEFINITION

"""
Options for the `type` argument of `@morphMany`.
"""
enum MorphManyType {
  """
  Offset-based pagination, similar to the Laravel default.
  """
  PAGINATOR

  """
  Offset-based pagination like the Laravel "Simple Pagination", which does not count the total number of records.
  """
  SIMPLE

  """
  Cursor-based pagination, compatible with the Relay specification.
  """
  CONNECTION
}

type Post {
  images: [Image!] @morphMany
}

type Image {
  imageable: Imageable! @morphTo
}

union Imageable = Post | User

# @morphOne

"""
Corresponds to [Eloquent's MorphOne-Relationship](https://laravel.com/docs/eloquent-relationships#one-to-one-polymorphic-relations).
"""
directive @morphOne(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

type Post {
  image: Image! @morphOne
}

type Image {
  imageable: Imageable! @morphTo
}

union Imageable = Post | User

# @morphTo

"""
Corresponds to [Eloquent's MorphTo-Relationship](https://laravel.com/docs/eloquent-relationships#one-to-one-polymorphic-relations).
"""
directive @morphTo(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) on FIELD_DEFINITION

type Image {
  imageable: Imageable! @morphTo
}

union Imageable = Post | User

# @morphToMany

"""
Corresponds to [Eloquent's ManyToMany-Polymorphic-Relationship](https://laravel.com/docs/eloquent-relationships#many-to-many-polymorphic-relations).
"""
directive @morphToMany(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  Allows to resolve the relation as a paginated list.
  """
  type: MorphToManyType

  """
  Allow clients to query paginated lists without specifying the amount of items.
  Overrules the `pagination.default_count` setting from `lighthouse.php`.
  Setting this to `null` means clients have to explicitly ask for the count.
  """
  defaultCount: Int

  """
  Limit the maximum amount of items that clients can request from paginated lists.
  Overrules the `pagination.max_count` setting from `lighthouse.php`.
  Setting this to `null` means the count is unrestricted.
  """
  maxCount: Int

  """
  Specify a custom type that implements the Edge interface
  to extend edge object.
  Only applies when using Relay style "connection" pagination.
  """
  edgeType: String
) on FIELD_DEFINITION

"""
Options for the `type` argument of `@morphToMany`.
"""
enum MorphToManyType {
  """
  Offset-based pagination, similar to the Laravel default.
  """
  PAGINATOR

  """
  Offset-based pagination like the Laravel "Simple Pagination", which does not count the total number of records.
  """
  SIMPLE

  """
  Cursor-based pagination, compatible with the Relay specification.
  """
  CONNECTION
}

# @namespace

"""
Redefine the default namespaces used in other directives.
The arguments are a map from directive names to namespaces.
"""
directive @namespace repeatable on FIELD_DEFINITION | OBJECT

The following example applies the namespace App\Blog to the @field directive used on the posts field.

type Query {
  posts: [Post!]!
    @field(resolver: "Post@resolveAll")
    @namespace(field: "App\\Blog")
}

When used upon an object type or an object type extension, the namespace applies to fields of the type as well. This allows you to specify a common namespace for a group of fields.

extend type Query @namespace(field: "App\\Blog") {
  posts: [Post!]! @field(resolver: "Post@resolveAll")
}

A @namespace directive defined on a field directive wins in case of a conflict.

# @neq

"""
Use the client given value to add an not-equal conditional to a database query.
"""
directive @neq(
  """
  Specify the database column to compare.
  Only required if database column has a different name than the attribute in your schema.
  """
  key: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

type User {
  posts(excludeCategory: String @neq(key: "category")): [Post!]! @hasMany
}

# @nest

"""
A no-op nested arg resolver that delegates all calls
to the ArgResolver directives attached to the children.
"""
directive @nest on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

This may be useful to logically group arg resolvers.

type Mutation {
  createUser(name: String, tasks: UserTasksOperations @nest): User @create
}

input UserTasksOperations {
  newTask: CreateTaskInput @create(relation: "tasks")
}

input CreateTaskInput {
  name: String
}

type Task {
  name: String!
}

type User {
  name: String
  tasks: [Task!]! @hasMany
}

# @node

"""
Register a type for Relay's global object identification.

When used without any arguments, Lighthouse will attempt
to resolve the type through a model with the same name.
"""
directive @node(
  """
  Reference to a function that receives the decoded `id` and returns a result.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.

  Mutually exclusive with `model`.
  """
  resolver: String

  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.

  Mutually exclusive with `resolver`.
  """
  model: String
) on OBJECT

Lighthouse defaults to resolving types through the underlying model, for example by calling User::find($id).

type User @node {
  id: ID! @globalId
}

You can also use a custom resolver function to resolve any kind of data.

type Country @node(resolver: "App\\Countries@byId") {
  name: String!
}

The resolver argument has to specify a function which will be passed the decoded id and resolves to a result.

public function byId($id): array {
    return [
        'DE' => ['name' => 'Germany'],
        'MY' => ['name' => 'Malaysia'],
    ][$id];
}

# @notIn

"""
Use the client given value to add a NOT IN conditional to a database query.
"""
directive @notIn(
  """
  Specify the database column to compare.
  Only required if database column has a different name than the attribute in your schema.
  """
  key: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

type Query {
  posts(excludeIds: [Int!] @notIn(key: "id")): [Post!]! @paginate
}

# @orderBy

"""
Sort a result list by one or more given columns.
"""
directive @orderBy(
  """
  Restrict the allowed column names to a well-defined list.
  This improves introspection capabilities and security.
  Mutually exclusive with `columnsEnum`.
  Only used when the directive is added on an argument.
  """
  columns: [String!]

  """
  Use an existing enumeration type to restrict the allowed columns to a predefined list.
  This allows you to re-use the same enum for multiple fields.
  Mutually exclusive with `columns`.
  Only used when the directive is added on an argument.
  """
  columnsEnum: String

  """
  Allow clients to sort by aggregates on relations.
  Only used when the directive is added on an argument.
  """
  relations: [OrderByRelation!]

  """
  The database column for which the order by clause will be applied on.
  Only used when the directive is added on a field.
  """
  column: String

  """
  The direction of the order by clause.
  Only used when the directive is added on a field.
  """
  direction: OrderByDirection = ASC
) on ARGUMENT_DEFINITION | FIELD_DEFINITION

"""
Options for the `direction` argument on `@orderBy`.
"""
enum OrderByDirection {
  """
  Sort in ascending order.
  """
  ASC

  """
  Sort in descending order.
  """
  DESC
}

"""
Options for the `relations` argument on `@orderBy`.
"""
input OrderByRelation {
  """
  Name of the relation.
  """
  relation: String!

  """
  Restrict the allowed column names to a well-defined list.
  This improves introspection capabilities and security.
  Mutually exclusive with `columnsEnum`.
  """
  columns: [String!]

  """
  Use an existing enumeration type to restrict the allowed columns to a predefined list.
  This allows you to re-use the same enum for multiple fields.
  Mutually exclusive with `columns`.
  """
  columnsEnum: String
}

See ordering.

# @paginate

"""
Query multiple entries as a paginated list.
"""
directive @paginate(
  """
  Which pagination style should be used.
  """
  type: PaginateType = PAGINATOR

  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  Mutually exclusive with `builder` and `resolver`.
  """
  model: String

  """
  Point to a function that provides a Query Builder instance.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  Mutually exclusive with `model` and `resolver`.
  """
  builder: String

  """
  Reference a function that resolves the field by directly returning data in a Paginator instance.
  Mutually exclusive with `builder` and `model`.
  Not compatible with `scopes` and builder arguments such as `@eq`.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  """
  resolver: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]

  """
  Allow clients to query paginated lists without specifying the amount of items.
  Overrules the `pagination.default_count` setting from `lighthouse.php`.
  Setting this to `null` means clients have to explicitly ask for the count.
  """
  defaultCount: Int

  """
  Limit the maximum amount of items that clients can request from paginated lists.
  Overrules the `pagination.max_count` setting from `lighthouse.php`.
  Setting this to `null` means the count is unrestricted.
  """
  maxCount: Int
) on FIELD_DEFINITION

"""
Options for the `type` argument of `@paginate`.
"""
enum PaginateType {
  """
  Offset-based pagination, similar to the Laravel default.
  """
  PAGINATOR

  """
  Offset-based pagination like the Laravel "Simple Pagination", which does not count the total number of records.
  """
  SIMPLE

  """
  Cursor-based pagination, compatible with the Relay specification.
  """
  CONNECTION
}

# Basic usage

This directive is meant to be used on root query fields:

type Query {
  posts: [Post!]! @paginate
}

The schema definition is automatically transformed to this:

type Query {
  posts(
    "Limits number of fetched items."
    first: Int!

    "The offset from which items are returned."
    page: Int
  ): PostPaginator
}

"A paginated list of Post items."
type PostPaginator {
  "A list of Post items."
  data: [Post!]!

  "Pagination information about the list of items."
  paginatorInfo: PaginatorInfo!
}

"Information about pagination using a fully featured paginator."
type PaginatorInfo {
  "Number of items in the current page."
  count: Int!

  "Index of the current page."
  currentPage: Int!

  "Index of the first item in the current page."
  firstItem: Int

  "Are there more pages after this one?"
  hasMorePages: Boolean!

  "Index of the last item in the current page."
  lastItem: Int

  "Index of the last available page."
  lastPage: Int!

  "Number of items per page."
  perPage: Int!

  "Number of total available items."
  total: Int!
}

It can be queried like this:

{
  posts(first: 10) {
    data {
      id
      title
    }
    paginatorInfo {
      currentPage
      lastPage
    }
  }
}

# Pagination type

The type of pagination defaults to PAGINATOR, but may also be set to SIMPLE (see Simple Pagination) or a Relay compliant CONNECTION.

type Query {
  posts: [Post!]! @paginate(type: CONNECTION)
}

The final schema will be transformed to this:

type Query {
  posts(
    "Limits number of fetched items."
    first: Int!

    "A cursor after which elements are returned."
    after: String
  ): PostConnection
}

"A paginated list of Post edges."
type PostConnection {
  "Pagination information about the list of edges."
  pageInfo: PageInfo!

  "A list of Post edges."
  edges: [PostEdge]
}

"An edge that contains a node of type Post and a cursor."
type PostEdge {
  "The Post node."
  node: Post

  "A unique cursor that can be used for pagination."
  cursor: String!
}

# Simple Pagination

In contrast to other pagination types, SIMPLE pagination only fires a single database query on every request. This improves performance, but means that the response does not hold information about the total number of items.

If you wish to use the simplePaginate method, set the type to SIMPLE.

type Query {
  posts: [Post!]! @paginate(type: SIMPLE)
}

The schema definition is automatically transformed to this:

type Query {
  posts(
    "Limits number of fetched items."
    first: Int!

    "The offset from which items are returned."
    page: Int
  ): PostSimplePaginator
}

"A paginated list of Post items."
type PostSimplePaginator {
  "A list of Post items."
  data: [Post!]!

  "Pagination information about the list of items."
  paginatorInfo: SimplePaginatorInfo!
}

"Information about pagination using a simple paginator."
type SimplePaginatorInfo {
  "Number of items in the current page."
  count: Int!

  "Index of the current page."
  currentPage: Int!

  "Index of the first item in the current page."
  firstItem: Int

  "Index of the last item in the current page."
  lastItem: Int

  "Number of items per page."
  perPage: Int!

  "Are there more pages after this one?"
  hasMorePages: Boolean!
}

It can be queried like this:

{
  posts(first: 10) {
    data {
      id
      title
    }
    paginatorInfo {
      currentPage
    }
  }
}

# Default count

You can supply a defaultCount to set a default count for any type of pagination.

type Query {
  posts: [Post!]! @paginate(type: CONNECTION, defaultCount: 25)
}

This lets you omit the count argument when querying:

query {
  posts {
    id
    name
  }
}

# Limit maximum count

Lighthouse allows you to specify a global maximum for the number of items a user can request through pagination through the config. You may also overwrite this per field with the maxCount argument:

type Query {
  posts: [Post!]! @paginate(maxCount: 10)
}

# Overwrite model

By default, Lighthouse looks for an Eloquent model in the configured default namespace, with the same name as the returned type. You can overwrite this by setting the model argument.

type Query {
  posts: [Post!]! @paginate(model: "App\\Blog\\BlogPost")
}

# Custom builder

If simply querying Eloquent does not fit your use-case, you can specify a custom builder.

type Query {
  blogStatistics: [BlogStatistic!]! @paginate(builder: "App\\Blog@statistics")
}

Your method receives the typical resolver arguments and has to return an instance of Illuminate\Database\Query\Builder.

namespace App\Models;

use Illuminate\Support\Facades\DB;
use Illuminate\Database\Query\Builder;
use GraphQL\Type\Definition\ResolveInfo;
use Nuwave\Lighthouse\Support\Contracts\GraphQLContext;

class Blog
{
    public function statistics($root, array $args, GraphQLContext $context, ResolveInfo $resolveInfo): Builder
    {
        return DB::table('posts')
            ->leftJoinSub(...)
            ->groupBy(...);
    }
}

# Custom resolver

You can provide your own function that resolves the field by directly returning data in a \Illuminate\Contracts\Pagination\Paginator instance.

This is mutually exclusive with builder and model. Not compatible with scopes and builder arguments such as @eq.

type Query {
  posts: [Post!]! @paginate(resolver: "App\\GraphQL\\Queries\\Posts")
}

A custom resolver function may look like the following:

namespace App\GraphQL\Queries;

use Illuminate\Pagination\LengthAwarePaginator;

final class Posts
{
    /**
     * @param  null  $root Always null, since this field has no parent.
     * @param  array{}  $args The field arguments passed by the client.
     * @param  \Nuwave\Lighthouse\Support\Contracts\GraphQLContext  $context Shared between all fields.
     * @param  \GraphQL\Type\Definition\ResolveInfo  $resolveInfo Metadata for advanced query resolution.
     */
    public function __invoke($root, array $args, GraphQLContext $context, ResolveInfo $resolveInfo): LengthAwarePaginator
    {
        //...apply your logic
        return new LengthAwarePaginator([
            [
                'id' => 1,
                'title' => 'Flying teacup found in solar orbit',
            ],
            [
                'id' => 2,
                'title' => 'What actually is the difference between cookies and biscuits?',
            ],
        ], 2, 15);
    }
}

# @rename

"""
Change the internally used name of a field or argument.

This does not change the schema from a client perspective.
"""
directive @rename(
  """
  The internal name of an attribute/property/key.
  """
  attribute: String!
) on FIELD_DEFINITION | ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

This can often be useful to ensure consistent naming of your schema without having to change the underlying models.

type User {
  createdAt: String! @rename(attribute: "created_at")
}

input UserInput {
  firstName: String! @rename(attribute: "first_name")
}

# @restore

"""
Un-delete one or more soft deleted models by their ID.
The field must have a single non-null argument that may be a list.
"""
directive @restore(
  """
  DEPRECATED use @globalId, will be removed in v6

  Set to `true` to use global ids for finding the model.
  If set to `false`, regular non-global ids are used.
  """
  globalId: Boolean = false

  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String
) on FIELD_DEFINITION

Use it on a root mutation field that returns an instance of the Model.

type Mutation {
  restorePost(id: ID!): Post @restore
}

Works very similar to the @delete directive.

# @rules

"""
Validate an argument using [Laravel validation](https://laravel.com/docs/validation).
"""
directive @rules(
  """
  Specify the validation rules to apply to the field.
  This can either be a reference to [Laravel's built-in validation rules](https://laravel.com/docs/validation#available-validation-rules),
  or the fully qualified class name of a custom validation rule.

  Rules that mutate the incoming arguments, such as `exclude_if`, are not supported
  by Lighthouse. Use ArgTransformerDirectives or FieldMiddlewareDirectives instead.
  """
  apply: [String!]!

  """
  Specify a custom attribute name to use in your validation message.
  """
  attribute: String

  """
  Specify the messages to return if the validators fail.
  """
  messages: [RulesMessage!]
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

"""
Input for the `messages` argument of `@rules`.
"""
input RulesMessage {
  """
  Name of the rule, e.g. `"email"`.
  """
  rule: String!

  """
  Message to display if the rule fails, e.g. `"Must be a valid email"`.
  """
  message: String!
}

For example, this rule ensures that users pass a valid 2 character country code:

type Query {
  users(countryCode: String @rules(apply: ["string", "size:2"])): [User!]! @all
}

# @rulesForArray

"""
Run validation on an array itself, using [Laravel built-in validation](https://laravel.com/docs/validation).
"""
directive @rulesForArray(
  """
  Specify the validation rules to apply to the field.
  This can either be a reference to any of Laravel's built-in validation rules: https://laravel.com/docs/validation#available-validation-rules,
  or the fully qualified class name of a custom validation rule.
  """
  apply: [String!]!

  """
  Specify a custom attribute name to use in your validation message.
  """
  attribute: String

  """
  Specify the messages to return if the validators fail.
  """
  messages: [RulesForArrayMessage!]
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

"""
Input for the `messages` argument of `@rulesForArray`.
"""
input RulesForArrayMessage {
  """
  Name of the rule, e.g. `"email"`.
  """
  rule: String!

  """
  Message to display if the rule fails, e.g. `"Must be a valid email"`.
  """
  message: String!
}

This is typically used to assert a certain number of elements is given in a list.

type Mutation {
  saveIcecream(
    flavors: [IcecreamFlavor!]! @rulesForArray(apply: ["min:3"])
  ): Icecream
}

# @scalar

"""
Reference a class implementing a scalar definition.
"""
directive @scalar(
  """
  Reference to a class that extends `\GraphQL\Type\Definition\ScalarType`.
  """
  class: String!
) on SCALAR

If you follow the namespace convention, you do not need this directive. Lighthouse looks into your configured scalar namespace for a class with the same name.

Learn how to implement your own scalar. (opens new window)

scalar DateTime @scalar(class: "DateTimeScalar")

If your class is not in the default namespace, pass a fully qualified class name.

scalar DateTime
  @scalar(class: "Nuwave\\Lighthouse\\Schema\\Types\\Scalars\\DateTime")

# @scope

"""
Adds a scope to the query builder.

The scope method will receive the client-given value of the argument as the second parameter.
"""
directive @scope(
  """
  The name of the scope.
  Defaults to the name of the argument.
  """
  name: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

type Query {
  posts(trending: Boolean @scope): [Post!]! @all
}

The scope will be passed the value of the client-given argument:

use Illuminate\Database\Eloquent\Builder;
use Illuminate\Database\Eloquent\Model;

class Post extends Model
{
    public function scopeTrending(Builder $query, bool $trending): Builder { ... }
}

You can use the name argument if your scope is named differently from your argument:

type Query {
  posts(isTrending: Boolean @scope(name: "trending")): [Post!] @all
}

# @search

"""
Perform a full-text search by the given input value.
"""
directive @search(
  """
  Specify a custom index to use for search.
  """
  within: String
) on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

The search() method of the model is called with the value of the argument, using the driver you configured for Laravel Scout (opens new window).

type Query {
  posts(search: String @search): [Post!]! @paginate
}

The @search directive only works in combination with filter directives that implement \Nuwave\Lighthouse\Scout\ScoutBuilderDirective:

@eq
@softDeletes

Normally the search will be performed using the index specified by the model's searchableAs method. However, in some situation a custom index might be needed, this can be achieved by using the argument within.

type Query {
  posts(search: String @search(within: "my.index")): [Post!]! @paginate
}

# @softDeletes

"""
Allows to filter if trashed elements should be fetched.
This manipulates the schema by adding the argument
`trashed: Trashed @trashed` to the field.
"""
directive @softDeletes on FIELD_DEFINITION

The following schema definition from a .graphql file:

type Query {
  tasks: [Tasks!]! @all @softDeletes
}

Will result in a schema that looks like this:

type Query {
  tasks(trashed: Trashed @trashed): [Tasks!]! @all
}

Find out how the added filter works: @trashed

# @spread

"""
Merge the fields of a nested input object into the arguments of its parent
when processing the field arguments given by a client.
"""
directive @spread on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

You may use @spread on field arguments or on input object fields:

type Mutation {
  updatePost(id: ID!, input: PostInput! @spread): Post @update
}

input PostInput {
  title: String!
  content: PostContent @spread
}

input PostContent {
  imageUrl: String
}

The schema does not change, client side usage works as if @spread was not there:

mutation {
  updatePost(
    id: 12
    input: {
      title: "My awesome title"
      content: { imageUrl: "https://some.site/image.jpg" }
    }
  ) {
    id
  }
}

Internally, the arguments will be transformed into a flat structure before they are passed along to the resolver:

[
    'id' => 12,
    'title' => 'My awesome title',
    'imageUrl' = 'https://some.site/image.jpg',
]

Note that Lighthouse spreads out the arguments after all other ArgDirectives have been applied, e.g. validation, transformation.

# @subscription

"""
Reference a class to handle the broadcasting of a subscription to clients.
The given class must extend `\Nuwave\Lighthouse\Schema\Types\GraphQLSubscription`.
"""
directive @subscription(
  """
  A reference to a subclass of `\Nuwave\Lighthouse\Schema\Types\GraphQLSubscription`.
  """
  class: String!
) on FIELD_DEFINITION

If you follow the default naming conventions for defining subscription fields you do not need this directive. It is only useful if you need to override the default namespace.

type Subscription {
  postUpdated(author: ID!): Post
    @subscription(class: "App\\GraphQL\\Blog\\PostUpdatedSubscription")
}

# @throttle

"""
Sets rate limit to access the field. Does the same as ThrottleRequests Laravel Middleware.
"""
directive @throttle(
  """
  Named preconfigured rate limiter. Requires Laravel 8.x or later.
  """
  name: String

  """
  Maximum number of attempts in a specified time interval.
  """
  maxAttempts: Int = 60

  """
  Time in minutes to reset attempts.
  """
  decayMinutes: Float = 1.0

  """
  Prefix to distinguish several field groups.
  """
  prefix: String
) on FIELD_DEFINITION

Allows use Laravel throttling on a per-field basis. See Laravel doc (opens new window) on how to configure named limiters.

Limiters that return response are not supported. Hashes are different from the ones of Laravel, so one can't use one named limiter to limit both Laravel route and GraphQL field.

# @trashed

"""
Allows to filter if trashed elements should be fetched.
"""
directive @trashed on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

The most convenient way to use this directive is through @softDeletes.

If you want to add it manually, make sure the argument is of the enum type Trashed:

type Query {
  flights(trashed: Trashed @trashed): [Flight!]! @all
}

# @trim

"""
Remove whitespace from the beginning and end of a given input.

This can be used on:
- a single argument or input field to sanitize that subtree
- a field to trim all strings
"""
directive @trim on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION | FIELD_DEFINITION

Whitespace around the passed in string will be removed.

type Mutation {
  createUser(name: String @trim): User
}

Usage on a field applies trim recursively to all inputs.

type Mutation {
  createUser(input: CreateUserInput): User @trim
}

If you want this for all your fields, consider adding this directive to your global field middleware in lighthouse.php:

    'field_middleware' => [
        \Nuwave\Lighthouse\Schema\Directives\TrimDirective::class,
        ...
    ],

# @union

"""
Use a custom function to determine the concrete type of unions.
"""
directive @union(
  """
  Reference a function that returns the implementing Object Type.
  Consists of two parts: a class name and a method name, seperated by an `@` symbol.
  If you pass only a class name, the method name defaults to `__invoke`.
  """
  resolveType: String!
) on UNION

Make sure you read the basics about Unions before deciding to use this directive, you probably don't need it.

type User {
  id: ID!
}

type Employee {
  employeeId: ID!
}

union Person @union(resolveType: "App\\GraphQL\\Unions\\Person@resolveType") =
    User
  | Employee

The function receives the value of the parent field as its single argument and must resolve an Object Type from Lighthouse's TypeRegistry.

namespace App\GraphQL\Unions;

use GraphQL\Type\Definition\Type;
use GraphQL\Type\Definition\ResolveInfo;
use Nuwave\Lighthouse\Schema\TypeRegistry;
use Nuwave\Lighthouse\Support\Contracts\GraphQLContext;

class Person
{
    /**
     * @var \Nuwave\Lighthouse\Schema\TypeRegistry
     */
    protected $typeRegistry;

    public function __construct(TypeRegistry $typeRegistry)
    {
        $this->typeRegistry = $typeRegistry;
    }

    /**
     * Decide which GraphQL type a resolved value has.
     *
     * @param  mixed  $rootValue The value that was resolved by the field. Usually an Eloquent model.
     * @param  \Nuwave\Lighthouse\Support\Contracts\GraphQLContext  $context
     * @param  \GraphQL\Type\Definition\ResolveInfo  $resolveInfo
     * @return \GraphQL\Type\Definition\Type
     */
    public function resolveType($rootValue, GraphQLContext $context, ResolveInfo $resolveInfo): Type
    {
        // Default to getting a type with the same name as the passed in root value
        // TODO implement your own resolver logic - if the default is fine, just delete this class
        return $this->typeRegistry->get(class_basename($rootValue));
    }
}

# @update

"""
Update an Eloquent model with the input values of the field.
"""
directive @update(
  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String

  """
  Specify the name of the relation on the parent model.
  This is only needed when using this directive as a nested arg
  resolver and if the name of the relation is not the arg name.
  """
  relation: String
) on FIELD_DEFINITION | ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

Use it on a root mutation field that returns an instance of the Model.

type Mutation {
  updatePost(id: ID!, content: String): Post @update
}

If the primary key of your model is not called id, it is recommended to rename it. Client libraries such as Apollo base their caching mechanism on that assumption.

type Mutation {
  updatePost(id: ID! @rename(attribute: "post_id"), content: String): Post
    @update
}

If the name of the Eloquent model does not match the return type of the field, or is located in a non-default namespace, set it with the model argument.

type Mutation {
  updateAuthor(id: ID!, name: String): Author @update(model: "App\\User")
}

This directive can also be used as a nested arg resolver.

# @upload

"""
Uploads given file to storage, removes the argument and sets
the returned path to the attribute key provided.

This does not change the schema from a client perspective.
"""
directive @upload(
  """
  The storage disk to be used, defaults to config value `filesystems.default`.
  """
  disk: String

  """
  The path where the file should be stored.
  """
  path: String! = "/"

  """
  Should the visibility be public?
  """
  public: Boolean! = false
) on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

This is useful when you want pass in a file as an argument but have it upload and resolve to a filepath. For example, you want to pass in a user avatar, have that file uploaded and the resulting filepath stored as a row in a database table.

type Mutation {
  createUser(
    avatar: Upload @upload(disk: "public", path: "images/avatars", public: true)
  ): User @create
}

type User {
  avatar: String!
}

# @upsert

"""
Create or update an Eloquent model with the input values of the field.
"""
directive @upsert(
  """
  Specify the class name of the model to use.
  This is only needed when the default model detection does not work.
  """
  model: String

  """
  DEPRECATED use @globalId, will be removed in v6

  Set to `true` to use global ids for finding the model.
  If set to `false`, regular non-global ids are used.
  """
  globalId: Boolean = false

  """
  Specify the name of the relation on the parent model.
  This is only needed when using this directive as a nested arg
  resolver and if the name of the relation is not the arg name.
  """
  relation: String
) on FIELD_DEFINITION | ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

Lighthouse will try to fetch the model by its primary key, just like @update. If the model doesn't exist, it will be newly created with a given id. In case no id is specified, an auto-generated fresh ID will be used instead.

type Mutation {
  upsertPost(post_id: ID!, content: String): Post @upsert
}

This directive can also be used as a nested arg resolver.

# @validator

"""
Provide validation rules through a PHP class.
"""
directive @validator(
  """
  The name of the class to use.

  If defined on an input, this defaults to a class called `{$inputName}Validator` in the
  default validator namespace. For fields, it uses the namespace of the parent type
  and the field name: `{$parent}\{$field}Validator`.
  """
  class: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION | FIELD_DEFINITION | INPUT_OBJECT

# @where

"""
Use an input value as a [where filter](https://laravel.com/docs/queries#where-clauses).
"""
directive @where(
  """
  Specify the operator to use within the WHERE condition.
  """
  operator: String = "="

  """
  Specify the database column to compare.
  Only required if database column has a different name than the attribute in your schema.
  """
  key: String

  """
  Use Laravel's where clauses upon the query builder.
  This only works for clauses with the signature (string $column, string $operator, mixed $value).
  """
  clause: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

You can specify simple operators:

type Query {
  postsSearchTitle(title: String! @where(operator: "like")): [Post!]! @all
}

Or use the additional clauses that Laravel provides:

type Query {
  postsByYear(created_at: Int! @where(clause: "whereYear")): [Post!]! @all
}

# @whereAuth

"""
Filter a type to only return instances owned by the current user.
"""
directive @whereAuth(
  """
  Name of the relationship that links to the user model.
  """
  relation: String!

  """
  Specify which guard to use, e.g. "api".
  When not defined, the default from `lighthouse.php` is used.
  """
  guard: String
) on FIELD_DEFINITION

The following query returns all posts that belong to the currently authenticated user.
Behind the scenes it is using a whereHas query.

type Query {
  posts: [Post!]! @all @whereAuth(relation: "user")
}

# @whereBetween

"""
Verify that a column's value is between two values.

The type of the input value this is defined upon should be
an `input` object with two fields.
"""
directive @whereBetween(
  """
  Specify the database column to compare.
  Only required if database column has a different name than the attribute in your schema.
  """
  key: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

This example defines an input to filter that a value is between two dates.

type Query {
  posts(created_at: DateRange @whereBetween): [Post!]! @all
}

input DateRange {
  from: Date!
  to: Date!
}

You may use any custom input type for the argument. Make sure it has exactly two required fields to ensure the query is valid.

# @whereConditions

The documentation for this directive is found in Complex Where Conditions.

# @whereHasConditions

The documentation for this directive is found in Complex Where Conditions.

# @whereJsonContains

"""
Use an input value as a [whereJsonContains filter](https://laravel.com/docs/queries#json-where-clauses).
"""
directive @whereJsonContains(
  """
  Specify the database column and path inside the JSON to compare.
  Only required if database column has a different name than the attribute in your schema.
  """
  key: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

type Query {
  posts(tags: [String]! @whereJsonContains): [Post!]! @all
}

You may use the key argument to look into the JSON content:

type Query {
  posts(tags: [String]! @whereJsonContains(key: "tags->recent")): [Post!]! @all
}

# @whereNotBetween

"""
Verify that a column's value lies outside of two values.

The type of the input value this is defined upon should be
an `input` object with two fields.
"""
directive @whereNotBetween(
  """
  Specify the database column to compare.
  Only required if database column has a different name than the attribute in your schema.
  """
  key: String
) repeatable on ARGUMENT_DEFINITION | INPUT_FIELD_DEFINITION

type Query {
  posts(
    notCreatedDuring: DateRange @whereNotBetween(key: "created_at")
  ): [Post!]! @all
}

input DateRange {
  from: Date!
  to: Date!
}

# @with

"""
Eager-load an Eloquent relation.
"""
directive @with(
  """
  Specify the relationship method name in the model class,
  if it is named different from the field in the schema.
  """
  relation: String

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) repeatable on FIELD_DEFINITION

This can be a useful optimization for fields that are not returned directly but rather used for resolving other fields.

type User {
  taskSummary: String! @with(relation: "tasks") @method(name: "getTaskSummary")
}

If you just want to return the relation itself as-is, look into handling Eloquent relationships.

# @withCount

"""
Eager-load the count of an Eloquent relation if the field is queried.

Note that this does not return a value for the field, the count is simply
prefetched, assuming it is used to compute the field value. Use `@count`
if the field should simply return the relation count.
"""
directive @withCount(
  """
  Specify the relationship method name in the model class.
  """
  relation: String!

  """
  Apply scopes to the underlying query.
  """
  scopes: [String!]
) repeatable on FIELD_DEFINITION

This can be a useful optimization for fields that use the count to compute a result.

type User {
  activityStatistics: ActivityStatistics! @withCount(relation: "posts")
}

If you just want to return the count itself as-is, use @count.

← Request Lifecycle Resolvers →