Writing a handler for AWS lambda in Scala can be as easy as...
import io.circe.generic.auto._
import io.github.mkotsur.aws.handler.Lambda._
import io.github.mkotsur.aws.handler.Lambda
import com.amazonaws.services.lambda.runtime.Context
case class Ping(inputMsg: String)
case class Pong(outputMsg: String)
class PingPongHandler extends Lambda[Ping, Pong] {
override def handle(ping: Ping, context: Context) = Right(Pong(ping.inputMsg.reverse))
}The input JSON will be automatically de-serialized into Ping, and the output into Pong. The handle() method is supposed to return Either[Throwable, Pong]: Right if the input was handled correctly, and Left otherwise.
This handler can be used in AWS Lambda as: io.github.mkotsur.example::handle.
Features:
- Return Futures right from the handler!
- JSON (de)serialization of case classes;
- Plain strings are supported too;
- AWS API Gateway proxy integration;
- Uncaught errors are logged with SLF4J and re-thrown.
import io.circe.generic.auto._
import io.github.mkotsur.aws.handler.Lambda._
import io.github.mkotsur.aws.handler.Lambda
import com.amazonaws.services.lambda.runtime.Context
import scala.concurrent.Future
case class Ping(inputMsg: String)
class PingFuturePongHandler extends Lambda[Ping, Future[Int]] {
override def handle(ping: Ping, context: Context) =
Right(Future.successful(ping.inputMsg.length))
}This lambda will accept an empty string, or string with null as an input.
import io.circe.generic.auto._
import io.github.mkotsur.aws.handler.Lambda._
import io.github.mkotsur.aws.handler.Lambda
import com.amazonaws.services.lambda.runtime.Context
class NothingToNothingHandler extends Lambda[None.type, None.type] {
override protected def handle(i: None.type, c: Context) = {
println("Only side effects")
Right(None)
}
}You can write less boilerplate when implementing a handler for API Gateway proxy events by extending Lambda.ApiProxy[I, C, O]. There are three type parameters there. The first one (I) corresponds to the body field of the API Gateway proxy event, the second one (C) corresponds to requestContext field, and the third one – to the body in the response object. More info about how the even looks like here.
import io.circe.generic.auto._
import io.circe.Json
import io.github.mkotsur.aws.handler.Lambda._
import io.github.mkotsur.aws.proxy._
import io.github.mkotsur.aws.handler.Lambda
import com.amazonaws.services.lambda.runtime.Context
import MyProxy._
object MyProxy {
case class MyRequestBody(name: String)
case class MyResponseBody(score: Int)
}
class MyProxy extends Lambda.ApiProxy[MyRequestBody, Json, MyResponseBody] {
override def handle(
i: ApiProxyRequest[MyRequestBody, Json],
c: Context
): Either[Throwable, ApiProxyResponse[MyResponseBody]] =
i.body match {
case Some(MyRequestBody("Bob")) =>
Right(ApiProxyResponse.success(Some(MyResponseBody(100))))
case Some(MyRequestBody("Alice")) =>
Right(ApiProxyResponse.success(Some(MyResponseBody(50))))
case Some(_) =>
Right(ApiProxyResponse(404))
case None =>
Left(new IllegalArgumentException)
}
}
Tip 1: of course, you can also pass a type defined by a case class into the second type parameter. Please check #24 and src/test/scala/io/github/mkotsur/aws/proxy/ProxyRequestTest.scala for an example.
Tip 2: Don't forget that Lambda.ApiProxy is a very thin wrapper around Lambda, so if something in ApiProxyRequest or ApiProxyResponse doesn't work for you - feel free to define your own case classes (and if you believe that the use case is generic enough – consider contributing back to the library).
Feel free to look at src/test/scala for more examples. And of course, contributions to the docs are welcome!
Scala versions supported: 2.11.x, 2.12.x, 2.13.x.
libraryDependencies += "io.github.mkotsur" %% "aws-lambda-scala" % {latest-version}Short answer: they complement each other. Long answer: read this blog post.