Skip to main content

Federation implementation for graphene

Project description

graphene-federation

Federation support for Graphene Logo Graphene following the Apollo Federation specifications.

PyPI version Unit Tests Status Coverage Status Integration Tests Status

This repository is heavily based on the repo it was forked from... Huge thanks to Preply for setting up the foundations.

WARNING: This version is not compatible with graphene version below v3. If you need to use a version compatible with graphene v2 I recommend using the version 1.0.0 of graphene_federation.


Supported Features

  • sdl (_service on field): enable to add schema in federation (as is)

Apollo Spec Supported

  • v1.0
  • v2.0
  • v2.1
  • v2.2
  • v2.3
  • v2.4
  • v2.5
  • v2.6 STABLE_VERSION . Rover dev supports only upto v2.6
  • v2.7 LATEST_VERSION

All directives could be easily integrated with the help of graphene-directives. Now every directive's values are validated at run time itself by graphene-directives.

Directives (v2.7)

directive @composeDirective(name: String!) repeatable on SCHEMA
directive @extends on OBJECT | INTERFACE
directive @external on OBJECT | FIELD_DEFINITION
directive @key(fields: FieldSet!, resolvable: Boolean = true) repeatable on OBJECT | INTERFACE
directive @inaccessible on
  | FIELD_DEFINITION
  | OBJECT
  | INTERFACE
  | UNION
  | ENUM
  | ENUM_VALUE
  | SCALAR
  | INPUT_OBJECT
  | INPUT_FIELD_DEFINITION
  | ARGUMENT_DEFINITION
directive @interfaceObject on OBJECT
directive @override(from: String!, label: String) on FIELD_DEFINITION
directive @provides(fields: FieldSet!) on FIELD_DEFINITION
directive @requires(fields: FieldSet!) on FIELD_DEFINITION
directive @shareable repeatable on FIELD_DEFINITION | OBJECT
directive @tag(name: String!) repeatable on
  | FIELD_DEFINITION
  | INTERFACE
  | OBJECT
  | UNION
  | ARGUMENT_DEFINITION
  | SCALAR
  | ENUM
  | ENUM_VALUE
  | INPUT_OBJECT
  | INPUT_FIELD_DEFINITION
directive @authenticated on
    FIELD_DEFINITION
  | OBJECT
  | INTERFACE
  | SCALAR
  | ENUM
directive @requiresScopes(scopes: [[federation__Scope!]!]!) on
    FIELD_DEFINITION
  | OBJECT
  | INTERFACE
  | SCALAR
  | ENUM
directive @policy(policies: [[federation__Policy!]!]!) on
  | FIELD_DEFINITION
  | OBJECT
  | INTERFACE
  | SCALAR
  | ENUM
scalar federation__Policy
scalar federation__Scope
scalar FieldSet

Read about directives in official documentation

Each type which is decorated with @key or @extends is added to the _Entity union. The __resolve_reference method can be defined for each type that is an entity. Note that since the notation with double underscores can be problematic in Python for model inheritance this resolver method can also be named _resolve_reference (the __resolve_reference method will take precedence if both are declared).

This method is called whenever an entity is requested as part of the fulfilling a query plan. If not explicitly defined, the default resolver is used. The default resolver just creates instance of type with passed fieldset as kwargs, see entity.get_entity_query for more details

  • You should define __resolve_reference, if you need to extract object before passing it to fields resolvers (example: FileNode)
  • You should not define __resolve_reference, if fields resolvers need only data passed in fieldset (example: FunnyText) Read more in official documentation.

Example

Here is an example of implementation based on the Apollo Federation introduction example. It implements a federation schema for a basic e-commerce application over three services: accounts, products, reviews.

Accounts

First add an account service that expose a User type that can then be referenced in other services by its id field:

from graphene import Field, Int, ObjectType, String

from graphene_federation import LATEST_VERSION, build_schema, key


@key("id")
class User(ObjectType):
    id = Int(required=True)
    username = String(required=True)

    def __resolve_reference(self, info, **kwargs):
        """
        Here we resolve the reference of the user entity referenced by its `id` field.
        """
        return User(id=self.id, email=f"user_{self.id}@mail.com")


class Query(ObjectType):
    me = Field(User)


schema = build_schema(query=Query, federation_version=LATEST_VERSION)

Product

The product service exposes a Product type that can be used by other services via the upc field:

from graphene import Argument, Int, List, ObjectType, String

from graphene_federation import LATEST_VERSION, build_schema, key


@key("upc")
class Product(ObjectType):
    upc = String(required=True)
    name = String(required=True)
    price = Int()

    def __resolve_reference(self, info, **kwargs):
        """
        Here we resolve the reference of the product entity referenced by its `upc` field.
        """
        return Product(upc=self.upc, name=f"product {self.upc}")


class Query(ObjectType):
    topProducts = List(Product, first=Argument(Int, default_value=5))


schema = build_schema(query=Query, federation_version=LATEST_VERSION)

Reviews

The reviews service exposes a Review type which has a link to both the User and Product types. It also has the ability to provide the username of the User. On top of that it adds to the User/Product types (that are both defined in other services) the ability to get their reviews.

from graphene import Field, Int, List, ObjectType, String

from graphene_federation import LATEST_VERSION, build_schema, external, key, provides


@key("id")
class User(ObjectType):
    id = external(Int(required=True))
    reviews = List(lambda: Review)

    def resolve_reviews(self, info, *args, **kwargs):
        """
        Get all the reviews of a given user. (not implemented here)
        """
        return []


@key("upc")
class Product(ObjectType):
    upc = external(String(required=True))
    reviews = List(lambda: Review)


class Review(ObjectType):
    body = String()
    author = provides(Field(User), fields="username")
    product = Field(Product)


class Query(ObjectType):
    review = Field(Review)


schema = build_schema(query=Query, federation_version=LATEST_VERSION)

Federation

Note that each schema declaration for the services is a valid graphql schema (it only adds the _Entity and _Service types). The best way to check that the decorator are set correctly is to request the service sdl:

from graphql import graphql

query = """
query {
    _service {
        sdl
    }
}
"""

result = graphql(schema, query)
print(result.data["_service"]["sdl"])

Those can then be used in a federated schema.

You can find more examples in the unit / integration tests and examples folder.

There is also a cool example of integration with Mongoengine.


Other Notes

build_schema new arguments

  • schema_directives (Collection[SchemaDirective]): Directives that can be defined at DIRECTIVE_LOCATION.SCHEMA with their argument values.
  • include_graphql_spec_directives (bool): Includes directives defined by GraphQL spec (@include, @skip, @deprecated, @specifiedBy)
  • federation_version (FederationVersion): Specify the version explicit (default STABLE_VERSION)

Directives Additional arguments

  • federation_version: (FederationVersion = LATEST_VERSION) : You can use this to take a directive from a particular federation version

Note: The federation_version in build_schema is given higher priority. If the directive you have chosen is not compatible, it will raise an error

Custom Directives

You can define custom directives as follows

from graphene import Field, ObjectType, String
from graphql import GraphQLArgument, GraphQLInt, GraphQLNonNull

from graphene_federation import ComposableDirective, DirectiveLocation, LATEST_VERSION
from graphene_federation import build_schema

CacheDirective = ComposableDirective(
    name="cache",
    locations=[DirectiveLocation.FIELD_DEFINITION, DirectiveLocation.OBJECT],
    args={
        "maxAge": GraphQLArgument(
            GraphQLNonNull(GraphQLInt), description="Specifies the maximum age for cache in seconds."
        ),
    },
    description="Caching directive to control cache behavior.",
    spec_url="https://specs.example.dev/directives/v1.0",
)

cache = CacheDirective.decorator()


@cache(max_age=20)
class Review(ObjectType):
    body = cache(field=String(), max_age=100)


class Query(ObjectType):
    review = Field(Review)


schema = build_schema(
    query=Query,
    directives=(CacheDirective,),
    federation_version=LATEST_VERSION ,
)

This will automatically add @link and @composeDirective to schema

extend schema
	@link(url: "https://specs.apollo.dev/federation/v2.6", import: ["@composeDirective"])
	@link(url: "https://specs.example.dev/directives/v1.0", import: ["@cache"])
	@composeDirective(name: "@cache")

"""Caching directive to control cache behavior."""
directive @cache(
  """Specifies the maximum age for cache in seconds."""
  maxAge: Int!
) on FIELD_DEFINITION | OBJECT

type Query {
  review: Review
  _service: _Service!
}

type Review  @cache(maxAge: 20) {
  body: String @cache(maxAge: 100)
}

If you wish to add the schema_directives @link @composeDirective manually. You can pass the add_to_schema_directives as False

from graphene import Field, ObjectType, String
from graphql import GraphQLArgument, GraphQLInt, GraphQLNonNull

from graphene_federation import (ComposableDirective, DirectiveLocation, LATEST_VERSION, build_schema,
                                 compose_directive, link_directive)

CacheDirective = ComposableDirective(
    name="cache",
    locations=[DirectiveLocation.FIELD_DEFINITION, DirectiveLocation.OBJECT],
    args={
        "maxAge": GraphQLArgument(
            GraphQLNonNull(GraphQLInt), description="Specifies the maximum age for cache in seconds."
        ),
    },
    description="Caching directive to control cache behavior.",
    add_to_schema_directives=False
)

cache = CacheDirective.decorator()


@cache(max_age=20)
class Review(ObjectType):
    body = cache(field=String(), max_age=100)


class Query(ObjectType):
    review = Field(Review)


schema = build_schema(
    query=Query,
    directives=(CacheDirective,),
    schema_directives=(
        link_directive(url="https://specs.example.dev/directives/v1.0", import_=['@cache']),
        compose_directive(name='@cache'),
    ),
    federation_version=LATEST_VERSION,
)

Custom field name

When using decorator on a field with custom name

Case 1 (auto_camelcase=False)

@key("identifier")
@key("validEmail")
class User(ObjectType):
    identifier = ID()
    email = String(name="validEmail")

class Query(ObjectType):
    user = Field(User)

schema = build_schema(query=Query, federation_version=LATEST_VERSION, auto_camelcase=False) # Disable auto_camelcase

This works correctly. By default fields of @key,@requires and @provides are not converted to camel case if auto_camelcase is set to False

Case 2 (auto_camelcase=True)

@key("identifier")
@key("valid_email")
class User(ObjectType):
    identifier = ID()
    email = String(name="valid_email")

class Query(ObjectType):
    user = Field(User)

schema = build_schema(query=Query, federation_version=LATEST_VERSION) # auto_camelcase Enabled

This will raise an error @key, field "validEmail" does not exist on type "User". Because The decorator auto camel-cased the field value of key, as schema has auto_camelcase=True (default)

To fix this, pass auto_case=False in the @key, @requires or @provides argument

@key("identifier")
@key("valid_email", auto_case=False)
class User(ObjectType):
    identifier = ID()
    email = String(name="valid_email")

class Query(ObjectType):
    user = Field(User)

schema = build_schema(query=Query, federation_version=LATEST_VERSION) # auto_camelcase=True

Known Issues

  • Using @composeDirective with @link in Federation v2.6 shows error in rover, rover cli only supports upto v2.5 as of 16/01/2024

Contributing

  • You can run the unit tests by doing: make tests.
  • You can run the integration tests by doing make integration-build && make integration-test.
  • You can get a development environment (on a Docker container) with make dev-setup.
  • You should use black to format your code.

The tests are automatically run on Travis CI on push to GitHub.


Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

graphene-federation-3.2.0.tar.gz (27.1 kB view hashes)

Uploaded Source

Built Distribution

graphene_federation-3.2.0-py3-none-any.whl (42.5 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page