Parse command-line arguments

From Rosetta Code
Parse command-line arguments is a draft programming task. It is not yet considered ready to be promoted as a complete task, for reasons that should be found in its talk page.

Command-line arguments can be quite complicated, as in:

nc   -v   -n   -z   -w   1   1-1000

Many languages provide a library (getopt or GetOpt) to parse the raw command line options in an intelligent way.


For AutoHotkey v1.1+

;Get Arguments as an array
if 0 > 0
Loop, %argc%
;if got no arguments, run self with arguments
Run,%a_scriptFullpath% -i Lib\* -c files.c --verbose -o files.o --Optimze
;Parse arguments
i:=0, msg:=""
while( i++ < argc ) {
if c in -,/ ; List all switch chars
if ( SubStr(args[i],1,2) == "--" ) ; if "--" is used like "--verbose"
msg:=msg args[i] "`t:`tTrue (Boolean)`n" ; parse as boolean
msg:=msg args[i] "`t:`t" args[++i] "`n"
msg:=msg args[i] "`t:`t(normal)`n"
MsgBox % "Parsed Arguments :`n" msg

Output (MsgBox):

Parsed Arguments :
-i        :  Lib\*
-c        :  files.c
--verbose :  True (Boolean)
-o        :  files.o
--Optimze :  True (Boolean)


Works with: gawk
#!/usr/bin/awk -E
# -E instead of -f so program arguments don't conflict with Gawk arguments
@include "getopt.awk"
while ((C = getopt(ARGC, ARGV, "ht:u:")) != -1) {
if(C == "h") {
if(C == "t")
tval = Optarg
if(C == "u")
uval = Optarg
print "There are " opti " arguments."
if(tval) print "-t = " tval
if(uval) print "-u = " uval


Per default, Bracmat treats all arguments as expressions and parses and evaluates them from left to right. A call to the function arg$ pops the next argument from the list of arguments and returns it as an inert string in no need of further parsing and evaluation.

bracmat arg$:?a 123 arg$:?b 77 !a+!b:?c out$!c




The man page for getopt (man 3 getopt) provides better option handling with examples. But if you just want to parse one argument... (adapted from simple database task):

#include <stdio.h>
int main(int argc, char **argv){
int i;
const char *commands[]={"-c", "-p", "-t", "-d", "-a", NULL};
if (argc<2) {
usage: printf ("Usage: %s [commands]\n"
"-c Create new entry.\n"
"-p Print the latest entry.\n"
"-t Sort by title.\n"
"-d Sort by date.\n"
"-a Sort by author.\n",argv[0]);
return 0;
for (i=0;commands[i]&&strcmp(argv[1],commands[i]);i++);
switch (i) {
case CREATE:
case PRINT:
printf ("Unknown command..." ...);
goto usage;
return 0;


See Parsing Command-Line Arguments from O'Reilly's Clojure Cookbook github.


The getopt module in D's standard library is inspired by Perl's Getopt::Long module. The syntax of Phobos getopt infers the expected parameter types from the static types of the passed-in pointers.

import std.stdio, std.getopt;
void main(string[] args) {
string data = "file.dat";
int length = 24;
bool verbose;
enum Color { no, yes }
Color color;
args.getopt("length", &length, // Integer.
"file", &data, // String.
"verbose", &verbose, // Boolean flag.
"color", &color); // Enum.
writeln("length: ", length);
writeln("file: ", data);
writeln("verbose: ", verbose);
writeln("color: ", color);
Usage example:
C:\getopt_test --verbose --length 12
length: 12
file: file.dat
verbose: true
color: no


Elixir provides an option parser in a library module called OptionParser.

#!/usr/bin/env elixir
IO.puts 'Arguments:'
IO.inspect OptionParser.parse(System.argv())
$ ./parse-args.exs --a --b --c=yes --no-flag --verbose -V -a=1 -b=t -- apple banana
{[a: true, b: true, c: "yes", no_flag: true, verbose: true],
["apple", "banana"], [{"-V", nil}, {"-a", "1"}, {"-b", "t"}]}


' FB 1.05.0 Win64
' Program (commandline.exe) invoked like this:
' commandline nc -v -n -z -w 1 1-1000
Dim argc As Integer = __FB_ARGC__
Dim argv As ZString Ptr Ptr = __FB_ARGV__
Print "The program was invoked with the following command line arguments:"
For i As Integer = 0 To argc - 1
Print "Arg"; i + 1; " = "; *argv[i]
Print "Press any key to quit"
The program was invoked with the following command line arguments:

Arg 1 = commandline
Arg 2 = nc
Arg 3 = -v
Arg 4 = -n
Arg 5 = -z
Arg 6 = -w
Arg 7 = 1
Arg 8 =
Arg 9 = 1-1000


Most simply, implementing the suggested example from the talk page:

package main
import (
func main() {
b := flag.Bool("b", false, "just a boolean")
s := flag.String("s", "", "any ol' string")
n := flag.Int("n", 0, "your lucky number")
fmt.Println("b:", *b)
fmt.Println("s:", *s)
fmt.Println("n:", *n)

Example runs:

> parse
b: false
n: 0

> parse -s bye -b
b: true
s: bye
n: 0

> parse -n 99 -s "say my name"
b: false
s: say my name
n: 99

Icon and Unicon[edit]

The Icon Programming Library provides a procedure for processing command line options. See the library reference for detailed documentation. The code below is an example.

link options
procedure main(ARGLIST)
/errproc := stop # special error procedure or stop()
opstring := "f!s:i+r.flag!string:integer+real." # example
opttable := options(ARGLIST,optstring,errproc)
if \opttable[flag] then ... # test a flag
r := opttable(real) # assign a real
r2 := opttable(r) # assign another real
s := opttable(s) # assign a string
i := opttable(i) # assign an integer

options.icn supports getting command-line options


When J starts up from the command line, the command line arguments are available in the array ARGV. On modern machines, the first command line argument is the name of the executable (the J interpeter, in this case).

Typically, the next argument (if present) is the name of a file whose contents will be executed.

Further command line analysis might include:

Test if an argument is present:
   (<'-b') e. ARGV
This is true if the argument is present and false, if it is not.
Or, find the name of an optional file:
   (ARGV i.<'-f') {:: }.ARGV,a:
This is the name of the first file named after the first -f argument, or empty if there was no such file.

Other concepts are also possible...


Works with: Julia version 0.6

Example taken from the official documentation of ArgParse docs.

using ArgParse
function parse_commandline()
s = ArgParseSettings()
@add_arg_table s begin
help = "an option with an argument"
"--opt2", "-o"
help = "another option with an argument"
arg_type = Int
default = 0
help = "an option without argument, i.e. a flag"
action = :store_true
help = "a positional argument"
required = true
return parse_args(s)
function main()
parsed_args = parse_commandline()
println("Parsed args:")
for (arg,val) in parsed_args
println(" $arg => $val")


// version 1.0.6 (packaged as parse_cla.jar)
fun main(args: Array<String>) = println(args.asList())
c:\kotlin-compiler-1.0.6>java -jar parse_cla.jar nc -v -n -z -w 1 1-1000
[nc, -v, -n, -z, -w, 1,, 1-1000]


The command line is parsed and stored into a list of strings to ease manual handling by list processing functions.

-> {math, -v, -n, -z, -w, 1,, 1-1000}


GP exists in a REPL and so it doesn't make sense to parse command-line arguments. But PARI can parse them just like C:

#include <pari/pari.h>
#include <stdio.h>
int main(int argc, char **argv){
pari_printf("8 + 1 = %Ps\n", addii(int2u(3), gen_1));
return 0;


import os
import parseopt
proc main =
# Directly accessing the app name and parameters
echo "app name: ", getAppFilename().extractFilename()
echo "# parameters: ", paramCount()
for ii in 1 .. paramCount(): # 1st param is at index 1
echo "param ", ii, ": ", paramStr(ii)
echo ""
# Using parseopt module to extract short and long options and arguments
var argCtr : int
for kind, key, value in getOpt():
case kind
of cmdArgument:
echo "Got arg ", argCtr, ": \"", key, "\""
of cmdLongOption, cmdShortOption:
case key
of "v", "n", "z", "w":
echo "Got a \"", key, "\" option with value: \"", value, "\""
echo "Unknown option: ", key
of cmdEnd:

Sample command line:

parsecmdline ab -z cd ef -w=abcd --w=1234 -v -e -x 1-1000


app name: parsecmdline
# parameters: 10
param 1: ab
param 2: -z
param 3: cd
param 4: ef
param 5: -w=abcd
param 6: --w=1234
param 7: -v
param 8: -e
param 9: -x
param 10: 1-1000

Got arg 0: "ab"
Got a "z" option with value: ""
Got arg 1: "cd"
Got arg 2: "ef"
Got a "w" option with value: "abcd"
Got a "w" option with value: "1234"
Got a "v" option with value: ""
Unknown option: e
Unknown option: x
Got arg 3: "1-1000"


Use the Getopt::Long module:

# Copyright Shlomi Fish, 2013 under the MIT/X11 License.
use strict;
use warnings;
use Getopt::Long qw(GetOptions);
my $output_path;
my $verbose = '';
my $length = 24;
"length=i" => \$length,
"output|o=s" => \$output_path,
"verbose!" => \$verbose,
) or die ("Error in command line arguments");
print "Outputting to '", ($output_path // '(undefined)'), "' path, with ",
($verbose ? "Verbosity" : "No verbosity"),
" and a length of $length.\n";

The output from it is:

$ perl --verbose -o foo.xml
Outputting to 'foo.xml' path, with Verbosity and a length of 24.
$ perl --verbose
Outputting to '(undefined)' path, with Verbosity and a length of 24.
$ perl --verbose --length=190
Outputting to '(undefined)' path, with Verbosity and a length of 190.
$ perl --verbose --length=190 -o test.txt
Outputting to 'test.txt' path, with Verbosity and a length of 190.

Perl 6[edit]

At the end of running any top-level code (which can preprocess the arguments if it likes), Perl 6 automatically examines any remaining arguments and transforms them into a call to a MAIN routine, if one is defined. The arguments are parsed based on the signature of the routine, so that options are mapped to named arguments.

sub MAIN (Bool :$b, Str :$s = '', Int :$n = 0, *@rest) {
say "Bool: $b";
say "Str: $s";
say "Num: $n";
say "Rest: @rest[]";
$ ./main -h
  ./main [-b] [-s=<Str>] [-n=<Int>] [<rest> ...]

$ ./main -b -n=42 -s=turtles all the way down
Bool: True
Str: turtles
Num: 42
Rest: all the way down

If there are multiple MAIN subs, they are differentiated by multiple dispatch. A help message can automatically be generated for all the variants. The intent of this mechanism is not to cover every possible switch structure, but just to make it drop-dead easy to handle most of the common ones.


sequence res = command_line()

Interpreted: res[1] is the interpreter, res[2] the source

> p test nc -v -n -z -w 1 1-1000
{"C:\\Program Files (x86)\\Phix\\p.exe","C:\\Program Files (x86)\\Phix\\test.exw","nc","-v","-n","-z","-w","1","","1-1000"}

Compiled: both res[1] and res[2] are the executable

> p -c test nc -v -n -z -w 1 1-1000
{"C:\\Program Files (x86)\\Phix\\test.exe","C:\\Program Files (x86)\\Phix\\test.exe","nc","-v","-n","-z","-w","1","","1-1000"}
> test nc -v -n -z -w 1 1-1000
{"C:\\Program Files (x86)\\Phix\\test.exe","C:\\Program Files (x86)\\Phix\\test.exe","nc","-v","-n","-z","-w","1","","1-1000"}


Version 2.3+

from optparse import OptionParser
parser = OptionParser()
parser.add_option("-f", "--file", dest="filename",
help="write report to FILE", metavar="FILE")
parser.add_option("-q", "--quiet",
action="store_false", dest="verbose", default=True,
help="don't print status messages to stdout")
(options, args) = parser.parse_args()
<yourscript> --file=outfile -q


PicoLisp doesn't have a library to get options. Instead, the command line is parsed at startup and handled in the following way: Each command line argument is executed (interpreted) as a Lisp source file, except that if the first character is a hypen '-', then that arguments is taken as a Lisp function call (without the surrounding parentheses). For example, the command line

$ ./pil abc.l -foo def.l -"bar 3 4" -'mumble "hello"' -bye

has the effect that

  1. The file "abc.l" is executed
  2. (foo) is called
  3. The file "def.l" is executed
  4. (bar 3 4) is called
  5. (mumble "hello") is called
  6. (bye) is called, resulting in program termination

Command line arguments like "-v", "-n" and "-z" can be implemented simply by defining three functions 'v', 'n' and 'z'.

In addition to the above mechanism, the command line can also be handled "manually", by either processing the list of arguments returned by 'argv', or by fetching arguments individually with 'opt'.


Powershell functions and filters handle options organically, with advanced .NET support to handle complex and advanced options including aliases, ranges, sets, datatypes, and more. See more here. However, to parse options 'classically', you can write a custom parser. A slightly messy version (inspired by ruby optparse) that can only handle switches and relies on RegEx:

$options = @{
opt1 = [bool] 0
opt2 = [bool] 0
opt3 = [bool] 0
$help = @"
FUNCTION usage: FUNCTION [-p] [-w] [-h] [-c] <int><float><string>PARAMETERS...
Lorem Ipsum blah blah blah
NOTE something yada yada
-p,--pxx Name Some option that has significance with the letter 'p'
-w,--wxx Name Some option that has significance with the letter 'w'
-c,--cxx Name Some option that has significance with the letter 'c'
-h,--help Help Prints this message
function parseOptions ($argv,$options) {
$opts = @()
if (!$argv) { return $null }
foreach ($arg in $argv) {
# Make sure the argument is something you are expecting
$test = ($arg -is [int]) -or
($arg -is [string]) -or
($arg -is [float])
if (!$test) {
Write-Host "Bad argument: $arg is not an integer, float, nor string." -ForegroundColor Red
throw "Error: Bad Argument"
if ($arg -like '-*') { $opts += $arg }
$argv = [Collections.ArrayList]$argv
if ($opts) {
foreach ($opt in $opts) {
switch ($opt) {
{'-p' -or '--pxx'} { $options.opt1 = [bool] 1 }
{'-w' -or '--wxx'} { $options.opt2 = [bool] 1 }
{'-c' -or '--cxx'} { $options.opt3 = [bool] 1 }
{'-h' -or '--help'} { Write-Host $help -ForegroundColor Cyan; break 1 }
default {
Write-Host "Bad option: $opt is not a valid option." -ForegroundColor Red
throw "Error: Bad Option"
return [array]$argv,$options

Usage (in some function or script):

$argv,$options = parseOptions $args $options
if ($options.opt3) {
$foo = $blah - ($yada * $options.opt1) + ($yada * $options.opt2)
$bar = $argv | SomeOtherFilter | Baz

Usage in shell:

$> function -c --wxx arg1 arg2

Note that this works in Powershell. All of the arguments after the function name will be passed as strings to the function, which then calls them as an array with the automatic variable, $args. The custom parser/function does the work from there, turning the strings into flags and typed arguments. WARNING: This is reinventing the wheel to an extreme degree.


Racket has a good command-line parsing library, the following demonstrates some of its features:

#!/usr/bin/env racket
#lang racket
(define loglevel 1)
(define mode 'new)
(define ops '())
(define root #f)
[("-v" "--verbose") "more verbose" (set! loglevel (add1 loglevel))]
[("-q" "--quiet") "be quiet" (set! loglevel 0)]
[("-i" "--in-place") "edit in-place" (set! mode 'in-place)]
[("-c" "--create-new") "create a new file" (set! mode 'new)]
[("-n" "--dry-run") "do nothing" (set! mode #f)]
[("-d" "--directory") dir "work in a given directory" (set! root dir)]
#:help-labels "operations to perform:"
[("+l" "++line") "add a line" (set! ops `(,@ops "add"))]
[("-l" "--line") "delete a line" (set! ops `(,@ops "delete"))]
[("-e" "--edit") "edit a line" (set! ops `(,@ops "edit"))]
#:args (file . files)
(printf "Running on: ~a\n" (string-join (cons file files) ", "))
(when root (printf "In Dir: ~a\n" root))
(printf "Mode: ~s\n" mode)
(printf "Log level: ~s\n" loglevel)
(printf "Operations: ~a\n" (string-join ops ", ")))

Sample runs:

$ ./foo -h
foo [ <option> ... ] <file> [<files>] ...
 where <option> is one of
* -v, --verbose : more verbose
* -q, --quiet : be quiet
/ -i, --in-place : edit in-place
| -c, --create-new : create a new file
\ -n, --dry-run : do nothing
  -d <dir>, --directory <dir> : work in a given directory
 operations to perform:
* +l, ++line : add a line
* -l, --line : delete a line
* -e, --edit : edit a line
  --help, -h : Show this help
  -- : Do not treat any remaining argument as a switch (at this level)
 * Asterisks indicate options allowed multiple times.
 /|\ Brackets indicate mutually exclusive options.
 Multiple single-letter switches can be combined after one `-'; for
  example: `-h-' is the same as `-h --'

$ /tmp/zz.rkt -viqvvd /tmp +ll -el foo bar baz
Running on: foo, bar, baz
In Dir:     /tmp
Mode:       in-place
Log level:  2
Operations: add, add, edit, delete


║ The subject of parsing text  (such as a command line)  is ingrained in the  ║
║ REXX language;  it has a PARSE instruction.  It's too rich to describe all  ║
║ its functionality/capabilities here,  but since the task isn't described,   ║
║ I'm assuming the    NC   [NetCat]  example  (or something like it)  is toe  ║
║ be parsed,  and I'm guessing at its syntax  (from other examples found on   ║
║ the web),  but I'll take a incomplete stab at it.                           ║
║                                                                             ║
║ For the most part,  every command appear to have their own rules for their  ║
║ operands (options),  as does the   NC   command.     For instance           ║
║                                                                             ║
║         nc  -u xxx -p port1 ···                                             ║
║                  and                                                        ║
║         nc  -u -p port1     ···                                             ║
║                                                                             ║
║ where the   -u   option has an operand in the first case,  but not the 2nd, ║
║ even though there is something following the   -u   option.                 ║
║                                                                             ║
║ It can only be assumed that any operand for the   -u   option can't start   ║
║ with a minus sign  [-].                                                     ║
/*REXX program demonstrates one method to parse options for a command (entered on the CL*/
parse arg opts /*this preserves the case of options. */
opts=space(opts) /*elide superfluous blanks in options. */
!.= /*all options to be "null" (default).*/
do while opts\=='' /*keep parsing 'til all opts examined*/
parse var opts x opts /*obtain a single keyword from options.*/
select /*hard-coded WHENs for option detection*/
when x=='-e' then parse var opts !.e_ opts
when x=='-p' then parse var opts !.p_ opts
when x=='-n' then !.z_=1
when x=='-u' then parse var opts !.uname_ !.unnn_ opts
when x=='-ul' then parse var opts !.ul_ opts
when x=='-vzu' then parse var opts !.vzu_  !.vzurange opts
when x=='-w' then parse var opts !.wStart_ !.waddr_ !.wrange_1 '-' !.wrange_2 opts
when x=='-z' then !.=1
otherwise call sayer 'option ' x " isn't a known option."
end /*select*/
end /*do while*/
/*check for conflicts here and/or validity of values.*/
if !.z_==1 & !.n_==1 then call sayer "N and Z can't both be specified."
if !.wrange_1\=='' then do /*see if it's a whole number (integer). */
if \isInt(!.wrange1_) then call sayer "wRange isn't an integer."
yada yada yada
...more stuff...
...and still more stuff...
exit /*stick a fork in it, we're all done. */
isInt: return datatype(arg(1), 'W') /*return 1 if argument is an integer*/
isNum: return datatype(arg(1), 'N') /*return 1 if argument is a number.*/
sayer: say; say '***error***' arg(1); exit 13
║ Note:  a programming trick is to append (say)  an underscore [_]  to  ║
║        an option's name as to not preclude that variable being used   ║
║        elsewhere in the REXX program.   That way, the option  J  can  ║
║        be used, as well as the variable  J  in the program.           ║


Ruby's standard library provides two different packages to parse command-line arguments.

  • 'getoptlong' resembles the libraries from other languages.
  • 'optparse' has more features.

Ruby with 'getoptlong'[edit]

#!/usr/bin/env ruby
# == Synopsis
# pargs: Phone a friend
# == Usage
# pargs [OPTIONS]
# --help, -h:
# show usage
# --eddy, -e <message>
# call eddy
# --danial, -d <message>
# call daniel
# --test, -t
# run unit tests
require "getoptlong"
require "rdoc/usage"
def phone(name, message)
puts "Calling #{name}..."
puts message
def test
phone("Barry", "Hi!")
phone("Cindy", "Hello!")
def main
mode = :usage
name = ""
message = ""
["--help", "-h", GetoptLong::NO_ARGUMENT],
["--eddy", "-e", GetoptLong::REQUIRED_ARGUMENT],
["--daniel", "-d", GetoptLong::REQUIRED_ARGUMENT],
["--test", "-t", GetoptLong::NO_ARGUMENT]
opts.each { |option, value|
case option
when "--help"
when "--eddy"
mode = :call
name = "eddy"
message = value
when "--daniel"
mode = :call
name = "daniel"
message = value
when "--test"
mode = :test
case mode
when :usage
when :call
phone(name, message)
when :test
if __FILE__==$0
rescue Interrupt => e
$ ./pargs.rb -h

pargs [OPTIONS]

--help, -h:

   show usage

--eddy, -e <message>

   call eddy

--daniel, -d <message>

   call daniel

--test, -t

   run unit tests

$ ./pargs.rb -e Yo!
Calling eddy...
$ ./pargs.rb --test
Calling Barry...
Calling Cindy...

Ruby with 'optparse'[edit]

require 'optparse'
sflag = false
longflag = false
count = 0
percent = 50
fruit = nil do |opts|
# Default banner is "Usage: #{opts.program_name} [options]".
opts.banner += " [arguments...]"
opts.separator "This demo prints the results of parsing the options."
opts.version = "0.0.1"
opts.on("-s", "Enable short flag") {sflag = true}
opts.on("--long", "Enable long flag") {longflag = true}
opts.on("-b", "--both", "Enable both -s and --long"
) {sflag = true; longflag = true}
opts.on("-c", "--count", "Add 1 to count") {count += 1}
# Argument must match a regular expression.
opts.on("-p", "--percent PERCENT", /[0-9]+%?/i,
"Percent [50%]") {|arg| percent = arg.to_i}
# Argument must match a list of symbols.
opts.on("-f", "--fruit FRUIT",
[:apple, :banana, :orange, :pear],
"Fruit (apple, banana, orange, pear)"
) {|arg| fruit = arg}
# Parse and remove options from ARGV.
rescue OptionParser::ParseError => error
# Without this rescue, Ruby would print the stack trace
# of the error. Instead, we want to show the error message,
# suggest -h or --help, and exit 1.
$stderr.puts error
$stderr.puts "(-h or --help will show valid options)"
exit 1
print <<EOF
Short flag: #{sflag}
Long flag: #{longflag}
Count: #{count}
Percent: #{percent}%
Fruit: #{fruit}
Arguments: #{ARGV.inspect}
$ ruby takeopts.rb -h
Usage: takeopts [options] [arguments...]
This demo prints the results of parsing the options.
    -s                               Enable short flag
        --long                       Enable long flag
    -b, --both                       Enable both -s and --long
    -c, --count                      Add 1 to count
    -p, --percent PERCENT            Percent [50%]
    -f, --fruit FRUIT                Fruit (apple, banana, orange, pear)
$ ruby takeopts.rb -v 
takeopts 0.0.1
$ ruby takeopts.rb -b -c
Short flag: true
Long flag: true
Count: 1
Percent: 50%
Arguments: []
$ ruby takeopts.rb -ccccp90% -f oran -- -arg
Short flag: false
Long flag: false
Count: 4
Percent: 90%
Fruit: orange
Arguments: ["-arg"]


Using the StructOpt:

use structopt::StructOpt;
struct Opt {
b: bool,
#[structopt(short, required = false, default_value = "")]
s: String,
#[structopt(short, required = false, default_value = "0")]
n: i32,
fn main() {
let opt = Opt::from_args();
println!("b: {}", opt.b);
println!("s: {}", opt.s);
println!("n: {}", opt.n);


> parse
b: false
n: 0

> parse -s bye -b
b: true
s: bye
n: 0

> parse -n 99 -s "say my name"
b: false
s: say my name
n: 99


Library: Scala
object CommandLineArguments extends App { 
println(s"Received the following arguments: + ${args.mkString("", ", ", ".")}")

Standard ML[edit]

Works with: SML/NJ
Works with: MLton

The following code listing can be compiled with both SML/NJ and MLton:

structure Test = struct
exception FatalError of string
fun main (prog, args) =
exception Args
val switch = ref false
fun do_A arg = print ("Argument of -A is " ^ arg ^ "\n")
fun do_B () = if !switch then print "switch is on\n" else print "switch is off\n"
fun usage () = print ("Usage: " ^ prog ^ " [-help] [-switch] [-A Argument] [-B]\n")
fun parseArgs nil = ()
| parseArgs ("-help"  :: ts) = (usage(); parseArgs ts)
| parseArgs ("-switch"  :: ts) = (switch := true; parseArgs ts)
| parseArgs ("-A" :: arg :: ts) = (do_A arg; parseArgs ts)
| parseArgs ("-B"  :: ts) = (do_B(); parseArgs ts)
| parseArgs _ = (usage(); raise Args)
parseArgs args handle Args => raise FatalError "Error parsing args. Use the -help option.";
(* Do something; *)
handle FatalError e => (print ("Fatal Error:\n"^e^"\n"); OS.Process.failure)
(* MLton *)
val _ = Test.main (, CommandLine.arguments())


SML/NJ can compile source code to a "heap file", witch can than be executed by the interpreter with arguments given (see this entry on for more information). The file should lock like this:


To compile the program, use ml-build This should create a "heap file" sources.x86-linux, depending on your architecture. The heap file can be executed with sml @SMLload=sources.x86-linux ARGUMENTS, or the script heap2exec can be used to make a single executable.


MLton compiles the source file directly to a executable by invoking mlton SOURCE_FILE.sml.


The following proc detects and removes argument-less (-b) and one-argument options from the argument vector.

proc getopt {_argv name {_var ""} {default ""}} {
upvar 1 $_argv argv $_var var
set pos [lsearch -regexp $argv ^$name]
if {$pos>=0} {
set to $pos
if {$_var ne ""} {set var [lindex $argv [incr to]]}
set argv [lreplace $argv $pos $to]
return 1
} else {
if {[llength [info level 0]] == 5} {set var $default}
return 0

Usage examples:

getopt argv -sep sep ";"     ;# possibly override default with user preference
set verbose [getopt argv -v] ;# boolean flag, no trailing word

Searching with -regexp allows to specify longer mnemonic names, so it still succeeds on longer flags, e.g.

$ myscript.tcl -separator '\t' ...


The Argh class provides command line parsing, it can do actions during parsing, leave it for you to do after parsing, print errors, the option list, short or long options, with or without option args, etc.

File myprogram.zkl:

var ip;
argh := Utils.Argh(
T("v","v","print version",fcn{println("Version stub")}),
T("+ip","","get IP address",fcn(arg){ip=arg}),
parsedArgs := argh.parse(vm.arglist);
println("Unparsed stuff: ",argh.loners);
println("The IP address is ",ip);
foreach option,arg in (parsedArgs){
switch(option) {
case("z") { println("zazzle") }
zkl myprogram nc -v -n -z --ip 1-1000
Version stub
Unparsed stuff: L("nc","1-1000")
The IP address is
zkl myprogram nc -v -n -z --ip
Option "ip" is missing an arg
  --ip <arg>: get IP address
  --n (-n) : ignored
  --v (-v) : print version
  --z (-z) : zazzle