[scala] Best way to parse command-line parameters?

What's the best way to parse command-line parameters in Scala? I personally prefer something lightweight that does not require external jar.

Related:

This question is related to scala command-line command-line-arguments

The answer is


freecli

package freecli
package examples
package command

import java.io.File

import freecli.core.all._
import freecli.config.all._
import freecli.command.all._

object Git extends App {

  case class CommitConfig(all: Boolean, message: String)
  val commitCommand =
    cmd("commit") {
      takesG[CommitConfig] {
        O.help --"help" ::
        flag --"all" -'a' -~ des("Add changes from all known files") ::
        O.string -'m' -~ req -~ des("Commit message")
      } ::
      runs[CommitConfig] { config =>
        if (config.all) {
          println(s"Commited all ${config.message}!")
        } else {
          println(s"Commited ${config.message}!")
        }
      }
    }

  val rmCommand =
    cmd("rm") {
      takesG[File] {
        O.help --"help" ::
        file -~ des("File to remove from git")
      } ::
      runs[File] { f =>
        println(s"Removed file ${f.getAbsolutePath} from git")
      }
    }

  val remoteCommand =
   cmd("remote") {
     takes(O.help --"help") ::
     cmd("add") {
       takesT {
         O.help --"help" ::
         string -~ des("Remote name") ::
         string -~ des("Remote url")
       } ::
       runs[(String, String)] {
         case (s, u) => println(s"Remote $s $u added")
       }
     } ::
     cmd("rm") {
       takesG[String] {
         O.help --"help" ::
         string -~ des("Remote name")
       } ::
       runs[String] { s =>
         println(s"Remote $s removed")
       }
     }
   }

  val git =
    cmd("git", des("Version control system")) {
      takes(help --"help" :: version --"version" -~ value("v1.0")) ::
      commitCommand ::
      rmCommand ::
      remoteCommand
    }

  val res = runCommandOrFail(git)(args).run
}

This will generate the following usage:

Usage


I'm going to pile on. I solved this with a simple line of code. My command line arguments look like this:

input--hdfs:/path/to/myData/part-00199.avro output--hdfs:/path/toWrite/Data fileFormat--avro option1--5

This creates an array via Scala's native command line functionality (from either App or a main method):

Array("input--hdfs:/path/to/myData/part-00199.avro", "output--hdfs:/path/toWrite/Data","fileFormat--avro","option1--5")

I can then use this line to parse out the default args array:

val nArgs = args.map(x=>x.split("--")).map(y=>(y(0),y(1))).toMap

Which creates a map with names associated with the command line values:

Map(input -> hdfs:/path/to/myData/part-00199.avro, output -> hdfs:/path/toWrite/Data, fileFormat -> avro, option1 -> 5)

I can then access the values of named parameters in my code and the order they appear on the command line is no longer relevant. I realize this is fairly simple and doesn't have all the advanced functionality mentioned above but seems to be sufficient in most cases, only needs one line of code, and doesn't involve external dependencies.


I just created my simple enumeration

val args: Array[String] = "-silent -samples 100 -silent".split(" +").toArray
                                              //> args  : Array[String] = Array(-silent, -samples, 100, -silent)
object Opts extends Enumeration {

    class OptVal extends Val {
        override def toString = "-" + super.toString
    }

    val nopar, silent = new OptVal() { // boolean options
        def apply(): Boolean = args.contains(toString)
    }

    val samples, maxgen = new OptVal() { // integer options
        def apply(default: Int) = { val i = args.indexOf(toString) ;  if (i == -1) default else args(i+1).toInt}
        def apply(): Int = apply(-1)
    }
}

Opts.nopar()                              //> res0: Boolean = false
Opts.silent()                             //> res1: Boolean = true
Opts.samples()                            //> res2: Int = 100
Opts.maxgen()                             //> res3: Int = -1

I understand that solution has two major flaws that may distract you: It eliminates the freedom (i.e. the dependence on other libraries, that you value so much) and redundancy (the DRY principle, you do type the option name only once, as Scala program variable and eliminate it second time typed as command line text).


I based my approach on the top answer (from dave4420), and tried to improve it by making it more general-purpose.

It returns a Map[String,String] of all command line parameters You can query this for the specific parameters you want (eg using .contains) or convert the values into the types you want (eg using toInt).

def argsToOptionMap(args:Array[String]):Map[String,String]= {
  def nextOption(
      argList:List[String], 
      map:Map[String, String]
    ) : Map[String, String] = {
    val pattern       = "--(\\w+)".r // Selects Arg from --Arg
    val patternSwitch = "-(\\w+)".r  // Selects Arg from -Arg
    argList match {
      case Nil => map
      case pattern(opt)       :: value  :: tail => nextOption( tail, map ++ Map(opt->value) )
      case patternSwitch(opt) :: tail => nextOption( tail, map ++ Map(opt->null) )
      case string             :: Nil  => map ++ Map(string->null)
      case option             :: tail => {
        println("Unknown option:"+option) 
        sys.exit(1)
      }
    }
  }
  nextOption(args.toList,Map())
}

Example:

val args=Array("--testing1","testing1","-a","-b","--c","d","test2")
argsToOptionMap( args  )

Gives:

res0: Map[String,String] = Map(testing1 -> testing1, a -> null, b -> null, c -> d, test2 -> null)

I am from Java world, I like args4j because its simple, specification is more readable( thanks to annotations) and produces nicely formatted output.

Here is my example snippet:

Specification

import org.kohsuke.args4j.{CmdLineException, CmdLineParser, Option}

object CliArgs {

  @Option(name = "-list", required = true,
    usage = "List of Nutch Segment(s) Part(s)")
  var pathsList: String = null

  @Option(name = "-workdir", required = true,
    usage = "Work directory.")
  var workDir: String = null

  @Option(name = "-master",
    usage = "Spark master url")
  var masterUrl: String = "local[2]"

}

Parse

//var args = "-listt in.txt -workdir out-2".split(" ")
val parser = new CmdLineParser(CliArgs)
try {
  parser.parseArgument(args.toList.asJava)
} catch {
  case e: CmdLineException =>
    print(s"Error:${e.getMessage}\n Usage:\n")
    parser.printUsage(System.out)
    System.exit(1)
}
println("workDir  :" + CliArgs.workDir)
println("listFile :" + CliArgs.pathsList)
println("master   :" + CliArgs.masterUrl)

On invalid arguments

Error:Option "-list" is required
 Usage:
 -list VAL    : List of Nutch Segment(s) Part(s)
 -master VAL  : Spark master url (default: local[2])
 -workdir VAL : Work directory.

I like the clean look of this code... gleaned from a discussion here: http://www.scala-lang.org/old/node/4380

object ArgParser {
  val usage = """
Usage: parser [-v] [-f file] [-s sopt] ...
Where: -v   Run verbosely
       -f F Set input file to F
       -s S Set Show option to S
"""

  var filename: String = ""
  var showme: String = ""
  var debug: Boolean = false
  val unknown = "(^-[^\\s])".r

  val pf: PartialFunction[List[String], List[String]] = {
    case "-v" :: tail => debug = true; tail
    case "-f" :: (arg: String) :: tail => filename = arg; tail
    case "-s" :: (arg: String) :: tail => showme = arg; tail
    case unknown(bad) :: tail => die("unknown argument " + bad + "\n" + usage)
  }

  def main(args: Array[String]) {
    // if there are required args:
    if (args.length == 0) die()
    val arglist = args.toList
    val remainingopts = parseArgs(arglist,pf)

    println("debug=" + debug)
    println("showme=" + showme)
    println("filename=" + filename)
    println("remainingopts=" + remainingopts)
  }

  def parseArgs(args: List[String], pf: PartialFunction[List[String], List[String]]): List[String] = args match {
    case Nil => Nil
    case _ => if (pf isDefinedAt args) parseArgs(pf(args),pf) else args.head :: parseArgs(args.tail,pf)
  }

  def die(msg: String = usage) = {
    println(msg)
    sys.exit(1)
  }

}

I like sliding over arguments for relatively simple configurations.

var name = ""
var port = 0
var ip = ""
args.sliding(2, 2).toList.collect {
  case Array("--ip", argIP: String) => ip = argIP
  case Array("--port", argPort: String) => port = argPort.toInt
  case Array("--name", argName: String) => name = argName
}

Command Line Interface Scala Toolkit (CLIST)

here is mine too! (a bit late in the game though)

https://github.com/backuity/clist

As opposed to scopt it is entirely mutable... but wait! That gives us a pretty nice syntax:

class Cat extends Command(description = "concatenate files and print on the standard output") {

  // type-safety: members are typed! so showAll is a Boolean
  var showAll        = opt[Boolean](abbrev = "A", description = "equivalent to -vET")
  var numberNonblank = opt[Boolean](abbrev = "b", description = "number nonempty output lines, overrides -n")

  // files is a Seq[File]
  var files          = args[Seq[File]](description = "files to concat")
}

And a simple way to run it:

Cli.parse(args).withCommand(new Cat) { case cat =>
    println(cat.files)
}

You can do a lot more of course (multi-commands, many configuration options, ...) and has no dependency.

I'll finish with a kind of distinctive feature, the default usage (quite often neglected for multi commands): clist


How to parse parameters without an external dependency. Great question! You may be interested in picocli.

Picocli is specifically designed to solve the problem asked in the question: it is a command line parsing framework in a single file, so you can include it in source form. This lets users run picocli-based applications without requiring picocli as an external dependency.

It works by annotating fields so you write very little code. Quick summary:

  • Strongly typed everything - command line options as well as positional parameters
  • Support for POSIX clustered short options (so it handles <command> -xvfInputFile as well as <command> -x -v -f InputFile)
  • An arity model that allows a minimum, maximum and variable number of parameters, e.g, "1..*", "3..5"
  • Fluent and compact API to minimize boilerplate client code
  • Subcommands
  • Usage help with ANSI colors

The usage help message is easy to customize with annotations (without programming). For example:

Extended usage help message (source)

I couldn't resist adding one more screenshot to show what kind of usage help messages are possible. Usage help is the face of your application, so be creative and have fun!

picocli demo

Disclaimer: I created picocli. Feedback or questions very welcome. It is written in java, but let me know if there is any issue using it in scala and I'll try to address it.


I liked the slide() approach of joslinm just not the mutable vars ;) So here's an immutable way to that approach:

case class AppArgs(
              seed1: String,
              seed2: String,
              ip: String,
              port: Int
              )
object AppArgs {
  def empty = new AppArgs("", "", "", 0)
}

val args = Array[String](
  "--seed1", "akka.tcp://seed1",
  "--seed2", "akka.tcp://seed2",
  "--nodeip", "192.167.1.1",
  "--nodeport", "2551"
)

val argsInstance = args.sliding(2, 1).toList.foldLeft(AppArgs.empty) { case (accumArgs, currArgs) => currArgs match {
    case Array("--seed1", seed1) => accumArgs.copy(seed1 = seed1)
    case Array("--seed2", seed2) => accumArgs.copy(seed2 = seed2)
    case Array("--nodeip", ip) => accumArgs.copy(ip = ip)
    case Array("--nodeport", port) => accumArgs.copy(port = port.toInt)
    case unknownArg => accumArgs // Do whatever you want for this case
  }
}

I've just found an extensive command line parsing library in scalac's scala.tools.cmd package.

See http://www.assembla.com/code/scala-eclipse-toolchain/git/nodes/src/compiler/scala/tools/cmd?rev=f59940622e32384b1e08939effd24e924a8ba8db


Poor man's quick-and-dirty one-liner for parsing key=value pairs:

def main(args: Array[String]) {
    val cli = args.map(_.split("=") match { case Array(k, v) => k->v } ).toMap
    val saveAs = cli("saveAs")
    println(saveAs)
}

I've attempted generalize @pjotrp's solution by taking in a list of required positional key symbols, a map of flag -> key symbol and default options:

def parseOptions(args: List[String], required: List[Symbol], optional: Map[String, Symbol], options: Map[Symbol, String]): Map[Symbol, String] = {
  args match {
    // Empty list
    case Nil => options

    // Keyword arguments
    case key :: value :: tail if optional.get(key) != None =>
      parseOptions(tail, required, optional, options ++ Map(optional(key) -> value))

    // Positional arguments
    case value :: tail if required != Nil =>
      parseOptions(tail, required.tail, optional, options ++ Map(required.head -> value))

    // Exit if an unknown argument is received
    case _ =>
      printf("unknown argument(s): %s\n", args.mkString(", "))
      sys.exit(1)
  }
}

def main(sysargs Array[String]) {
  // Required positional arguments by key in options
  val required = List('arg1, 'arg2)

  // Optional arguments by flag which map to a key in options
  val optional = Map("--flag1" -> 'flag1, "--flag2" -> 'flag2)

  // Default options that are passed in
  var defaultOptions = Map()

  // Parse options based on the command line args
  val options = parseOptions(sysargs.toList, required, optional, defaultOptions)
}

Here's a scala command line parser that is easy to use. It automatically formats help text, and it converts switch arguments to your desired type. Both short POSIX, and long GNU style switches are supported. Supports switches with required arguments, optional arguments, and multiple value arguments. You can even specify the finite list of acceptable values for a particular switch. Long switch names can be abbreviated on the command line for convenience. Similar to the option parser in the Ruby standard library.


Here is mine 1-liner

    def optArg(prefix: String) = args.drop(3).find { _.startsWith(prefix) }.map{_.replaceFirst(prefix, "")}
    def optSpecified(prefix: String) = optArg(prefix) != None
    def optInt(prefix: String, default: Int) = optArg(prefix).map(_.toInt).getOrElse(default)

It drops 3 mandatory arguments and gives out the options. Integers are specified like notorious -Xmx<size> java option, jointly with the prefix. You can parse binaries and integers as simple as

val cacheEnabled = optSpecified("cacheOff")
val memSize = optInt("-Xmx", 1000)

No need to import anything.


scala-optparse-applicative

I think scala-optparse-applicative is the most functional command line parser library in Scala.

https://github.com/bmjames/scala-optparse-applicative


As everyone posted it's own solution here is mine, cause I wanted something easier to write for the user : https://gist.github.com/gwenzek/78355526e476e08bb34d

The gist contains a code file, plus a test file and a short example copied here:

import ***.ArgsOps._


object Example {
    val parser = ArgsOpsParser("--someInt|-i" -> 4, "--someFlag|-f", "--someWord" -> "hello")

    def main(args: Array[String]){
        val argsOps = parser <<| args
        val someInt : Int = argsOps("--someInt")
        val someFlag : Boolean = argsOps("--someFlag")
        val someWord : String = argsOps("--someWord")
        val otherArgs = argsOps.args

        foo(someWord, someInt, someFlag)
    }
}

There is not fancy options to force a variable to be in some bounds, cause I don't feel that the parser is the best place to do so.

Note : you can have as much alias as you want for a given variable.


I'd suggest to use http://docopt.org/. There's a scala-port but the Java implementation https://github.com/docopt/docopt.java works just fine and seems to be better maintained. Here's an example:

import org.docopt.Docopt

import scala.collection.JavaConversions._
import scala.collection.JavaConverters._

val doc =
"""
Usage: my_program [options] <input>

Options:
 --sorted   fancy sorting
""".stripMargin.trim

//def args = "--sorted test.dat".split(" ").toList
var results = new Docopt(doc).
  parse(args()).
  map {case(key, value)=>key ->value.toString}

val inputFile = new File(results("<input>"))
val sorted = results("--sorted").toBoolean

another library: scarg


There's also JCommander (disclaimer: I created it):

object Main {
  object Args {
    @Parameter(
      names = Array("-f", "--file"),
      description = "File to load. Can be specified multiple times.")
    var file: java.util.List[String] = null
  }

  def main(args: Array[String]): Unit = {
    new JCommander(Args, args.toArray: _*)
    for (filename <- Args.file) {
      val f = new File(filename)
      printf("file: %s\n", f.getName)
    }
  }
}

I have never liked ruby like option parsers. Most developers that used them never write a proper man page for their scripts and end up with pages long options not organized in a proper way because of their parser.

I have always preferred Perl's way of doing things with Perl's Getopt::Long.

I am working on a scala implementation of it. The early API looks something like this:

def print_version() = () => println("version is 0.2")

def main(args: Array[String]) {
  val (options, remaining) = OptionParser.getOptions(args,
    Map(
      "-f|--flag"       -> 'flag,
      "-s|--string=s"   -> 'string,
      "-i|--int=i"      -> 'int,
      "-f|--float=f"    -> 'double,
      "-p|-procedure=p" -> { () => println("higher order function" }
      "-h=p"            -> { () => print_synopsis() }
      "--help|--man=p"  -> { () => launch_manpage() },
      "--version=p"     -> print_version,
    ))

So calling script like this:

$ script hello -f --string=mystring -i 7 --float 3.14 --p --version world -- --nothing

Would print:

higher order function
version is 0.2

And return:

remaining = Array("hello", "world", "--nothing")

options = Map('flag   -> true,
              'string -> "mystring",
              'int    -> 7,
              'double -> 3.14)

The project is hosted in github scala-getoptions.


This is largely a shameless clone of my answer to the Java question of the same topic. It turns out that JewelCLI is Scala-friendly in that it doesn't require JavaBean style methods to get automatic argument naming.

JewelCLI is a Scala-friendly Java library for command-line parsing that yields clean code. It uses Proxied Interfaces Configured with Annotations to dynamically build a type-safe API for your command-line parameters.

An example parameter interface Person.scala:

import uk.co.flamingpenguin.jewel.cli.Option

trait Person {
  @Option def name: String
  @Option def times: Int
}

An example usage of the parameter interface Hello.scala:

import uk.co.flamingpenguin.jewel.cli.CliFactory.parseArguments
import uk.co.flamingpenguin.jewel.cli.ArgumentValidationException

object Hello {
  def main(args: Array[String]) {
    try {
      val person = parseArguments(classOf[Person], args:_*)
      for (i <- 1 to (person times))
        println("Hello " + (person name))
    } catch {
      case e: ArgumentValidationException => println(e getMessage)
    }
  }
}

Save copies of the files above to a single directory and download the JewelCLI 0.6 JAR to that directory as well.

Compile and run the example in Bash on Linux/Mac OS X/etc.:

scalac -cp jewelcli-0.6.jar:. Person.scala Hello.scala
scala -cp jewelcli-0.6.jar:. Hello --name="John Doe" --times=3

Compile and run the example in the Windows Command Prompt:

scalac -cp jewelcli-0.6.jar;. Person.scala Hello.scala
scala -cp jewelcli-0.6.jar;. Hello --name="John Doe" --times=3

Running the example should yield the following output:

Hello John Doe
Hello John Doe
Hello John Doe

This is what I cooked. It returns a tuple of a map and a list. List is for input, like input file names. Map is for switches/options.

val args = "--sw1 1 input_1 --sw2 --sw3 2 input_2 --sw4".split(" ")
val (options, inputs) = OptParser.parse(args)

will return

options: Map[Symbol,Any] = Map('sw1 -> 1, 'sw2 -> true, 'sw3 -> 2, 'sw4 -> true)
inputs: List[Symbol] = List('input_1, 'input_2)

Switches can be "--t" which x will be set to true, or "--x 10" which x will be set to "10". Everything else will end up in list.

object OptParser {
  val map: Map[Symbol, Any] = Map()
  val list: List[Symbol] = List()

  def parse(args: Array[String]): (Map[Symbol, Any], List[Symbol]) = _parse(map, list, args.toList)

  private [this] def _parse(map: Map[Symbol, Any], list: List[Symbol], args: List[String]): (Map[Symbol, Any], List[Symbol]) = {
    args match {
      case Nil => (map, list)
      case arg :: value :: tail if (arg.startsWith("--") && !value.startsWith("--")) => _parse(map ++ Map(Symbol(arg.substring(2)) -> value), list, tail)
      case arg :: tail if (arg.startsWith("--")) => _parse(map ++ Map(Symbol(arg.substring(2)) -> true), list, tail)
      case opt :: tail => _parse(map, list :+ Symbol(opt), tail)
    }
  }
}

I realize that the question was asked some time ago, but I thought it might help some people, who are googling around (like me), and hit this page.

Scallop looks quite promising as well.

Features (quote from the linked github page):

  • flag, single-value and multiple value options
  • POSIX-style short option names (-a) with grouping (-abc)
  • GNU-style long option names (--opt)
  • Property arguments (-Dkey=value, -D key1=value key2=value)
  • Non-string types of options and properties values (with extendable converters)
  • Powerful matching on trailing args
  • Subcommands

And some example code (also from that Github page):

import org.rogach.scallop._;

object Conf extends ScallopConf(List("-c","3","-E","fruit=apple","7.2")) {
  // all options that are applicable to builder (like description, default, etc) 
  // are applicable here as well
  val count:ScallopOption[Int] = opt[Int]("count", descr = "count the trees", required = true)
                .map(1+) // also here work all standard Option methods -
                         // evaluation is deferred to after option construction
  val properties = props[String]('E')
  // types (:ScallopOption[Double]) can be omitted, here just for clarity
  val size:ScallopOption[Double] = trailArg[Double](required = false)
}


// that's it. Completely type-safe and convenient.
Conf.count() should equal (4)
Conf.properties("fruit") should equal (Some("apple"))
Conf.size.get should equal (Some(7.2))
// passing into other functions
def someInternalFunc(conf:Conf.type) {
  conf.count() should equal (4)
}
someInternalFunc(Conf)

For most cases you do not need an external parser. Scala's pattern matching allows consuming args in a functional style. For example:

object MmlAlnApp {
  val usage = """
    Usage: mmlaln [--min-size num] [--max-size num] filename
  """
  def main(args: Array[String]) {
    if (args.length == 0) println(usage)
    val arglist = args.toList
    type OptionMap = Map[Symbol, Any]

    def nextOption(map : OptionMap, list: List[String]) : OptionMap = {
      def isSwitch(s : String) = (s(0) == '-')
      list match {
        case Nil => map
        case "--max-size" :: value :: tail =>
                               nextOption(map ++ Map('maxsize -> value.toInt), tail)
        case "--min-size" :: value :: tail =>
                               nextOption(map ++ Map('minsize -> value.toInt), tail)
        case string :: opt2 :: tail if isSwitch(opt2) => 
                               nextOption(map ++ Map('infile -> string), list.tail)
        case string :: Nil =>  nextOption(map ++ Map('infile -> string), list.tail)
        case option :: tail => println("Unknown option "+option) 
                               exit(1) 
      }
    }
    val options = nextOption(Map(),arglist)
    println(options)
  }
}

will print, for example:

Map('infile -> test/data/paml-aln1.phy, 'maxsize -> 4, 'minsize -> 2)

This version only takes one infile. Easy to improve on (by using a List).

Note also that this approach allows for concatenation of multiple command line arguments - even more than two!


Examples related to scala

Intermediate language used in scalac? Why does calling sumr on a stream with 50 tuples not complete Select Specific Columns from Spark DataFrame Joining Spark dataframes on the key Provide schema while reading csv file as a dataframe how to filter out a null value from spark dataframe Fetching distinct values on a column using Spark DataFrame Can't push to the heroku Spark - Error "A master URL must be set in your configuration" when submitting an app Add jars to a Spark Job - spark-submit

Examples related to command-line

Git is not working after macOS Update (xcrun: error: invalid active developer path (/Library/Developer/CommandLineTools) Flutter command not found Angular - ng: command not found how to run python files in windows command prompt? How to run .NET Core console app from the command line Copy Paste in Bash on Ubuntu on Windows How to find which version of TensorFlow is installed in my system? How to install JQ on Mac by command-line? Python not working in the command line of git bash Run function in script from command line (Node JS)

Examples related to command-line-arguments

How to pass arguments to Shell Script through docker run How can I pass variable to ansible playbook in the command line? Use Robocopy to copy only changed files? mkdir's "-p" option What is Robocopy's "restartable" option? How do you run a .exe with parameters using vba's shell()? Bash command line and input limit Check number of arguments passed to a Bash script Parsing boolean values with argparse Import SQL file by command line in Windows 7