NAME

CLI::Framework - Build standardized, flexible, testable command-line applications

OVERVIEW

CLI::Framework ("CLIF") provides a framework and conceptual pattern for building full-featured command line applications. It intends to make this process simple and consistent. It assumes the responsibility of implementing details that are common to all command-line applications, making it possible for new applications adhering to well-defined conventions to be built without the need to repeatedly write the same command-line interface code.

For instance, a complete application supporting commands and subcommands, with options and arguments for the application itself as well as its commands, can be built by writing concise, understandable code in packages that are easy to test and maintain. The classes can focus on implementation of unique aspects essential to the command's purpose without being concerned with the many details involved in building an interface around those commands. This methodology for building command-line applications also establishes a valuable standard for an organization (or an individual developer).

LEARNING CLIF: RECOMMENDATIONS

CLIF has a rich set of features and offers many alternative approaches to building applications, but if you are new to using it, you may want a succinct introduction. For this reason, the CLI::Framework::Tutorial is provided and is the recommended starting point.

After you gain a basic understanding, the other documents can be used as references.

MOTIVATION

There are a few other distributions on CPAN intended to simplify building modular command line applications. I have not found any that meet my requirements, which are documented in DESIGN GOALS AND FEATURES.

DESIGN GOALS AND FEATURES

CLIF was designed to offer the following features...

  • A clear conceptual pattern for creating command-line applications

  • Guiding documentation and examples

  • Convenience for simple cases, flexibility for complex cases

  • Support for both non-interactive and interactive modes (with almost no additional work -- define the necessary hooks and both modes will be supported)

  • A design that naturally encourages MVC applications: decouple data model, control flow, and presentation

  • Commands that can be shared between applications (and uploaded to CPAN)

  • The possibility to share some components with MVC web applications

  • Validation of application options

  • Validation of command options and arguments

  • A model that encourages easily-testable applications

  • A flexible means to provide usage/help information for the application as a whole and for individual commands

  • Support for subcommands that can be added as a natural extension to commands

  • Support for recursively-defined subcommands (sub-sub-...commands to any level of depth)

  • Support for aliases to commands and subcommands

  • Allow Application and [sub]commands to be defined inline (some or all packages involved may be defined in the same file) or split across multiple files

  • Support the concept of a default command for the application

  • Exception handling that allows individual applications to define custom exception handlers

  • Performance. Core framework code should load as quickly as a simple script; individual commands should be initialized only when invoked.

CONCEPTS AND DEFINITIONS

  • Application Script - The wrapper program that invokes the CLIF Application's run method. The file it is defined in may or may not also contain the definition of Application or Command packages.

  • Metacommand - An application-aware command. Metacommands are subclasses of CLI::Framework::Command::Meta. They are identical to regular commands except they hold a reference to the application within which they are running. This means they are able to "know about" and affect the application. For example, the built-in command "Menu" is a Metacommand because it needs to produce a list of the other commands in its application.

    In general, your commands should be designed to operate independently of the application, so they should simply inherit from CLI::Framework::Command. This encourages looser coupling. However, in exceptional cases, the use of Metacommands is warranted (For an example, see the built-in "Menu" command).

  • Non-interactive Command - In interactive mode, some commands need to be disabled. For instance, the built-in "console" command (which is used to start interactive mode, presenting a command menu and responding to user selections) should not be presented as a menu option in interactive mode because it is already running. You can designate which commands are non-interactive by overriding the noninteractive_commands method.

  • Registration of commands - Each CLIF application defines the commands it will support. These may be built-in CLIF commands or custom CLIF commands. These commands are lazily "registered" as they are called upon for use.

APPLICATION RUN SEQUENCE

When a command of the form:

$ app [app-opts] <cmd> [cmd-opts] { <cmd> [cmd-opts] {...} } [cmd-args]

examples:

        app      |             [app-opts]            { <cmd>       |   [cmd-opts]    } [cmd-args]
`````````````````|```````````````````````````````````|`````````````|`````````````````|``````````````
$ examples/queue |--qin=/tmp/qfile --qout=/tmp/qfile | enqueue     | --tag=x --tag=y | 'item'
`````````````````|```````````````````````````````````|`````````````|`````````````````|``````````````
$ gen-report     |              --html               | stats       |  --role=admin   |
                 |                                   | usage       |   --time='2d'   | '/tmp/stats.html'
````````````````````````````````````````````````````````````````````````````````````````````````````

...causes your application script, <app>, to invoke the run method in your application class, CLI::Framework::Application performs the following actions:

  1. Parse the command request

  2. Validate application options

  3. Initialize application

  4. Prepare command

  5. Invoke command pre-dispatch hook

  6. Dispatch command

These steps are explained in more detail below...

Request parsing

Parse the application options [app-opts], command name <cmd>, command options [cmd-opts], and the remaining part of the command line, which includes command arguments [cmd-args] for the last command and may include multiple subcommands. Everything between the inner brackets ({ ... }) represents recursive subcommand processing -- the "..." represents another string of "<cmd> [cmd-opts] {...}".

The second example above shows a command request that requires recursive subcommand processing. The command might cause an HTML report to be generated with usage statistics for admin users (of some application) for the past two days, writing the report to a file. In one line, it would look like this:

$ gen-report --html stats --role=admin usage --time='2d' '/tmp/stats.html'

This fictional gen-report application could be designed with such an interface because it could offer various types of reports (as opposed to the statistics report). There might be other statistics reports (as opposed to 'usage'). The stats might be available for users with other roles. The usage report might need to accept custom time frames.

CLIF allows you to choose whether various parts of your data should be supplied as options or as arguments -- these interface decisions are left to your discretion. CLIF also makes it easy to validate command requests and to provide usage information so users know what to change if a command request fails validation.

In general, if a command request is not well-formed, it is replaced with the default command and any arguments present are ignored. The default command prints a help or usage message (you may change this behavior if desired).

Validation of application options

Your application class can optionally define the validate_options method.

If your application class does not override this method, validation is skipped -- any received options are considered to be valid.

Application initialization

Your application class can optionally override the init method. This is a hook that can be used to perform any application-wide initialization that needs to be done independent of individual commands. For example, your application may use the init method to connect to a database and store a connection handle which may be needed by some or all of the commands in your application.

Preparing the command

The requested command is now loaded (if not already done). The command's cache is set (using a reference to the same cache object used by the application).

Command pre-dispatch

Your application class can optionally have a pre_dispatch method that is called with one parameter: the Command object that is about to be dispatched.

Dispatching the command

CLIF uses the dispatch method to actually dispatch a specific command. That method is responsible for running the command or delegating responsibility to a subcommand, if applicable.

INTERACTIVITY

After building your CLIF application, in addition to basic non-interactive functionality, you will instantly benefit from the ability to (optionally) run your application in interactive mode. A readline-enabled application command console with an event loop, a command menu, and built-in debugging commands is provided by default.

Inside interactive mode, only steps 4, 5, and 6 above (APPLICATION RUN SEQUENCE) are performed for each command request.

Supporting interactivity in your application is as simple as adding the built-in command CLI::Framework::Command::Console to your command_map.

BUILT-IN COMMANDS INCLUDED IN THIS DISTRIBUTION

This distribution comes with some default built-in commands, and more CLIF built-ins can be installed as they become available on CPAN.

Use of the built-ins is optional in most cases, but certain features require specific built-in commands (e.g. the Help command is a fundamental feature of all applications and the Menu command is required in interactive mode). You can override any of the built-ins.

A new application that does not override the command_map hook will include all of the built-ins listed below.

The existing built-ins and their corresponding packages are as follows:

help: Print application or command-specific usage messages

CLI::Framework::Command::Help

Note: This command is registered automatically. All CLIF applications must have the "help" command defined (though this built-in can replaced by your subclass to change the "help" command behavior or to do nothing if you specifically do not want a help command).

list: Print a list of commands available to the running application

CLI::Framework::Command::List

dump: Show the internal state of a running application

CLI::Framework::Command::Dump

tree: Display a tree representation of the commands that are currently registered with the running application

CLI::Framework::Command::Tree

alias: Display the command aliases that are in effect for the running application and its commands

CLI::Framework::Command::Alias

console: Invoke CLIF's interactive mode

CLI::Framework::Command::Console

CLI::Framework::Command::Menu

Note: This command is registered automatically when an application runs in interactive mode. This built-in may be replaced by a user-defined "menu" command, but any command class to be used for the "menu" command MUST be a subclass of this one.

CLIF ARCHITECTURE AT A GLANCE

The class diagram below shows the relationships of the major classes of CLI Framework, including some of their methods. This is not intended to be a comprehensive diagram, only an aid to understanding CLIF at a glance.

class diagram from docs/images dir

SEE ALSO

CLI::Framework::Application

CLI::Framework::Command

CLI::Framework::Tutorial

LICENSE AND COPYRIGHT

Copyright (c) 2009 Karl Erisman (kerisman@cpan.org). All rights reserved.

This is free software; you can redistribute it and/or modify it under the same terms as Perl itself. See perlartistic.

AUTHOR

Karl Erisman (kerisman@cpan.org)