@aws-cdk/aws-appsync

The CDK Construct Library for AWS::AppSync

Stats

stars 🌟issues ⚠️updated 🛠created 🐣size 🏋️‍♀️
6,8951,942Jun 23, 2021Oct 4, 2017Minified + gzip package size for @aws-cdk/aws-appsync in KB

Readme

AWS AppSync Construct Library


cfn-resources: Stable

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

cdk-constructs: Experimental

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 This means that while you may use them, you may need to update your source code when upgrading to a newer version of this package.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:

import * as appsync from '@aws-cdk/aws-appsync';
import * as db from '@aws-cdk/aws-dynamodb';

const api = new appsync.GraphqlApi(stack, 'Api', {
  name: 'demo',
  schema: appsync.Schema.fromAsset(join(__dirname, 'schema.graphql')),
  authorizationConfig: {
    defaultAuthorization: {
      authorizationType: appsync.AuthorizationType.IAM
    },
  },
  xrayEnabled: true,
});

const demoTable = new db.Table(stack, 'DemoTable', {
  partitionKey: {
    name: 'id',
    type: db.AttributeType.STRING,
  },
});

const demoDS = api.addDynamoDbDataSource('demoDataSource', demoTable);

// Resolver for the Query "getDemos" that scans the DynamoDb table and returns the entire list.
demoDS.createResolver({
  typeName: 'Query',
  fieldName: 'getDemos',
  requestMappingTemplate: appsync.MappingTemplate.dynamoDbScanTable(),
  responseMappingTemplate: appsync.MappingTemplate.dynamoDbResultList(),
});

// Resolver for the Mutation "addDemo" that puts the item into the DynamoDb table.
demoDS.createResolver({
  typeName: 'Mutation',
  fieldName: 'addDemo',
  requestMappingTemplate: appsync.MappingTemplate.dynamoDbPutItem(
    appsync.PrimaryKey.partition('id').auto(),
    appsync.Values.projecting('input')
  ),
  responseMappingTemplate: appsync.MappingTemplate.dynamoDbResultItem(),
});

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.

// Create username and password secret for DB Cluster
const secret = new rds.DatabaseSecret(stack, 'AuroraSecret', {
  username: 'clusteradmin',
});

// The VPC to place the cluster in
const vpc = new ec2.Vpc(stack, 'AuroraVpc');

// Create the serverless cluster, provide all values needed to customise the database.
const cluster = new rds.ServerlessCluster(stack, 'AuroraCluster', {
  engine: rds.DatabaseClusterEngine.AURORA_MYSQL,
  vpc,
  credentials: { username: 'clusteradmin' },
  clusterIdentifier: 'db-endpoint-test',
  defaultDatabaseName: 'demos',
});

// Build a data source for AppSync to access the database.
const rdsDS = api.addRdsDataSource('rds', cluster, secret, 'demos');

// Set up a resolver for an RDS query.
rdsDS.createResolver({
  typeName: 'Query',
  fieldName: 'getDemosRds',
  requestMappingTemplate: MappingTemplate.fromString(`
  {
    "version": "2018-05-29",
    "statements": [
      "SELECT * FROM demos"
    ]
  }
  `),
  responseMappingTemplate: MappingTemplate.fromString(`
    $util.rds.toJsonObject($ctx.result)
  `),
});

// Set up a resolver for an RDS mutation.
rdsDS.createResolver({
  typeName: 'Mutation',
  fieldName: 'addDemoRds',
  requestMappingTemplate: MappingTemplate.fromString(`
  {
    "version": "2018-05-29",
    "statements": [
      "INSERT INTO demos VALUES (:id, :version)",
      "SELECT * WHERE id = :id"
    ],
    "variableMap": {
      ":id": $util.toJson($util.autoId()),
      ":version": $util.toJson($ctx.args.version)
    }
  }
  `),
  responseMappingTemplate: MappingTemplate.fromString(`
    $util.rds.toJsonObject($ctx.result)
  `),
});

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:

import * as appsync from '@aws-cdk/aws-appsync';

const api = new appsync.GraphqlApi(scope, 'api', {
  name: 'api',
  schema: appsync.Schema.fromFile(join(__dirname, 'schema.graphql')),
});

const httpDs = api.addHttpDataSource(
  'ds',
  'https://states.amazonaws.com',
  {
    name: 'httpDsWithStepF',
    description: 'from appsync to StepFunctions Workflow',
    authorizationConfig: {
      signingRegion: 'us-east-1',
      signingServiceName: 'states'
    }
  }
);

httpDs.createResolver({
  typeName: 'Mutation',
  fieldName: 'callStepFunction',
  requestMappingTemplate: MappingTemplate.fromFile('request.vtl'),
  responseMappingTemplate: MappingTemplate.fromFile('response.vtl')
});

Elasticsearch

AppSync has builtin support for Elasticsearch from domains that are provisioned through your AWS account. You can use AppSync resolvers to perform GraphQL operations such as queries, mutations, and subscriptions.

const user = new User(stack, 'User');
const domain = new es.Domain(stack, 'Domain', {
  version: es.ElasticsearchVersion.V7_1,
  removalPolicy: cdk.RemovalPolicy.DESTROY,
  fineGrainedAccessControl: { masterUserArn: user.userArn },
  encryptionAtRest: { enabled: true },
  nodeToNodeEncryption: true,
  enforceHttps: true,
});

const ds = api.addElasticsearchDataSource('ds', domain);

ds.createResolver({
  typeName: 'Query',
  fieldName: 'getTests',
  requestMappingTemplate: appsync.MappingTemplate.fromString(JSON.stringify({
    version: '2017-02-28',
    operation: 'GET',
    path: '/id/post/_search',
    params: {
      headers: {},
      queryString: {},
      body: { from: 0, size: 50 },
    },
  })),
  responseMappingTemplate: appsync.MappingTemplate.fromString(`[
    #foreach($entry in $context.result.hits.hits)
    #if( $velocityCount > 1 ) , #end
    $utils.toJson($entry.get("_source"))
    #end
  ]`),
});

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.

const api = new 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.

const schema = new appsync.Schema();
schema.addObjectType('demo', {
  definition: { id: appsync.GraphqlType.id() },
});
const api = new appsync.GraphqlApi(stack, 'api', {
  name: 'myApi',
  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.

const api = appsync.GraphqlApi(stack, 'api', {
  name: 'myApi',
  schema: appsync.Schema.fromAsset(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.

const importedApi = appsync.GraphqlApi.fromGraphqlApiAttributes(stack, 'IApi', {
  graphqlApiId: api.apiId,
  graphqlArn: api.arn,
});
importedApi.addDynamoDbDataSource('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:

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.

const role = new iam.Role(stack, 'Role', {
  assumedBy: new iam.ServicePrincipal('lambda.amazonaws.com'),
});
const api = new appsync.GraphqlApi(stack, 'API', {
  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)
// For generic types
api.grantMutation(role, 'updateExample');

// For custom types and granular design
api.grant(role, appsync.IamResource.ofType('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.

const appsyncFunction = new appsync.AppsyncFunction(stack, 'function', {
  name: 'appsync_function',
  api: api,
  dataSource: apiDataSource,
  requestMappingTemplate: appsync.MappingTemplate.fromFile('request.vtl'),
  responseMappingTemplate: appsync.MappingTemplate.fromFile('response.vtl'),
});

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

const pipelineResolver = new appsync.Resolver(stack, 'pipeline', {
  name: 'pipeline_resolver',
  api: api,
  dataSource: apiDataSource,
  requestMappingTemplate: appsync.MappingTemplate.fromFile('beforeRequest.vtl'),
  pipelineConfig: [appsyncFunction],
  responseMappingTemplate: appsync.MappingTemplate.fromFile('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.

import { GraphqlType } from '@aws-cdk/aws-appsync';

export const string = appsync.GraphqlType.string();
export const 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.

const pluralize = require('pluralize');
import * as scalar from './scalar-types.ts';
import * as appsync from '@aws-cdk/aws-appsync';

export const args = {
  after: scalar.string,
  first: scalar.int,
  before: scalar.string,
  last: scalar.int,
};

export const Node = new appsync.InterfaceType('Node', {
  definition: { id: scalar.string }
});
export const FilmNode = new appsync.ObjectType.implementInterface('FilmNode', {
  interfaceTypes: [Node],
  definition: { filmName: scalar.string }
});

export function generateEdgeAndConnection(base: appsync.ObjectType) {
  const edge = new appsync.ObjectType(`${base.name}Edge`, {
    definition: { node: base.attribute(), cursor: scalar.string }
  });
  const connection = new appsync.ObjectType(`${base.name}Connection`, {
    definition: {
      edges: edges.attribute({ isList: true }),
      [pluralize(base.name)]: base.attribute({ isList: true }),
      totalCount: scalar.int,
    }
  });
  return { edge: edge, connection: connection };
}

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

import * as appsync from '@aws-cdk/aws-appsync';
import * as schema from './object-types';

const api = new appsync.GraphqlApi(stack, 'Api', {
  name: 'demo',
});

this.objectTypes = [ schema.Node, schema.Film ];

const filmConnections = schema.generateEdgeAndConnection(schema.Film);

api.addQuery('allFilms', new appsync.ResolvableField({
    returnType: filmConnections.connection.attribute(),
    args: schema.args,
    dataSource: dummyDataSource,
    requestMappingTemplate: dummyRequest,
    responseMappingTemplate: dummyResponse,
  }),
});

this.objectTypes.map((t) => api.addType(t));
Object.keys(filmConnections).forEach((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:

const field = new appsync.Field({
  returnType: appsync.GraphqlType.string(),
  args: {
    argument: appsync.GraphqlType.string(),
  },
});
const type = new 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:

const info = new appsync.ObjectType('Info', {
  definition: {
    node: new appsync.ResolvableField({
      returnType: appsync.GraphqlType.string(),
      args: {
        id: appsync.GraphqlType.string(),
      },
      dataSource: api.addNoneDataSource('none'),
      requestMappingTemplate: dummyRequest,
      responseMappingTemplate: dummyResponse,
    }),
  },
});

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:

const query = new appsync.ObjectType('Query', {
  definition: {
    get: new appsync.ResolvableField({
      returnType: appsync.GraphqlType.string(),
      args: {
        argument: appsync.GraphqlType.string(),
      },
      dataSource: api.addNoneDataSource('none'),
      requestMappingTemplate: dummyRequest,
      responseMappingTemplate: dummyResponse,
    }),
  },
});

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

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.

const node = new appsync.InterfaceType('Node', {
  definition: {
    id: appsync.GraphqlType.string({ isRequired: 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.

    const api = new appsync.GraphqlApi(stack, 'Api', {
      name: 'demo',
    });
    const demo = new appsync.ObjectType('Demo', {
      definition: {
        id: appsync.GraphqlType.string({ isRequired: true }),
        version: appsync.GraphqlType.string({ isRequired: true }),
      },
    });
    
    api.addType(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

    export const required_string = appsync.GraphqlType.string({ isRequired: true });
    

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

    import { required_string } from './scalar-types';
    export const demo = new appsync.ObjectType('Demo', {
      definition: {
        id: required_string,
        version: required_string,
      },
    });
    

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

    import { demo } from './object-types';
    api.addType(demo);
    
  2. Object Types can be created externally from an Interface Type.

    const node = new appsync.InterfaceType('Node', {
      definition: {
        id: appsync.GraphqlType.string({ isRequired: true }),
      },
    });
    const demo = new appsync.ObjectType('Demo', {
      interfaceTypes: [ node ],
      definition: {
        version: appsync.GraphqlType.string({ isRequired: 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:

const episode = new appsync.EnumType('Episode', {
  definition: [
    'NEWHOPE',
    'EMPIRE',
    'JEDI',
  ],
});
api.addType(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:

const review = new appsync.InputType('Review', {
  definition: {
    stars: GraphqlType.int({ isRequired: true }),
    commentary: GraphqlType.string(),
  },
});
api.addType(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:

const string = appsync.GraphqlType.string();
const human = new appsync.ObjectType('Human', { definition: { name: string } });
const droid = new appsync.ObjectType('Droid', { definition: { name: string } });
const starship = new appsync.ObjectType('Starship', { definition: { name: string } }););
const search = new appsync.UnionType('Search', {
  definition: [ human, droid, starship ],
});
api.addType(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.

const string = appsync.GraphqlType.string();
const int = appsync.GraphqlType.int();
api.addQuery('allFilms', new appsync.ResolvableField({
  returnType: filmConnection.attribute(),
  args: { after: string, first: int, before: string, last: int},
  dataSource: api.addNoneDataSource('none'),
  requestMappingTemplate: dummyRequest,
  responseMappingTemplate: dummyResponse,
}));

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.

const string = appsync.GraphqlType.string();
const int = appsync.GraphqlType.int();
api.addMutation('addFilm', new appsync.ResolvableField({
  returnType: film.attribute(),
  args: { name: string, film_number: int },
  dataSource: api.addNoneDataSource('none'),
  requestMappingTemplate: dummyRequest,
  responseMappingTemplate: dummyResponse,
}));

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.

api.addSubscription('addedFilm', new appsync.ResolvableField({
  returnType: film.attribute(),
  args: { id: appsync.GraphqlType.id({ isRequired: true }) },
  directive: [appsync.Directive.subscribe('addFilm')],
}));

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

If you find any bugs or have a feature request, please open an issue on github!

The npm package download data comes from npm's download counts api and package details come from npms.io.