Deploying to AWS Lambda using the Serverless Application Model (SAM)

This guide illustrates how to deploy a Server-Side Swift workload on AWS using the AWS Serverless Application Model (SAM) toolkit. The workload is a REST API for tracking a To Do List. It deploys the API using Amazon API Gateway. The API methods store and retrieve data in a Amazon DynamoDB database using AWS Lambda functions.

Architecture

Architecture

  • Amazon API Gateway receives API requests
  • API Gateway invokes Lambda functions to process PUT and GET events
  • Lambda functions use the AWS SDK for Swift and the Swift AWS Lambda Runtime to retrieve and save items to the database

Prerequisites

To build this sample application, you need:

Step 1: Create a new SAM project

The SAM project creates resources (Lambda functions, API Gateway, and DynamoDB table) in your AWS account. You define the resources in a YAML template.

Create a folder for your project and a new template.yml file.

mkdir swift-lambda-api && cd swift-lambda-api
touch template.yml

Open the template.yml file and add the following code. Review the comments in the code to determine what it created in each section.

AWSTemplateFormatVersion: '2010-09-09'
Transform: AWS::Serverless-2016-10-31

Resources:
  # DynamoDB table to store your data
  SwiftAPITable:
    Type: AWS::Serverless::SimpleTable
    Properties:
      PrimaryKey:
        Name: id
        Type: String

  # Lambda function to put items to the database
  PutItemFunction:
    Type: AWS::Serverless::Function
    Properties:
      # package the function as a Docker image
      PackageType: Image
      Policies:
        # allow function to read and write to database table
        - DynamoDBCrudPolicy:
            TableName: !Ref SwiftAPITable
      Environment:
        # store database table name as an environment variable
        Variables:
          TABLE_NAME: !Ref SwiftAPITable
      Events:
        # handles the POST /item method of the REST API
        Api:
          Type: HttpApi
          Properties:
            Method: post
            Path: /item
    Metadata:
      # location of the code and Docker file for function
      DockerContext: ./src/put-item
      Dockerfile: Dockerfile
      DockerBuildArgs:
        TARGET_NAME: put-item

  # Lambda function to retrieve items from database
  GetItemsFunction:
    Type: AWS::Serverless::Function
    Properties:
      # package the function as a Docker image
      PackageType: Image
      Policies:
        # allow function to read and write to database table
        - DynamoDBCrudPolicy:
            TableName: !Ref SwiftAPITable
      Environment:
        # store database table name as an environment variable
        Variables:
          TABLE_NAME: !Ref SwiftAPITable
      Events:
        # handles the GET /items method of the REST API
        Api:
          Type: HttpApi
          Properties:
            Method: get
            Path: /items
    Metadata:
      # location of the code and Docker file for function
      DockerContext: ./src/get-items
      Dockerfile: Dockerfile
      DockerBuildArgs:
        TARGET_NAME: get-items

# print API endpoint and name of database table
Outputs:
  SwiftAPIEndpoint:
    Description: "API Gateway endpoint URL for your application"
    Value: !Sub "https://${ServerlessHttpApi}.execute-api.${AWS::Region}.amazonaws.com"
  SwiftAPITable:
    Description: "DynamoDB Table Name"
    Value: !Ref SwiftAPITable

Step 2: Initialize Lambda functions with SwiftPM

Lambda functions, written in Swift, process the API events. The PutItem function processes POST requests to add items to the database. The GetItems function processes GET requests to retrieve items from the database.

Use the Swift Package Manager to initialize a project for each function. You also add a Dockerfile to each folder.

mkdir -p src/put-item
cd src/put-item
swift package init --type executable
touch Dockerfile

cd ../..

mkdir -p src/get-items
cd src/get-items
swift package init --type executable
touch Dockerfile

Step 3: Update the Dockerfile

Docker is used to compile your Swift code and deploy the image to Lambda. Copy the following code into the Dockerfile you created in each function’s folder.

# image used to compile your Swift code
FROM --platform=linux/amd64 public.ecr.aws/docker/library/swift:5.7.2-amazonlinux2 as builder

ARG TARGET_NAME

RUN yum -y install git jq tar zip openssl-devel
WORKDIR /build-lambda
RUN mkdir -p /Sources/$TARGET_NAME/
RUN mkdir -p /Tests/$TARGET_NAME/
ADD /Sources/ ./Sources/
ADD /Tests/ ./Tests/
COPY Package.swift .
RUN cd /build-lambda && swift package clean && swift build --static-swift-stdlib -c release

# image deplpoyed to AWS Lambda with your compiled executable
FROM public.ecr.aws/lambda/provided:al2-x86_64

ARG TARGET_NAME

RUN mkdir -p /var/task/
RUN mkdir -p /var/runtime/
COPY --from=builder /build-lambda/.build/release/$TARGET_NAME /var/task/lambdaExec
RUN chmod 755 /var/task/lambdaExec
RUN ln -s /var/task/lambdaExec /var/runtime/bootstrap
RUN chmod 755 /var/runtime/bootstrap
WORKDIR /var/task
CMD ["/var/task/lambdaExec"]

Step 4: Update the Swift dependencies

Your project requires 3 libraries.

  • swift-aws-lambda-runtime
  • swift-aws-lambda-events
  • aws-sdk-swift

You define these in the Package.swift file. Replace the contents of the Package.swift file in each function’s folder with the following code.

src/put-item/Sources/put-item/Package.swift

// swift-tools-version: 5.7
// The swift-tools-version declares the minimum version of Swift required to build this package.

import PackageDescription

let package = Package(
    name: "put-item",
    platforms: [.macOS(.v12)],
    dependencies: [
        .package(url: "https://github.com/swift-server/swift-aws-lambda-runtime", branch: "main"),
        .package(url: "https://github.com/swift-server/swift-aws-lambda-events", branch: "main"),
        .package(url: "https://github.com/awslabs/aws-sdk-swift", from: "0.9.1")
    ],
    targets: [
        .executableTarget(
            name: "put-item",
            dependencies: [
                .product(name: "AWSLambdaRuntime",package: "swift-aws-lambda-runtime"),
                .product(name: "AWSLambdaEvents", package: "swift-aws-lambda-events"),
                .product(name: "AWSDynamoDB", package: "aws-sdk-swift")
            ]),
        .testTarget(
            name: "put-itemTests",
            dependencies: ["put-item"]),
    ]
)

src/get-items/Sources/get-items/Package.swift

// swift-tools-version: 5.7
// The swift-tools-version declares the minimum version of Swift required to build this package.

import PackageDescription

let package = Package(
    name: "get-items",
    platforms: [.macOS(.v12)],
    dependencies: [
        .package(url: "https://github.com/swift-server/swift-aws-lambda-runtime", branch: "main"),
        .package(url: "https://github.com/swift-server/swift-aws-lambda-events", branch: "main"),
        .package(url: "https://github.com/awslabs/aws-sdk-swift", from: "0.9.1")
    ],
    targets: [
        .executableTarget(
            name: "get-items",
            dependencies: [
                .product(name: "AWSLambdaRuntime",package: "swift-aws-lambda-runtime"),
                .product(name: "AWSLambdaEvents", package: "swift-aws-lambda-events"),
                .product(name: "AWSDynamoDB", package: "aws-sdk-swift")
            ]),
        .testTarget(
            name: "get-itemsTests",
            dependencies: ["get-items"]),
    ]
)

Step 5: Update the Lambda function source code

Replace the contents of the main code file for each Swift project with the following code.

src/put-item/Sources/put-item/put_item.swift

// import the packages required by our function
import Foundation
import AWSLambdaRuntime
import AWSLambdaEvents
import AWSDynamoDB

// define Codable struct for function response
struct Item : Codable {
    var id: String?
    let itemName: String
}

enum FunctionError: Error {
    case envError
}

@main
struct PutItemFunction: SimpleLambdaHandler {

    // Lambda Function handler
    func handle(_ event: APIGatewayV2Request, context: LambdaContext) async throws -> Item {
        
        print("event received:\(event)")
        
        // create a client to interact with DynamoDB
        let client = try await DynamoDBClient()

        // obtain DynamoDB table name from function's environment variables
        guard let tableName = ProcessInfo.processInfo.environment["TABLE_NAME"] else {
            throw FunctionError.envError
        }

        // decode data from APIGateway POST into a codable struct
        var item = try JSONDecoder().decode(
            Item.self, 
            from: event.body!.data(using: .utf8)!
        )

        // generate a unique id for the key of the item
        item.id = UUID().uuidString

        // use SDK to put the item into the database and return the item with key value
        let input = PutItemInput(item: ["id": .s(item.id!), "itemName": .s(item.itemName)], tableName: tableName)

        _ = try await client.putItem(input: input)

        return item
    }
}

src/get-items/Sources/get_items/get_items.swift

// import the packages required by our function
import Foundation
import AWSLambdaRuntime
import AWSLambdaEvents
import AWSDynamoDB

// define Codable struct for function response
struct Item : Codable {
    var id: String = ""
    var itemName: String = ""
}

enum FunctionError: Error {
    case envError
}

@main
struct GetItemsFunction: SimpleLambdaHandler {

    // Lambda Function handler
    func handle(_ event: APIGatewayV2Request, context: LambdaContext) async throws -> [Item] {
        
        print("event received:\(event)")
        
        // create a client to interact with DynamoDB
        let client = try await DynamoDBClient()

        // obtain DynamoDB table name from function's environment variables
        guard let tableName = ProcessInfo.processInfo.environment["TABLE_NAME"] else {
            throw FunctionError.envError
        }

        // use SDK to retrieve items from table
        let input = ScanInput(tableName: tableName)
        let response = try await client.scan(input: input)
        
        // return items in an array
        return response.items!.map() {i in
            var item = Item()

            if case .s(let value) = i["id"] {
                item.id = value
            }

            if case .s(let value) = i["itemName"] {
                item.itemName = value
            }

            return item
        }
    }
}

Step 6: Build the SAM project

Building your SAM project uses Docker on your machine to compile your Swift code into Docker images. From the root folder of your project (swift-lambda-api) run the following command.

sam build

Step 7: Deploy the SAM project

Deploying your SAM project creates the Lambda functions, API Gateway, and DynamoDB database in your AWS account.

sam deploy --guided

Accept the default response to every prompt, except the following two:

PutItemFunction may not have authorization defined, Is this okay? [y/N]: y
GetItemsFunction may not have authorization defined, Is this okay? [y/N]: y

The project creates a publicly accessible API endpoint. These are warnings to inform you the API does not have authorization. If you are interested in adding authorization to the API, please refer to the SAM Documentation.

Step 8: Use your API

At the end of deployment, SAM displays the endpoint of your API Gateway:

Outputs
----------------------------------------------------------------------------------------
Key                 SwiftAPIEndpoint
Description         API Gateway endpoint URL for your application
Value               https://[your-api-id].execute-api.[your-aws-region].amazonaws.com
----------------------------------------------------------------------------------------

Use cURL or a tool such as Postman to interact with your API. Replace [your-api-endpoint] with the SwiftAPIEndpoint value from the deployment output.

Add a To Do List item

curl --request POST 'https://[your-api-endpoint]/item' --header 'Content-Type: application/json' --data-raw '{"itemName": "my todo item"}'

Retrieve To Do List items

curl https://[your-api-endpoint]/items

Cleanup

When finished with your application, use SAM to delete it from your AWS account. Answer Yes (y) to all prompts.

sam delete