How to create a task that takes an argument in sbt

Something that comes up when you're coming to an end of a project is the simple question of how to deploy it. For the hobbyist, deploying might be as simple as installing libraries or services to a machine and sftp-ing the contents of their project up to a server. For a professional, it may involve setting up builds on Jenkin's servers and deploy key's on github, then creating a deployed and auto-scaling environment on something like Amazon or Rackspace.

If you're using SBT it's natural that to deploy something with a build in jenkins you would use something like SBT Assembly or SBT Native Packager. If you're also using the PlayFramework you're likely familiar with the dist task which creates a deployment for your application. No matter what you use to create your final application, if you're doing an enterprise launch of some kind, you'll likely have more than one environment. A possible setup might include:

  1. A development environment to test feature branches on

  2. An integration environment to ensure features work with each other

  3. A production environment that is live.

With those in mind, your build task will likely use different configuration files, and so you'll want to change that for each environment. But when making an sbt task, how do you do that? It's not too hard if you read the documentation on Input Tasks and on Parsers for a few minutes. Here's the long and skinny:

  1. Imports go at the top of your build.sbt file. 

import sbt.complete._
import complete.DefaultParsers._

  1. Next, define your task as inputKey since it takes input: 

val myTask = inputKey[Unit]("This task takes a parameter!")

  1. Define the input to your task, in our example let's say we'll have the environment as the variable: 

val stageEnv: Parser[String] = " staging" 
val intEnv : Parser[String] = " integration" 
val prodEnv : Parser[String] = " production"
val combinedParser: Parser[String] = stageEnv | intEnv | prodEnv

  1. Define what your task is going to do! 

myTask := {
  val environment = combinedParser.parsed.trim
  val s = streams.value
  s.log.info(s"Parameter was $environment")
}

  1. If neccesary, have dependent tasks run first. For example, to run myTask after the dist task in play try: 

myTask <<= myTask.dependsOn(dist in Universal)

  1. Run sbt and myTask <environment> where environment is staging, integration or production.

And that's the step by step process. I find myself learning from simple examples like these much better than other ways. The documentation on combining parsers states you can do things like 

val color: Parser[String] = "blue" | "green"

But I found that sbt complained when doing so. The other thing to notice here is that I trimed the output of the parsed environment. Why? Because it has an extra space in the parser see? The trick with the regular Parser[String] is that it takes the literal string.

This is all I wanted to cover in this blog post, but depending on interest, I might write up a small series on going from small examples to bigger ones with SBT, or just provide some simple templates or examples of SBT usage that I've seen come in handy.