Getopt::ArgParse(3) Parsing command line arguments with a richer and


version 1.0.6


use Getopt::ArgParse;
$ap = Getopt::ArgParse->new_parser(
prog => 'MyProgramName',
description => 'This is a program',
epilog => 'This appears at the bottom of usage',
# Parse an option: '--foo value' or '-f value'
$ap->add_arg('--foo', '-f', required => 1);
# Parse a boolean: '--bool' or '-b' using a different name from
# the option
$ap->add_arg('--bool', '-b', type => 'Bool', dest => 'boo');
# Parse a positonal option.
# But in this case, better using subcommand. See below
$ap->add_arg('command', required => 1);
# $ns is also accessible via $ap->namespace
$ns = $ap->parse_args(split(' ', 'test -f 1 -b'));
say $ns->command; # 'test'
say $ns->foo; # false
say $ns->boo; # false
say $ns->no_boo; # true - 'no_' is added for boolean options
# You can continue to add arguments and parse them again
# $ap->namespace is accumulatively populated
# Parse an Array type option and split the value into an array of values
$ap->add_arg('--emails', type => 'Array', split => ',');
$ns = $ap->parse_args(split(' ', '--emails [email protected],[email protected],[email protected]'));
# Because this is an array option, this also allows you to specify the
# option multiple times and splitting
$ns = $ap->parse_args(split(' ', '--emails [email protected],[email protected] --emails [email protected]'));
# Below will print: [email protected]|[email protected]|[email protected]|[email protected]|[email protected]|[email protected]
# Because Array types are appended
say join('|', $ns->emails);
# Parse an option as key,value pairs
$ap->add_arg('--param', type => 'Pair', split => ',');
$ns = $ap->parse_args(split(' ', '--param a=1,b=2,c=3'));
say $ns->param->{a}; # 1
say $ns->param->{b}; # 2
say $ns->param->{c}; # 3
# You can use choice to restrict values
$ap->add_arg('--env', choices => [ 'dev', 'prod' ],);
# or use case-insensitive choices
# Override the previous option with reset
$ap->add_arg('--env', choices_i => [ 'dev', 'prod' ], reset => 1);
# or use a coderef
# Override the previous option
choices => sub {
die "--env invalid values" if $_[0] !~ /^(dev|prod)$/i;
reset => 1,
# subcommands
$ap->add_subparsers(title => 'subcommands'); # Must be called to initialize subcommand parsing
$list_parser = $ap->add_parser(
help => 'List directory entries',
description => 'A multiple paragraphs long description.',
'--verbose', '-v',
type => 'Count',
help => 'Verbosity',
help => 'depth',
$ns = $ap->parse_args(split(' ', 'list -v'));
say $ns->current_command(); # current_command stores list,
# Don't use this name for your own option
$ns =$ap->parse_args(split(' ', 'help list')); # This will print the usage for the list command
# help subcommand is automatically added for you
say $ns->help_command(); # list
# Copy parsing
$common_args = Getopt::ArgParse->new_parser();
type => 'Bool',
help => 'Dry run',
$sp = $ap->add_parser(
aliases => [qw(rm)], # prog remove or prog rm
parents => [ $command_args ], # prog rm --dry-run
# Or copy explicitly
$sp = $ap->add_parser(
aliases => [qw(cp)], # prog remove or prog rm
$sp->copy_args($command_parser); # You can also copy_parsers() but in this case
# $common_parser doesn't have subparsers


Getopt::ArgParse, Getopt::ArgParse::Parser and related classes together aim to provide user-friendly interfaces for writing command-line interfaces. A user should be able to use it without looking up the document most of the time. It allows applications to define argument specifications and it will parse them out of @AGRV by default or a command line if provided. It implements both named arguments, using Getopt::Long for parsing, and positional arguments. The class also generates help and usage messages.

The parser has a namespace property, which is an object of ArgParser::Namespace. The parsed argument values are stored in this namespace property. Moreover, the values are stored accumulatively when parse_args() is called multiple times.

Though inspired by Python's argparse and names and ideas are borrowed from it, there is a lot of difference from the Python one.


This is the underlying parser that does the heavylifting.

Getopt::ArgParse::Parser is a Moo class.


  my $parser = Getopt::ArgParse->new_parser(
    help        => 'short description',
    description => 'long description',

The former calls Getopt::ArgParser::Parser->new to create a parser object. The parser constructor accepts the following parameters.

All parsers are created with a predefined Bool option --help|-h. The program can choose to reset it, though.

  • prog

    The program's name. Default $0.

  • help

    A short description of the program.

  • description

    A long description of the program.

  • namespace

    An object of Getopt::ArgParse::Namespace. An empty namespace is created if not provided. The parsed values are stored in it, and they can be refered to by their argument names as the namespace's properties, e.g. $parser->namespace->boo. See also Getopt::ArgParse::Namespace

  • parser_configs

    The Getopt::Long configurations. See also Getopt::Long

  • parents

    Parent parsents, whose argument and subparser specifications the new parser will copy. See copy() below

  • error_prefix

    Customize the message prefixed to error messages thrown by Getop::ArgParse, default to 'Getopt::ArgParse: '

  • print_usage_if_help

    Set this to false to not display usage messages even if --help is on or the subcommand help is called. The default behavior is to display usage messages if help is set.

add_arg, add_argument, add_args, and add_arguments

    [ '--foo', required => 1, type => 'Array', split => ',' ],
    [ 'boo', required => 1, nargs => '+' ],

The object method, arg_arg or the longer version add_argument, defines the specification of an argument. It accepts the following parameters.

add_args or add_arguments() is to add multiple multiple arguments.

name or flags

Either a name or a list of option strings, e.g. foo or -f, --foo.

If dest is not specified, the name or the first option without leading dashes will be used as the name for retrieving values. If a name is given, this argument is a positional argument. Otherwise, it's an option argument.

Hyphens can be used in names and flags, but they will be replaced with underscores '_' when used as option names. For example:

    $parser->add_argument( [ '--dry-run', type => 'Bool' ]);
    # command line: prog --dry-run
    $parser->namespace->dry_run; # The option's name is dry_run

A name or option strings are following by named parameters.


The name of the attribute to be added to the namespace populated by parse_args().

type => $type

Specify the type of the argument. It can be one of the following values:

  • Scalar

    The option takes a scalar value.

  • Array

    The option takes a list of values. The option can appear multiple times in the command line. Each value is appended to the list. It's stored in an arrayref in the namespace.

  • Pair

    The option takes a list of key-value pairs separated by the equal sign '='. It's stored in a hashref in the namespace.

  • Bool

    The option does not take an argument. It's set to true if the option is present or false otherwise. A 'no_bool' option is also available, which is the negation of bool().

    For example:

        $parser->add_argument('--dry-run', type => 'Bool');
        $ns = $parser->parse_args(split(' ', '--dry-run'));
        print $ns->dry_run; # true
        print $ns->no_dry_run; # false
  • Count

    The option does not take an argument and its value will be incremented by 1 every time it appears on the command line.


split should work with types 'Array' and 'Pair' only.

split specifies a string by which to split the argument string e.g. if split => ',', a,b,c will be split into [ 'a', 'b', 'c' ].When split works with type 'Pair', the parser will split the argument string and then parse each of them as pairs.

choices or choices_i

choices specifies a list of the allowable values for the argument or a subroutine that validates input values.

choices_i specifies a list of the allowable values for the argument, but case insenstive, and it doesn't allow one to use a subroutine for validation.

Either choices or chioces_i can be present or completely omitted, but not both at the same time.


The value produced if the argument is absent from the command line.

Only one value is allowed for scalar argument types: Scalar, Count, and Bool.


Whether or not the command-line option may be omitted (optionals only). This has no effect on types 'Bool' and 'Count'. An named option is marked by the question mark ? in the generated usage, e.g.
    --help, -h             ? show this help message and exit

This parameter is ignored for Bool and Count types for they will already have default values.


A brief description of what the argument does.


A name for the argument in usage messages.


Set reset to override the existing definition of an option. This will clear the value in the namspace as well.

nargs - Positional option only

This only instructs how many arguments the parser consumes. The program still needs to specify the right type to achieve the desired result.

  • n

    1 if not specified

  • ?

    1 or 0

  • +

    1 or more

  • *

    0 or many. This will consume the rest of arguments.


  $namespace = $parser->parse_args(@command_line);

This object method accepts a list of arguments or @ARGV if unspecified, parses them for values, and stores the values in the namespace object.

A few things may be worth noting about parse_args().

First, parsing for Named Arguments is done by Getopt::Long

Second, parsing for positional arguments takes place after that for named arguments. It will consume what's still left in the command line.

Finally, the Namespace object is accumulatively poplulated. If parse_args() is called multiple times to parse a number of command lines, the same namespace object is accumulatively populated. For Scalar and Bool options, this means the previous value will be overwrittend. For Pair and Array options, values will be appended. And for a Count option, it will add on top of the previous value.

In face, the program can choose to pass a already populated namespace when creating a parser object. This is to allow the program to pre-load values to a namespace from conf files before parsing the command line.

And finally, it does NOT display usage messages if the argument list is empty. This may be contrary to many other implementations of argument parsing.


  @argv = $parser->argv; # called after parse_args

Call this after parse_args() is invoked to get the unconsumed arguments. It's up to the application to decide what to do if there is a surplus of arguments.

The Namespace Object

The parsed values are stored in a namespace object. Any class with the following three methods:

  * A constructor new()
  * set_attr(name => value)
  * get_attr(name)

can be used as the Namespace class.

The default one is Getopt::ArgParse::Namespace. It uses autoload to provide a readonly accessor method using dest names to access parsed values. However, this is not required for user-defined namespace. So within the implementation, $namespace->get_attr($dest) should always be used.

Subcommand Support

Note only one level of subcommand parsing is supported. Subcommands cannot have subcommands.

Call add_subparsers() first to initialize the current parser for subcommand support. A help subcommand is created as part of the initialization. The help subcommand has the following options:

required positional arguments:
     COMMAND      ? Show the usage for this command optional named arguments:
    --help, -h     ? show this help message and exit
    --all, -a      ? Show the full usage

Call add_parser() to add a subparser for each subcommand. Use the parser object returned by add_parser() to add the options to the subcommand.

Once subcommand support is on, if the first argument is not a flag, i.e. starting with a dash '-', the parser's parse_args() will treat it as a subcommand. Otherwise, the parser parses for the defined arguments.

The namespace's current_command() will contain the subcommand after parsing successfully.

Unlike arguments, subparsers cannot be reset.


    title       => 'Subcommands',
    description => 'description about providing subcommands',

add_subparsers must be called to initialize subcommand support.

  • title

    A title message to mark the beginning of subcommand usage in the usage message

  • description

    A general description appearing about the title


  $subparser = $parser->add_parser(
     aliases     => [qw(ls)],
     help        => 'short description',
     description => 'a long one',
     parents => [ $common_args ], # inherit common args from
                                  # $common_args
  • $command

    The first argument is the name of the new command.

  • help

    A short description of the subcommand.

  • description

    A long description of the subcommand.

  • aliases

    An array reference containing a list of command aliases.

  • parents

    An array reference containing a list of parsers whose specification will be copied by the new parser.


   $subparser = $parser->get_parser('ls');

Return the parser for parsing the $alias command if exsist.

Copying Parsers

A parser can copy argument specification or subcommand specifciation for existing parsers. A use case for this is that the program wants all subcommands to have a command set of arguments.



Copy argument specification from the $parent parser



Copy parser specification for subcommands from the $parent parser



Copy both arguments and subparsers.

Usage Messages and Related Methods


  $usage = $parser->format_usage;

Return the formatted usage message for the whole program in an array reference.



Print the usage message returned by format_usage().


  $usage = $parser->format_command_usage($subcommand);

Return the formatted usage message for the command in an array reference.



Print the usage message returned by format_command_usage(). If $command is not given, it will first try to use $self->namespace->help_command, which will be present for the help subcommand, and then $self->namespace->current_command.


Mytram <[email protected]> (original author)


Robbin Bonthond (rbonthond@github) Adam Pfeiffer (apinco@github)


This software is Copyright (c) 2015 by Mytram.

This is free software, licensed under:

  The Artistic License 2.0 (GPL Compatible)