Integration with Cats Effect

Cats Effect is a library for writing side-effectful programs in a pure functional style.

The module decline-effect defines a thin integration between decline and Cats Effect. In particular, CommandIOApp combines the simple and rich CLI from decline’s CommandApp and the pure effect management of cats-effect’s IOApp

but instead of using the CommandApp, we are going to use a newly defined CommandIOApp.

In the following lines we are going to show how to do this by following an example.

Building an IO-based application

As an example, we’ll reimplement a small Docker-like command-line interface: it’s fairly well known CLI tool, and has a nice mix of options, flags, arguments and subcommands. We’ll focus only on the ps and build commands – just enough to get the point across.

First, we’ll add the module to our dependencies:

libraryDependencies += "com.monovore" %% "decline-effect" % "2.4.2-SNAPSHOT"

And add the necessary imports:

import cats.effect._
import cats.implicits._

import com.monovore.decline._
import com.monovore.decline.effect._

Defining the command line interface

Let’s now define our interface as a data type. We’re aiming for the following very-simplified Docker-like interface:

$ docker ps --help
Usage: docker ps [--all]

Lists docker processes running!

    --all
            Whether to show all running processes.
    --help
            Display this help text.
$ docker build --help
Usage: docker build [--file <name>] path

Builds a docker image!

    --file <name>
            The name of the Dockerfile.
    --help
            Display this help text.

If we’re translating that interface into data types, we’ll end up with something like the following:

case class ShowProcesses(all: Boolean)
case class BuildImage(dockerFile: Option[String], path: String)

Now we’ll build our parser, composing the individual elements for each of the components. Here’s the ps subcommand:

val showProcessesOpts: Opts[ShowProcesses] =
  Opts.subcommand("ps", "Lists docker processes running!") {
    Opts.flag("all", "Whether to show all running processes.", short = "a")
      .orFalse
      .map(ShowProcesses)
  }
// showProcessesOpts: Opts[ShowProcesses] = Opts(ps)

And the build command would be as follows:

val dockerFileOpts: Opts[Option[String]] =
  Opts.option[String]( "file", "The name of the Dockerfile.", short = "f" ).orNone
// dockerFileOpts: Opts[Option[String]] = Opts([--file <string>])

val pathOpts: Opts[String] =
  Opts.argument[String](metavar = "path")
// pathOpts: Opts[String] = Opts(<path>)

val buildOpts: Opts[BuildImage] =
  Opts.subcommand("build", "Builds a docker image!") {
    (dockerFileOpts, pathOpts).mapN(BuildImage)
  }
// buildOpts: Opts[BuildImage] = Opts(build)

Interpreting our command line interface

Now we’ll build an interpreter for the data type we just created. This could be done using the CommandIOApp as follows:

object DockerApp extends CommandIOApp(
  name = "docker",
  header = "Faux docker command line",
  version = "0.0.x"
) {

  override def main: Opts[IO[ExitCode]] =
    (showProcessesOpts orElse buildOpts).map {
      case ShowProcesses(all) => ???
      case BuildImage(dockerFile, path) => ???
    }
}

The main: Opts[IO[ExitCode]] is what aggregates all the bits and pieces of our command line interpreter. In this case, we just take the previously-defined subcommand options, and map into IO actions that correspond to the given command line arguments. (It’s usually handy to define these actions within the CommandIOApp itself… it puts a ContextShift and Timer in implicit scope, which are required by lots of other code in the Cats Effect ecosystem.)