Perinci::CmdLine - Rinci/Riap-based command-line application framework


version 0.59


In your command-line script:

 use 5.010;
 use Log::Any '$log';
 use Perinci::CmdLine;

 our %SPEC;
 $SPEC{foo} = {
     v => 1.1,
     summary => 'Does foo to your computer',
     args => {
         bar => {
             schema=>['str*', {in=>[qw/aa bb cc/]}],
         baz => {
 sub foo {
     my %args = @_;
     $log->debugf("Arguments are %s", \%args);
     [200, "OK", $args{bar} . ($args{baz} ? "and $args{baz}" : "")];

 Perinci::CmdLine->new(url => '/main/foo')->run;

To run this program:

 % foo --help ;# display help message
 % LANG=id_ID foo --help ;# display help message in Indonesian
 % foo --version ;# display version
 % foo --bar aa ;# run function and display the result
 % foo --bar aa --debug ;# turn on debug output
 % foo --baz x  ;# fail because required argument 'bar' not specified

To do bash tab completion:

 % complete -C foo foo ;# can be put in ~/.bashrc
 % foo <tab> ;# completes to --help, --version, --bar, --baz and others
 % foo --b<tab> ;# completes to --bar and --baz
 % foo --bar <tab> ;# completes to aa, bb, cc

See also the peri-run script which provides a command-line interface for Perinci::CmdLine.


Perinci::CmdLine is a command-line application framework. It accesses functions using Riap protocol (Perinci::Access) so you get transparent remote access. It utilizes Rinci metadata in the code so the amount of plumbing that you have to do is quite minimal.

What you'll get:

  • Command-line options parsing

    Non-scalar arguments (array, hash, other nested) can also be passed as JSON or YAML (both will be attempted). For example, the tags argument is defined as 'array':

     % mycmd --tags '[foo, bar, baz]' ; # interpreted as YAML
     % mycmd --tags '["foo","bar"]'   ; # interpreted as JSON
     % mycmd --tags '[foo, bar, baz'  ; # fails both
  • Help message (utilizing information from metadata, supports translation)

  • Tab completion for bash (including completion from remote code)

  • Undo/redo/history

This module uses Log::Any and Log::Any::App for logging.

This module uses Moo for OO.


program_name => STR (default from $0)

url => STR

Required if you only want to run one function. URL should point to a function entity.

Alternatively you can provide multiple functions from which the user can select using the first argument (see subcommands).

summary => STR

If unset, will be retrieved from function metadata when needed.

subcommands => {NAME => {ARGUMENT=>...}, ...} | CODEREF

Should be a hash of subcommand specifications or a coderef.

Each subcommand specification is also a hash(ref) and should contain these keys: url. It can also contain these keys: summary (str, will be retrieved from function metadata if unset), tags (array of str, for categorizing subcommands), log_any_app (bool, whether to load Log::Any::App, default is true, for subcommands that need fast startup you can try turning this off for said subcommands), undo (bool, can be set to 0 to disable transaction for this subcommand; this is only relevant when undo attribute is set to true), pass_cmdline_object (bool, to override pass_cmdline_object attribute on a per-subcommand basis).

Subcommands can also be a coderef, for dynamic list of subcommands. The coderef will be called as a method with hash arguments. It can be called in two cases. First, if called without argument name (usually when doing --list) it must return a hashref of subcommand specifications. If called with argument name it must return subcommand specification for subcommand with the requested name only.

default_subcommand => NAME

If set, subcommand will always be set to this instead of from the first argument. To use other subcommands, you will have to use --cmd option.

extra_opts => HASH

Optional. Used to let program recognize extra command-line options. Currently not well-documented. For example:

 extra_opts => {
     halt => {
         handler => sub {
             my ($self, $val) = @_;
             $self->{_selected_subcommand} = 'shutdown';

This will make:

 % cmd --halt

equivalent to executing the 'shutdown' subcommand:

 % cmd shutdown

As an alternative to using this attribute, you can also subclass and override gen_common_opts(), like this:

 sub gen_common_opts {
     my ($self) = @_;
     my $go = $self->SUPER::gen_common_opts;
     push @$go, (
         halt => sub {
             $self->{_selected_subcommand} = 'shutdown';

exit => BOOL (default 1)

If set to 0, instead of exiting with exit(), run() will return the exit code instead.

log_any_app => BOOL

Whether to load Log::Any::App. Default is yes, or to look at LOG environment variable. For faster startup, you might want to disable this or just use LOG=0 when running your scripts.

custom_completer => CODEREF

Will be passed to Perinci::BashComplete's bash_complete_riap_func_arg. See its documentation for more details.

custom_arg_completer => CODEREF | {ARGNAME=>CODEREF, ...}

Will be passed to Perinci::BashComplete. See its documentation for more details.

dash_to_underscore => BOOL (optional, default 1)

If set to 1, subcommand like a-b-c will be converted to a_b_c. This is for convenience when typing in command line.

pass_cmdline_object => BOOL (optional, default 0)

Whether to pass special argument -cmdline containing the Perinci::CmdLine object to function. This can be overriden using the pass_cmdline_object on a per-subcommand basis.

Passing the cmdline object can be useful, e.g. to call run_help(), etc.

undo => BOOL (optional, default 0)

Whether to enable undo/redo functionality. Some things to note if you intend to use undo:

  • These command-line options will be recognized

    --undo, --redo, --history, --clear-history.

  • Transactions will be used

    use_tx=>1 will be passed to Perinci::Access, which will cause it to initialize the transaction manager. Riap requests begin_tx and commit_tx will enclose the call request to function.

  • Called function will need to support transaction and undo

    Function which do not meet qualifications will refuse to be called.

    Exception is when subcommand is specified with undo=>0, where transaction will not be used for that subcommand. For an example of disabling transaction for some subcommands, see bin/u-trash in the distribution.

undo_dir => STR (optional, default ~/.<program_name>/.undo)

Where to put undo data. This is actually the transaction manager's data dir.


new(%opts) => OBJ

Create an instance.

run() -> INT

The main routine. Its job is to parse command-line options in @ARGV and determine which action method (e.g. run_subcommand(), run_help(), etc) to run. Action method should return an integer containing exit code. If action method returns undef, the next action candidate method will be tried.

After that, exit() will be called with the exit code from the action method (or, if exit attribute is set to false, routine will return with exit code instead).


To do bash completion, first create your script, e.g. myscript, that uses Perinci::CmdLine:

 use Perinci::CmdLine;

then execute this in bash (or put it in bash startup files like /etc/bash.bashrc or ~/.bashrc for future sessions):

 % complete -C myscript myscript; # myscript must be in PATH


This module interprets the following result metadata keys:

cmdline.display_result => BOOL

If you don't want to display function output (for example, function output is a detailed data structure which might not be important for end users), you can set cmdline.display_result result metadata to false. Example:

 $SPEC{foo} = { ... };
 sub foo {
     [200, "OK", $data, {"cmdline.display_result"=>0}];

cmdline.page_result => BOOL

If you want to filter the result through pager (currently defaults to $ENV{PAGER} or less -FRS), you can set cmdline.page_result in result metadata to true.

For example:

 $SPEC{doc} = { ... };
 sub doc {
     [200, "OK", $doc, {"cmdline.page_result"=>1}];

cmdline.pager => STR

Instruct Perinci::CmdLine to use specified pager instead of $ENV{PAGER} or the default less or more.


PERINCI_CMDLINE_PROGRAM_NAME. Can be used to set CLI program name.


How does Perinci::CmdLine compare with other CLI-app frameworks?

The main difference is that Perinci::CmdLine accesses your code through Riap protocol, not directly. This means that aside from local Perl code, Perinci::CmdLine can also provide CLI for code in remote hosts/languages. For a very rough demo, download and run this PHP Riap::TCP server on your system. After that, try running:

 % peri-run riap+tcp://localhost:9090/terbilang --help
 % peri-run riap+tcp://localhost:9090/terbilang 1234

Everything from help message, calling, argument checking, tab completion works for remote code as well as local Perl code.

Aside from this difference, there are several others:

  • Non-OO, function-centric

    If you want OO, there are already several frameworks out there for you, e.g. App::Cmd, App::Rad, MooX::Cmd, etc.

  • Configuration file support is currently missing

    Coming soon, most probably will be based on Config::Ini::OnDrugs.

  • Also lacking is more documentation and more plugins

How to add support for new output format (e.g. XML, HTML)?

See Perinci::Result::Format.

How to accept input from STDIN (or files)?

If you specify 'cmdline_src' to 'stdin' to a 'str' argument, the argument's value will be retrieved from standard input if not specified. Example:

 use Perinci::CmdLine;
 $SPEC{cmd} = {
     v => 1.1,
     args => {
         arg => {
             schema => 'str*',
             cmdline_src => 'stdin',
 sub cmd {
     my %args = @_;
     [200, "OK", "arg is $args{arg}"];

When run from command line:

 % cmd --arg v1


Perinci, Rinci, Riap.

Other CPAN modules to write command-line applications: App::Cmd, App::Rad, MooseX::Getopt.


Steven Haryanto <>


This software is copyright (c) 2012 by Steven Haryanto.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.