cdk-appsync-transformer
TypeScript icon, indicating that this package has built-in type declarations

1.138.3 • Public • Published

AppSync Transformer Construct for AWS CDK

build codecov dependencies Status npm

npm version PyPI version

Notice

For CDK versions < 1.64.0 please use aws-cdk-appsync-transformer.

Why This Package

In April 2020 I wrote a blog post on using the AWS Cloud Development Kit with AppSync. I wrote my own transformer in order to emulate AWS Amplify's method of using GraphQL directives in order to template a lot of the Schema Definition Language.

This package is my attempt to convert all of that effort into a separate construct in order to clean up the process.

How Do I Use It

Example Usage

API With Default Values

import { AppSyncTransformer } from 'cdk-appsync-transformer';
...
new AppSyncTransformer(this, "my-cool-api", {
    schemaPath: 'schema.graphql'
});

schema.graphql

type Customer
  @model
  @auth(
    rules: [
      { allow: groups, groups: ["Admins"] }
      { allow: private, provider: iam, operations: [read, update] }
    ]
  ) {
  id: ID!
  firstName: String!
  lastName: String!
  active: Boolean!
  address: String!
}

type Product
  @model
  @auth(
    rules: [
      { allow: groups, groups: ["Admins"] }
      { allow: public, provider: iam, operations: [read] }
    ]
  ) {
  id: ID!
  name: String!
  description: String!
  price: String!
  active: Boolean!
  added: AWSDateTime!
  orders: [Order] @connection
}

type Order @model @key(fields: ["id", "productID"]) {
  id: ID!
  productID: ID!
  total: String!
  ordered: AWSDateTime!
}

Supported Amplify Directives

Tested:

Experimental:

Not Yet Supported:

Custom Transformers & Directives

This is an advanced feature

It is possible to add pre/post custom transformers that extend the Amplify ITransformer. To see a simple example please look at mapped-transformer.ts in the tests section.

This allows you to modify the data either before or after the cdk-transformer is run.

Limitation: Due to some limitations with jsii we are unable to export the ITransformer interface from graphql-transformer-core to ensure complete type safety. Instead, there is a validation method that will check for name, directive and typeDefinitions members in the transformers that are passed in.

import { PreTransformer, PostTransformer } from "./customTransformers";
new AppSyncTransformer(this, "my-cool-api", {
  schemaPath: "schema.graphql",
  preCdkTransformers: [new PreTransformer()],
  postCdkTransformers: [new PostTransformer()],
});

Custom VTL Transformer

Can be used to create custom NONE datasource resolvers.This allows for custom or special logic to be used and added via a transformer.

Example:

type Thing {
  fooBar: String
}

type Query {
  listThingCustom: Thing
    @custom(request: "test/custom-resolvers/Test/request.vtl", response: "test/custom-resolvers/Test/response.vtl")
}

The above will generate a Query.listThingCustom request and response resolver. You can customize the location of custom resolvers using the customVtlTransformerRootDirectory property.

Overriding generated vtl

This is an advanced feature

You can override generated request and response mapping templates using the overrideResolver convenience method.

const appsyncTransformer = new AppSyncTransformer(this, "my-cool-api", {
  schemaPath: "schema.graphql",
});

// You can override the just the request, just the response, or BOTH
appsyncTransformer.overrideResolver({
  typeName: 'Query',
  fieldName: 'listThings',
  requestMappingTemplateFile: path.join(process.cwd(), 'custom-resolvers', 'Things', 'request.vtl'),
  responseMappingTemplateFile: path.join(process.cwd(), 'custom-resolvers', 'Things', 'response.vtl'),
});

Authentication

User Pool Authentication

const userPool = new UserPool(this, 'my-cool-user-pool', {
    ...
})
...
const userPoolClient = new UserPoolClient(this, `${id}-client`, {
    userPool: this.userPool,
    ...
})
...
new AppSyncTransformer(this, "my-cool-api", {
    schemaPath: 'schema.graphql',
    authorizationConfig: {
        defaultAuthorization: {
            authorizationType: AuthorizationType.USER_POOL,
            userPoolConfig: {
                userPool: userPool,
                appIdClientRegex: userPoolClient.userPoolClientId,
                defaultAction: UserPoolDefaultAction.ALLOW
            }
        }
    }
});

IAM

Unauth Role

You can grant access to the public policies generated from the @auth transformer by using appsyncTransformer.grantPublic(...). In the example below you can see we give public iam read access for the Product type. This will generate permissions for listProducts, getProduct and Product (to get all the fields). We then attach it to our publicRole using the grantPublic method.

Example:

type Product
    @model
    @auth(rules: [
        { allow: groups, groups: ["Admins"] },
        { allow: public, provider: iam, operations: [read] }
    ])
    @key(name: "productsByName", fields: ["name", "added"], queryField: "productsByName") {
        id: ID!
        name: String!
        description: String!
        price: String!
        active: Boolean!
        added: AWSDateTime!
        orders: [Order] @connection
}
const identityPool = new CfnIdentityPool(stack, 'test-identity-pool', {
    identityPoolName: 'test-identity-pool',
    cognitoIdentityProviders: [
      {
        clientId: userPoolClient.userPoolClientId,
        providerName: `cognito-idp.${stack.region}.amazonaws.com/${userPool.userPoolId}`,
      },
    ],
    allowUnauthenticatedIdentities: true,
  });

const publicRole = new Role(stack, 'public-role', {
  assumedBy: new WebIdentityPrincipal('cognito-identity.amazonaws.com')
    .withConditions({
      'StringEquals': { 'cognito-identity.amazonaws.com:aud': `${identityPool.ref}` },
      'ForAnyValue:StringLike': { 'cognito-identity.amazonaws.com:amr': 'unauthenticated' },
    }),
});

appSyncTransformer.grantPublic(publicRole);
Auth Role

You can grant access to the private policies generated from the @auth transformer by using appsyncTransformer.grantPrivate(...). In the example below you can see we give private iam read and update access for the Customer type. This will generate permissions for listCustomers, getCustomer, updateCustomer and Customer (to get all the fields). We then attach it to our privateFunction using the grantPrivate method. You could also use an identity pool as in the unauth example above, I just wanted to show a varied range of use

type Customer 
    @model
    @auth(rules: [
        { allow: groups, groups: ["Admins"] },
        { allow: private, provider: iam, operations: [read, update] }
    ]) {
        id: ID!
        firstName: String!
        lastName: String!
        active: Boolean!
        address: String!
}
const privateFunction = new Function(stack, 'test-function', {
  runtime: Runtime.NODEJS_12_X,
  code: Code.fromInline('export function handler() { }'),
  handler: 'handler',
});

appSyncTransformer.grantPrivate(privateFunction);

Functions

Directive Example

type Query {
  listUsers: UserConnection @function(name: "myFunction")
  getUser(id: ID!): User @function(name: "myFunction")
}

There are two ways to add functions as data sources (and their resolvers)

Construct Convenience Method

const myFunction = new Function(...);

// first argument is the name in the @function directive
appsyncTransformer.addLambdaDataSourceAndResolvers('myFunction', 'unique-id', myFunction, {
  name: 'lambdaDatasourceName'
})

addLambdaDataSourceAndResolvers does the same thing as the manual version below. However, if you want to customize mapping templates you will have to bypass this and set up the data source and resolvers yourself

Manually

Fields with the @function directive will be accessible via appsyncTransformer.functionResolvers. It will return a map like so:

{
  'user-function': [
    { typeName: 'Query', fieldName: 'listUsers' },
    { typeName: 'Query', fieldName: 'getUser' },
    { typeName: 'Mutation', fieldName: 'createUser' },
    { typeName: 'Mutation', fieldName: 'updateUser' }
  ]
}

You can grab your function resolvers via the map and assign them your own function(s). Example might be something like:

const userFunction = new Function(...);
const userFunctionDataSource = appsyncTransformer.appsyncAPI.addLambdaDataSource('some-id', userFunction);

const dataSourceMap = {
  'user-function': userFunctionDataSource
};

for (const [functionName, resolver] of Object.entries(appsyncTransformer.functionResolvers)) {
  const dataSource = dataSourceMap[functionName];
  new Resolver(this.nestedAppsyncStack, `${resolver.typeName}-${resolver.fieldName}-resolver`, {
    api: appsyncTransformer.appsyncAPI,
    typeName: resolver.typeName,
    fieldName: resolver.fieldName,
    dataSource: dataSource,
    requestMappingTemplate: resolver.defaultRequestMappingTemplate,
    responseMappingTemplate: resolver.defaultResponseMappingTemplate // This defaults to allow errors to return to the client instead of throwing
  });
}

Table Name Map

Often you will need to access your table names in a lambda function or elsewhere. The cdk-appsync-transformer will return these values as a map of table names to cdk tokens. These tokens will be resolved at deploy time. They can be accessed via appSyncTransformer.tableNameMap.

{
  CustomerTable: '${Token[TOKEN.1300]}',
  ProductTable: '${Token[TOKEN.1346]}',
  OrderTable: '${Token[TOKEN.1392]}',
  BlogTable: '${Token[TOKEN.1442]}',
  PostTable: '${Token[TOKEN.1492]}',
  CommentTable: '${Token[TOKEN.1546]}',
  UserTable: '${Token[TOKEN.1596]}'
}

Table Map

You may need to access your dynamo table L2 constructs. These can be accessed via appSyncTransformer.tableMap.

Custom Table Names

If you do not like autogenerated names for your Dynamo tables you can pass in props to specify them. Use the tableKey value derived from the @model directive. Example, if you have type Foo @model you would use FooTable as the key value.

const appSyncTransformer = new AppSyncTransformer(stack, 'test-transformer', {
  schemaPath: testSchemaPath,
  tableNames: {
    CustomerTable: customerTableName,
    OrderTable: orderTableName
  },
});

DynamoDB Streams

There are two ways to enable DynamoDB streams for a table. The first version is probably most preferred. You pass in the @model type name and the StreamViewType as properties when creating the AppSyncTransformer. This will also allow you to access the tableStreamArn property of the L2 table construct from the tableMap.

const appSyncTransformer = new AppSyncTransformer(stack, 'test-transformer', {
  schemaPath: testSchemaPath,
  dynamoDbStreamConfig: {
    Order: StreamViewType.NEW_IMAGE,
    Blog: StreamViewType.NEW_AND_OLD_IMAGES
  }
});

const orderTable = appSyncTransformer.tableMap.OrderTable;
// Do something with the table stream arn - orderTable.tableStreamArn

A convenience method is also available. It returns the stream arn because the L2 Table construct does not seem to update with the value since we are updating the underlying CfnTable. Normally a Table construct must pass in the stream specification as a prop

const streamArn = appSyncTransformer.addDynamoDBStream({
  modelTypeName: 'Order',
  streamViewType: StreamViewType.NEW_IMAGE,
});

// Do something with the streamArn

DataStore Support

  1. Pass syncEnabled: true to the AppSyncTransformerProps
  2. Generate necessary exports (see Code Generation below)

Cfn Outputs

  • appsyncGraphQLEndpointOutput - the appsync graphql endpoint

Code Generation

I've written some helpers to generate code similarly to how AWS Amplify generates statements and types. You can find the code here.

Versioning

I will attempt to align the major and minor version of this package with [AWS CDK], but always check the release descriptions for compatibility.

I currently support GitHub package.json dependency version (prod)

Contributing

See CONTRIBUTING for details

License

Distributed under Apache License, Version 2.0

References

Sponsors

Stedi

Readme

Keywords

Package Sidebar

Install

npm i cdk-appsync-transformer

Weekly Downloads

822

Version

1.138.3

License

Apache-2.0

Unpacked Size

81.2 MB

Total Files

6557

Last publish

Collaborators

  • kwinner