aws-cdk.aws-appsync 1.87.1

The CDK Construct Library for AWS::AppSync

AWS AppSync Construct Library

---

All classes with the Cfn prefix in this module (CFN Resources) are always stable and safe to use.

The APIs of higher level constructs in this module are experimental and under active development. They are subject to non-backward compatible changes or removal in any future version. These are not subject to the Semantic Versioning model and breaking changes will be announced in the release notes. This means that while you may use them, you may need to update your source code when upgrading to a newer version of this package.

---

The @aws-cdk/aws-appsync package contains constructs for building flexible APIs that use GraphQL.

Example

DynamoDB

Example of a GraphQL API with AWS_IAM authorization resolving into a DynamoDb backend data source.

GraphQL schema file schema.graphql:

type demo {
  id: String!
  version: String!
}
type Query {
  getDemos: [ demo! ]
}
input DemoInput {
  version: String!
}
type Mutation {
  addDemo(input: DemoInput!): demo
}

CDK stack file app-stack.ts:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
import aws_cdk.aws_appsync as appsync
import aws_cdk.aws_dynamodb as db

api = appsync.GraphqlApi(stack, "Api",
    name="demo",
    schema=appsync.Schema.from_asset(join(__dirname, "schema.graphql")),
    authorization_config=AuthorizationConfig(
        default_authorization=AuthorizationMode(
            authorization_type=appsync.AuthorizationType.IAM
        )
    ),
    xray_enabled=True
)

demo_table = db.Table(stack, "DemoTable",
    partition_key=Attribute(
        name="id",
        type=db.AttributeType.STRING
    )
)

demo_dS = api.add_dynamo_db_data_source("demoDataSource", demo_table)

# Resolver for the Query "getDemos" that scans the DynamoDb table and returns the entire list.
demo_dS.create_resolver(
    type_name="Query",
    field_name="getDemos",
    request_mapping_template=appsync.MappingTemplate.dynamo_db_scan_table(),
    response_mapping_template=appsync.MappingTemplate.dynamo_db_result_list()
)

# Resolver for the Mutation "addDemo" that puts the item into the DynamoDb table.
demo_dS.create_resolver(
    type_name="Mutation",
    field_name="addDemo",
    request_mapping_template=appsync.MappingTemplate.dynamo_db_put_item(
        appsync.PrimaryKey.partition("id").auto(),
        appsync.Values.projecting("demo")),
    response_mapping_template=appsync.MappingTemplate.dynamo_db_result_item()
)

Aurora Serverless

AppSync provides a data source for executing SQL commands against Amazon Aurora Serverless clusters. You can use AppSync resolvers to execute SQL statements against the Data API with GraphQL queries, mutations, and subscriptions.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
# Create username and password secret for DB Cluster
secret = rds.DatabaseSecret(stack, "AuroraSecret",
    username="clusteradmin"
)

# Create the DB cluster, provide all values needed to customise the database.
cluster = rds.DatabaseCluster(stack, "AuroraCluster",
    engine=rds.DatabaseClusterEngine.aurora_mysql(version=rds.AuroraMysqlEngineVersion.VER_2_07_1),
    credentials={"username": "clusteradmin"},
    cluster_identifier="db-endpoint-test",
    default_database_name="demos"
)

# Build a data source for AppSync to access the database.
rds_dS = api.add_rds_data_source("rds", "The rds data source", cluster, secret)

# Set up a resolver for an RDS query.
rds_dS.create_resolver(
    type_name="Query",
    field_name="getDemosRds",
    request_mapping_template=MappingTemplate.from_string("\n  {\n    \"version\": \"2018-05-29\",\n    \"statements\": [\n      \"SELECT * FROM demos\"\n    ]\n  }\n  "),
    response_mapping_template=MappingTemplate.from_string("\n    $util.rds.toJsonObject($ctx.result)\n  ")
)

# Set up a resolver for an RDS mutation.
rds_dS.create_resolver(
    type_name="Mutation",
    field_name="addDemoRds",
    request_mapping_template=MappingTemplate.from_string("\n  {\n    \"version\": \"2018-05-29\",\n    \"statements\": [\n      \"INSERT INTO demos VALUES (:id, :version)\",\n      \"SELECT * WHERE id = :id\"\n    ],\n    \"variableMap\": {\n      \":id\": $util.toJson($util.autoId()),\n      \":version\": $util.toJson($ctx.args.version)\n    }\n  }\n  "),
    response_mapping_template=MappingTemplate.from_string("\n    $util.rds.toJsonObject($ctx.result)\n  ")
)

HTTP Endpoints

GraphQL schema file schema.graphql:

type job {
  id: String!
  version: String!
}

input DemoInput {
  version: String!
}

type Mutation {
  callStepFunction(input: DemoInput!): job
}

GraphQL request mapping template request.vtl:

{
  "version": "2018-05-29",
  "method": "POST",
  "resourcePath": "/",
  "params": {
    "headers": {
      "content-type": "application/x-amz-json-1.0",
      "x-amz-target":"AWSStepFunctions.StartExecution"
    },
    "body": {
      "stateMachineArn": "<your step functions arn>",
      "input": "{ \"id\": \"$context.arguments.id\" }"
    }
  }
}

GraphQL response mapping template response.vtl:

{
  "id": "${context.result.id}"
}

CDK stack file app-stack.ts:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
import aws_cdk.aws_appsync as appsync

api = appsync.GraphqlApi(scope, "api",
    name="api",
    schema=appsync.Schema.from_file(join(__dirname, "schema.graphql"))
)

http_ds = api.add_http_data_source("ds", "https://states.amazonaws.com",
    name="httpDsWithStepF",
    description="from appsync to StepFunctions Workflow",
    authorization_config=AwsIamConfig(
        signing_region="us-east-1",
        signing_service_name="states"
    )
)

http_ds.create_resolver(
    type_name="Mutation",
    field_name="callStepFunction",
    request_mapping_template=MappingTemplate.from_file("request.vtl"),
    response_mapping_template=MappingTemplate.from_file("response.vtl")
)

Schema

Every GraphQL Api needs a schema to define the Api. CDK offers appsync.Schema for static convenience methods for various types of schema declaration: code-first or schema-first.

Code-First

When declaring your GraphQL Api, CDK defaults to a code-first approach if the schema property is not configured.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
api = appsync.GraphqlApi(stack, "api", name="myApi")

CDK will declare a Schema class that will give your Api access functions to define your schema code-first: addType, addObjectType, addToSchema, etc.

You can also declare your Schema class outside of your CDK stack, to define your schema externally.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
schema = appsync.Schema()
schema.add_object_type("demo",
    definition={"id": appsync.GraphqlType.id()}
)
api = appsync.GraphqlApi(stack, "api",
    name="myApi",
    schema=schema
)

See the code-first schema section for more details.

Schema-First

You can define your GraphQL Schema from a file on disk. For convenience, use the appsync.Schema.fromAsset to specify the file representing your schema.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
api = appsync.GraphqlApi(stack, "api",
    name="myApi",
    schema=appsync.Schema.from_asset(join(__dirname, "schema.graphl"))
)

Imports

Any GraphQL Api that has been created outside the stack can be imported from another stack into your CDK app. Utilizing the fromXxx function, you have the ability to add data sources and resolvers through a IGraphqlApi interface.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
imported_api = appsync.GraphqlApi.from_graphql_api_attributes(stack, "IApi",
    graphql_api_id=api.api_id,
    graphql_arn=api.arn
)
imported_api.add_dynamo_db_data_source("TableDataSource", table)

If you don't specify graphqlArn in fromXxxAttributes, CDK will autogenerate the expected arn for the imported api, given the apiId. For creating data sources and resolvers, an apiId is sufficient.

Permissions

When using AWS_IAM as the authorization type for GraphQL API, an IAM Role with correct permissions must be used for access to API.

When configuring permissions, you can specify specific resources to only be accessible by IAM authorization. For example, if you want to only allow mutability for IAM authorized access you would configure the following.

In schema.graphql:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
type Mutation {
  updateExample(...): ...@aws_iam

In IAM:

{
   "Version": "2012-10-17",
   "Statement": [
      {
         "Effect": "Allow",
         "Action": [
            "appsync:GraphQL"
         ],
         "Resource": [
            "arn:aws:appsync:REGION:ACCOUNT_ID:apis/GRAPHQL_ID/types/Mutation/fields/updateExample"
         ]
      }
   ]
}

See documentation for more details.

To make this easier, CDK provides grant API.

Use the grant function for more granular authorization.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
role = iam.Role(stack, "Role",
    assumed_by=iam.ServicePrincipal("lambda.amazonaws.com")
)
api = appsync.GraphqlApi(stack, "API",
    definition=definition
)

api.grant(role, appsync.IamResource.custom("types/Mutation/fields/updateExample"), "appsync:GraphQL")

IamResource

In order to use the grant functions, you need to use the class IamResource.

• IamResource.custom(...arns) permits custom ARNs and requires an argument.
• IamResouce.ofType(type, ...fields) permits ARNs for types and their fields.
• IamResource.all() permits ALL resources.

Generic Permissions

Alternatively, you can use more generic grant functions to accomplish the same usage.

These include:

• grantMutation (use to grant access to Mutation fields)
• grantQuery (use to grant access to Query fields)
• grantSubscription (use to grant access to Subscription fields)

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
# For generic types
api.grant_mutation(role, "updateExample")

# For custom types and granular design
api.grant(role, appsync.IamResource.of_type("Mutation", "updateExample"), "appsync:GraphQL")

Pipeline Resolvers and AppSync Functions

AppSync Functions are local functions that perform certain operations onto a backend data source. Developers can compose operations (Functions) and execute them in sequence with Pipeline Resolvers.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
appsync_function = appsync.AppsyncFunction(stack, "function",
    name="appsync_function",
    api=api,
    data_source=api_data_source,
    request_mapping_template=appsync.MappingTemplate.from_file("request.vtl"),
    response_mapping_template=appsync.MappingTemplate.from_file("response.vtl")
)

AppSync Functions are used in tandem with pipeline resolvers to compose multiple operations.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
pipeline_resolver = appsync.Resolver(stack, "pipeline",
    name="pipeline_resolver",
    api=api,
    data_source=api_data_source,
    request_mapping_template=appsync.MappingTemplate.from_file("beforeRequest.vtl"),
    pipeline_config=[appsync_function],
    response_mapping_template=appsync.MappingTemplate.from_file("afterResponse.vtl")
)

Learn more about Pipeline Resolvers and AppSync Functions here.

Code-First Schema

CDK offers the ability to generate your schema in a code-first approach. A code-first approach offers a developer workflow with:

• modularity: organizing schema type definitions into different files
• reusability: simplifying down boilerplate/repetitive code
• consistency: resolvers and schema definition will always be synced

The code-first approach allows for dynamic schema generation. You can generate your schema based on variables and templates to reduce code duplication.

Code-First Example

To showcase the code-first approach. Let's try to model the following schema segment.

interface Node {
  id: String
}

type Query {
  allFilms(after: String, first: Int, before: String, last: Int): FilmConnection
}

type FilmNode implements Node {
  filmName: String
}

type FilmConnection {
  edges: [FilmEdge]
  films: [Film]
  totalCount: Int
}

type FilmEdge {
  node: Film
  cursor: String
}

Above we see a schema that allows for generating paginated responses. For example, we can query allFilms(first: 100) since FilmConnection acts as an intermediary for holding FilmEdges we can write a resolver to return the first 100 films.

In a separate file, we can declare our scalar types: scalar-types.ts.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
from aws_cdk.aws_appsync import GraphqlType

string = appsync.GraphqlType.string()
int = appsync.GraphqlType.int()

In another separate file, we can declare our object types and related functions. We will call this file object-types.ts and we will have created it in a way that allows us to generate other XxxConnection and XxxEdges in the future.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
pluralize = require("pluralize")
import ..scalar_types.ts as scalar
import aws_cdk.aws_appsync as appsync

args = {
    "after": scalar.string,
    "first": scalar.int,
    "before": scalar.string,
    "last": scalar.int
}

Node = appsync.InterfaceType("Node", {
    "definition": {"id": scalar.string}
})
FilmNode = appsync.ObjectType.implement_interface("FilmNode",
    interface_types=[Node],
    definition={"film_name": scalar.string}
)

def generate_edge_and_connection(base):
    edge = appsync.ObjectType(f"{base.name}Edge",
        definition={"node": base.attribute(), "cursor": scalar.string}
    )
    connection = appsync.ObjectType(f"{base.name}Connection",
        definition={
            "edges": edges.attribute(is_list=True),
            pluralize(base.name): base.attribute(is_list=True),
            "total_count": scalar.int
        }
    )return {"edge": edge, "connection": connection}

Finally, we will go to our cdk-stack and combine everything together to generate our schema.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
import aws_cdk.aws_appsync as appsync
import ..object_types as schema

api = appsync.GraphqlApi(stack, "Api",
    name="demo"
)

self.object_types = [schema.Node, schema.Film]

film_connections = schema.generate_edge_and_connection(schema.Film)

api.add_query("allFilms", appsync.ResolvableField(
    return_type=film_connections.connection.attribute(),
    args=schema.args,
    data_source=dummy_data_source,
    request_mapping_template=dummy_request,
    response_mapping_template=dummy_response
))

self.object_types.map((t) => api.addType(t))
Object.keys(film_connections).for_each((key) => api.addType(filmConnections[key]))

Notice how we can utilize the generateEdgeAndConnection function to generate Object Types. In the future, if we wanted to create more Object Types, we can simply create the base Object Type (i.e. Film) and from there we can generate its respective Connections and Edges.

Check out a more in-depth example here.

GraphQL Types

One of the benefits of GraphQL is its strongly typed nature. We define the types within an object, query, mutation, interface, etc. as GraphQL Types.

GraphQL Types are the building blocks of types, whether they are scalar, objects, interfaces, etc. GraphQL Types can be:

• Scalar Types: Id, Int, String, AWSDate, etc.
• Object Types: types that you generate (i.e. demo from the example above)
• Interface Types: abstract types that define the base implementation of other Intermediate Types

More concretely, GraphQL Types are simply the types appended to variables. Referencing the object type Demo in the previous example, the GraphQL Types is String! and is applied to both the names id and version.

Directives

Directives are attached to a field or type and affect the execution of queries, mutations, and types. With AppSync, we use Directives to configure authorization. CDK provides static functions to add directives to your Schema.

• Directive.iam() sets a type or field's authorization to be validated through Iam

• Directive.apiKey() sets a type or field's authorization to be validated through a Api Key

• Directive.oidc() sets a type or field's authorization to be validated through OpenID Connect

• Directive.cognito(...groups: string[]) sets a type or field's authorization to be validated through Cognito User Pools

  • groups the name of the cognito groups to give access

To learn more about authorization and directives, read these docs here.

Field and Resolvable Fields

While GraphqlType is a base implementation for GraphQL fields, we have abstractions on top of GraphqlType that provide finer grain support.

Field

Field extends GraphqlType and will allow you to define arguments. Interface Types are not resolvable and this class will allow you to define arguments, but not its resolvers.

For example, if we want to create the following type:

type Node {
  test(argument: string): String
}

The CDK code required would be:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
field = appsync.Field(
    return_type=appsync.GraphqlType.string(),
    args={
        "argument": appsync.GraphqlType.string()
    }
)
type = appsync.InterfaceType("Node",
    definition={"test": field}
)

Resolvable Fields

ResolvableField extends Field and will allow you to define arguments and its resolvers. Object Types can have fields that resolve and perform operations on your backend.

You can also create resolvable fields for object types.

type Info {
  node(id: String): String
}

The CDK code required would be:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
info = appsync.ObjectType("Info",
    definition={
        "node": appsync.ResolvableField(
            return_type=appsync.GraphqlType.string(),
            args={
                "id": appsync.GraphqlType.string()
            },
            data_source=api.add_none_data_source("none"),
            request_mapping_template=dummy_request,
            response_mapping_template=dummy_response
        )
    }
)

To nest resolvers, we can also create top level query types that call upon other types. Building off the previous example, if we want the following graphql type definition:

type Query {
  get(argument: string): Info
}

The CDK code required would be:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
query = appsync.ObjectType("Query",
    definition={
        "get": appsync.ResolvableField(
            return_type=appsync.GraphqlType.string(),
            args={
                "argument": appsync.GraphqlType.string()
            },
            data_source=api.add_none_data_source("none"),
            request_mapping_template=dummy_request,
            response_mapping_template=dummy_response
        )
    }
)

Learn more about fields and resolvers here.

Intermediate Types

Intermediate Types are defined by Graphql Types and Fields. They have a set of defined fields, where each field corresponds to another type in the system. Intermediate Types will be the meat of your GraphQL Schema as they are the types defined by you.

Intermediate Types include:

• Interface Types
• Object Types
• Enum Types
• Input Types
• Union Types

Interface Types

Interface Types are abstract types that define the implementation of other intermediate types. They are useful for eliminating duplication and can be used to generate Object Types with less work.

You can create Interface Types externally.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
node = appsync.InterfaceType("Node",
    definition={
        "id": appsync.GraphqlType.string(is_required=True)
    }
)

To learn more about Interface Types, read the docs here.

Object Types

Object Types are types that you declare. For example, in the code-first example the demo variable is an Object Type. Object Types are defined by GraphQL Types and are only usable when linked to a GraphQL Api.

You can create Object Types in three ways:

1. Object Types can be created externally.

   # Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
   api = appsync.GraphqlApi(stack, "Api",
       name="demo"
   )
   demo = appsync.ObjectType("Demo",
       defintion={
           "id": appsync.GraphqlType.string(is_required=True),
           "version": appsync.GraphqlType.string(is_required=True)
       }
   )
   
   api.add_type(object)
   

   This method allows for reusability and modularity, ideal for larger projects. For example, imagine moving all Object Type definition outside the stack.

   scalar-types.ts - a file for scalar type definitions

   # Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
   required_string = appsync.GraphqlType.string(is_required=True)
   

   object-types.ts - a file for object type definitions

   # Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
   from ..scalar_types import required_string
   demo = appsync.ObjectType("Demo",
       defintion={
           "id": required_string,
           "version": required_string
       }
   )
   

   cdk-stack.ts - a file containing our cdk stack

   # Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
   from ..object_types import demo
   api.add_type(demo)
   

2. Object Types can be created externally from an Interface Type.

   # Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
   node = appsync.InterfaceType("Node",
       definition={
           "id": appsync.GraphqlType.string(is_required=True)
       }
   )
   demo = appsync.ObjectType("Demo",
       interface_types=[node],
       defintion={
           "version": appsync.GraphqlType.string(is_required=True)
       }
   )
   

   This method allows for reusability and modularity, ideal for reducing code duplication.

To learn more about Object Types, read the docs here.

Enum Types

Enum Types are a special type of Intermediate Type. They restrict a particular set of allowed values for other Intermediate Types.

enum Episode {
  NEWHOPE
  EMPIRE
  JEDI
}

This means that wherever we use the type Episode in our schema, we expect it to be exactly one of NEWHOPE, EMPIRE, or JEDI.

The above GraphQL Enumeration Type can be expressed in CDK as the following:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
episode = appsync.EnumType("Episode",
    definition=["NEWHOPE", "EMPIRE", "JEDI"
    ]
)
api.add_type(episode)

To learn more about Enum Types, read the docs here.

Input Types

Input Types are special types of Intermediate Types. They give users an easy way to pass complex objects for top level Mutation and Queries.

input Review {
  stars: Int!
  commentary: String
}

The above GraphQL Input Type can be expressed in CDK as the following:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
review = appsync.InputType("Review",
    definition={
        "stars": GraphqlType.int(is_required=True),
        "commentary": GraphqlType.string()
    }
)
api.add_type(review)

To learn more about Input Types, read the docs here.

Union Types

Union Types are a special type of Intermediate Type. They are similar to Interface Types, but they cannot specify any common fields between types.

Note: the fields of a union type need to be Object Types. In other words, you can't create a union type out of interfaces, other unions, or inputs.

union Search = Human | Droid | Starship

The above GraphQL Union Type encompasses the Object Types of Human, Droid and Starship. It can be expressed in CDK as the following:

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
string = appsync.GraphqlType.string()
human = appsync.ObjectType("Human", definition={"name": string})
droid = appsync.ObjectType("Droid", definition={"name": string})
starship = appsync.ObjectType("Starship", definition={"name": string})
search = appsync.UnionType("Search",
    definition=[human, droid, starship]
)
api.add_type(search)

To learn more about Union Types, read the docs here.

Query

Every schema requires a top level Query type. By default, the schema will look for the Object Type named Query. The top level Query is the only exposed type that users can access to perform GET operations on your Api.

To add fields for these queries, we can simply run the addQuery function to add to the schema's Query type.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
string = appsync.GraphqlType.string()
int = appsync.GraphqlType.int()
api.add_query("allFilms", appsync.ResolvableField(
    return_type=film_connection.attribute(),
    args={"after": string, "first": int, "before": string, "last": int},
    data_source=api.add_none_data_source("none"),
    request_mapping_template=dummy_request,
    response_mapping_template=dummy_response
))

To learn more about top level operations, check out the docs here.

Mutation

Every schema can have a top level Mutation type. By default, the schema will look for the ObjectType named Mutation. The top level Mutation Type is the only exposed type that users can access to perform mutable operations on your Api.

To add fields for these mutations, we can simply run the addMutation function to add to the schema's Mutation type.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
string = appsync.GraphqlType.string()
int = appsync.GraphqlType.int()
api.add_mutation("addFilm", appsync.ResolvableField(
    return_type=film.attribute(),
    args={"name": string, "film_number": int},
    data_source=api.add_none_data_source("none"),
    request_mapping_template=dummy_request,
    response_mapping_template=dummy_response
))

To learn more about top level operations, check out the docs here.

Subscription

Every schema can have a top level Subscription type. The top level Subscription Type is the only exposed type that users can access to invoke a response to a mutation. Subscriptions notify users when a mutation specific mutation is called. This means you can make any data source real time by specify a GraphQL Schema directive on a mutation.

Note: The AWS AppSync client SDK automatically handles subscription connection management.

To add fields for these subscriptions, we can simply run the addSubscription function to add to the schema's Subscription type.

# Example automatically generated without compilation. See https://github.com/aws/jsii/issues/826
api.add_subscription("addedFilm", appsync.ResolvableField(
    return_type=film.attribute(),
    args={"id": appsync.GraphqlType.id(is_required=True)},
    directive=[appsync.Directive.subscribe("addFilm")]
))

To learn more about top level operations, check out the docs here.