AWS SDK for the Swift programming language working on Linux, macOS and iOS.
AWSSDKSwift works on both Linux, macOS and iOS. Version 4 is dependent on swift-nio 2. Libraries/frameworks that are dependent on an earlier version of swift-nio will not work with version 4 of AWSSDKSwift. In this case Version 3 can be used. For example Vapor 3 uses swift-nio 1.13 so you can only use versions 3.x of AWSSDKSwift with Vapor 3. Below is a compatibility table for versions 3 and 4 of AWSSDKSwift.
Version | Swift | MacOS | iOS | Linux | Vapor |
---|---|---|---|---|---|
3.x | 4.2 - | ✓ | Ubuntu 14.04-18.04 | 3.0 | |
4.x | 5.0 - | ✓ | 12.0 - | Ubuntu 14.04-18.04 | 4.0 |
Visit the aws-sdk-swift
documentation to browse the api reference.
AWSSDKSwift uses the Swift Package Manager to manage its code dependencies. To use AWSSDKSwift in your codebase it is recommended you do the same. Add a dependency to the package in your own Package.swift dependencies.
dependencies: [
.package(url: "https://github.com/swift-aws/aws-sdk-swift.git", from: "4.0.0")
],
Then add target dependencies for each of the AWSSDKSwift targets you want to use.
targets: [
.target(name: "MyAWSApp", dependencies: ["S3", "SES", "CloudFront", "ELBV2", "IAM", "Kinesis"]),
]
)
Alternatively if you are using Xcode 11+ you can use the Swift Package integration and add a dependency to AWSSDKSwift through that.
All developers should feel welcome and encouraged to contribute to aws-sdk-swift
.
As contributors and maintainers of this project, and in the interest of fostering an open and welcoming community, we pledge to respect all people who contribute through reporting issues, posting feature requests, updating documentation, submitting pull requests or patches, and other activities.
To contribute a feature or idea to aws-sdk-swift
, submit an issue and fill in the template. If the request is approved, you or one of the members of the community can start working on it.
If you find a bug, please submit a pull request with a failing test case displaying the bug or create an issue.
If you find a security vulnerability, please contact yuki@miketokyo.com and reach out on the #aws channel on the Vapor Discord as soon as possible. We take these matters seriously.
Before using the SDK, you will need AWS credentials to sign all your requests. Credentials can be provided in the following ways.
If you are running your code on an AWS EC2 instance, you can setup an IAM role as the server's Instance Profile to automatically grant credentials via the metadata service.
There are no code changes or configurations to specify in the code, it will automatically pull and use them.
If you are running your code as an AWS ECS container task, you can setup an IAM role for your container task to automatically grant credentials via the metadata service.
There are no code changes or configurations to specify in the code, it will automatically pull and use them.
You can set shared credentials in the home directory for the user running the app
in ~/.aws/credentials,
[default]
aws_access_key_id = YOUR_AWS_ACCESS_KEY_ID
aws_secret_access_key = YOUR_AWS_SECRET_ACCESS_KEY
Alternatively, you can set the following environment variables:
AWS_ACCESS_KEY_ID=YOUR_AWS_ACCESS_KEY_ID
AWS_SECRET_ACCESS_KEY=YOUR_AWS_SECRET_ACCESS_KEY
All of the AWS Services's initializers accept accessKeyId
and secretAccessKey
let ec2 = EC2(
accessKeyId: "YOUR_AWS_ACCESS_KEY_ID",
secretAccessKey: "YOUR_AWS_SECRET_ACCESS_KEY"
)
Some services like CognitoIdentityProvider don't require credentials to access some of their functions. Explicitly set accessKeyId
and secretAccessKey
to "". This will disable all other credential access functions and send requests unsigned.
AWS Swift Modules can be imported into any swift project. Each module provides a struct that can be initialized, with instance methods to call aws services. See documentation for details on specific services.
The underlying aws-sdk-swift httpclient returns a swift-nio EventLoopFuture object. An EventLoopFuture is not the response, but rather a container object that will be populated with the response sometime later. In this manner calls to AWS do not block the main thread.
The recommended manner to interact with futures is chaining. The following function returns an EventLoopFuture that creates an S3 bucket, puts a file in the bucket, reads the file back from the bucket and finally prints the contents of the file. Each of these operations are chained together. The output of one being the input of the next.
import S3 //ensure this module is specified as a dependency in your package.swift
let bucket = "my-bucket"
let s3 = S3(accessKeyId: "Your-Access-Key", secretAccessKey: "Your-Secret-Key", region: .uswest2)
func createBucketPutGetObject() -> EventLoopFuture<S3.GetObjectOutput> {
// Create Bucket, Put an Object, Get the Object
let createBucketRequest = S3.CreateBucketRequest(bucket: bucket)
s3.createBucket(createBucketRequest)
.flatMap { response -> Future<S3.PutObjectOutput> in
// Upload text file to the s3
let bodyData = "hello world".data(using: .utf8)!
let putObjectRequest = S3.PutObjectRequest(acl: .publicRead, body: bodyData, bucket: bucket, contentLength: Int64(bodyData.count), key: "hello.txt")
return s3.putObject(putObjectRequest)
}
.flatMap { response -> Future<S3.GetObjectOutput> in
let getObjectRequest = S3.GetObjectRequest(bucket: bucket, key: "hello.txt")
return s3.getObject(getObjectRequest)
}
.whenSuccess { response in
if let body = response.body {
print(String(data: body, encoding: .utf8)!)
}
}
}
The simplest way to upgrade from an existing 1.0 or 2.0 implementation is to call .wait()
on existing synchronous calls. However it is recommend to rewrite your synchronous code to work with the returned future objects. It is no longer necessary to use a DispatchQueue.
The AWS SDK has its own EventLoopGroup
but it is recommended that you provide your own EventLoopGroup
for the SDK to work off. You can do this when you construct your client.
let s3 = S3(region:.uswest2, eventLoopGroupProvider: .shared(myEventLoopGroup)
The EventLoopGroup types you can use depend on the platform you are running on. On Linux use MultiThreadedEventLoopGroup
, on macOS use MultiThreadedEventLoopGroup
or NIOTSEventLoopGroup
and iOS use NIOTSEventLoopGroup
. Using the NIOTSEventLoopGroup
will mean you use NIO Transport Services and the Apple Network framework.
Integration with Vapor is pretty straight forward. Although be sure you use the correct version of AWSSDKSwift depending on which version of Vapor you are using. See the compatibility section for details. Below is a simple Vapor 3 example that extracts an email address, subject and message from a request and then sends an email using these details. Take note of the hopTo(eventLoop:)
call. If your AWS SDK is not working off the same EventLoopGroup
as the Vapor Request
this is a requirement.
import Vapor
import HTTP
import SES
final class MyController {
struct EmailData: Content {
let address: String
let subject: String
let message: String
}
func sendUserEmailFromJSON(_ req: Request) throws -> Future<HTTPStatus> {
return try req.content.decode(EmailData.self)
.flatMap { (emailData)->EventLoopFuture<SES.SendEmailResponse> in
let client = SES(region: .uswest1)
let destination = SES.Destination(toAddresses: [emailData.address])
let message = SES.Message(body:SES.Body(text:SES.Content(data:emailData.message)), subject:SES.Content(data:emailData.subject))
let sendEmailRequest = SES.SendEmailRequest(destination: destination, message: message, source:"awssdkswift@me.com")
return client.sendEmail(sendEmailRequest)
}
.hopTo(eventLoop: req.eventLoop)
.map { response -> HTTPResponseStatus in
return HTTPStatus.ok
}
}
}
By specifying only those modules necessary for your application, only those modules will compile which makes for fast compilation.
If you want to create a module for your service, you can try using the module-exporter to build a separate repo for any of the modules.
aws-sdk-swift
is released under the Apache License, Version 2.0. See LICENSE for details.