piotrmurach/tty

TTY is a toolbox for developing beautiful command line clients in Ruby with a fluid interface for gathering input, querying terminal properties and displaying information.

All too often libraries that interact with terminals create their own interface logic that gathers input from users and displays information back. Many times utility files are created that contain methods for reading system or terminal properties. Shouldn't we focus our energy on building the actual client?

Building terminal tools takes time. I believe that modular components put together in a single package with project scaffolding will help people build things faster and produce higher quality results. It is easy to jump start a new project with available scaffolding and mix and match components to create new tooling.

• Jump-start development of your command line app the Unix way with scaffold provided by teletype.
• Fully modular, choose out of many components to suit your needs or use any 3rd party ones.
• All tty components are small packages that do one thing well.
• Fully tested with major ruby interpreters.

Add this line to your application's Gemfile to install all components:

gem 'tty'

or install a particular component:

gem 'tty-*'

And then execute:

$ bundle

Or install it yourself as:

$ gem install tty

• 1. Overview
• 2. Bootstrapping
  • 2.1 new command
    • 2.1.1 --author, -a flag
    • 2.1.2 --ext flag
    • 2.1.3 --license, -l flag
    • 2.1.4 --test, -t flag
  • 2.2 add command
    • 2.2.1 --args flag
    • 2.2.2 --desc flag
    • 2.2.3 --force flag
  • 2.3 Working with Commands
  • 2.4 Arguments
  • 2.5 Description
  • 2.6 Options and Flags
  • 2.7 Global Flags
  • 2.8 Working with Subcommands
• 3. Components
• 4. Contributing

TTY provides you with commands and many components to get you onto the path of building awesome terminal applications in next to no time.

To simply jump start a new command line application use teletype executable:

$ teletype new app

Move in to your new app, and then add more commands:

$ cd app
$ teletype add config

Throughout the rest of this guide, I will assume a generated application called app, that you are in the working directory of 'app/', and a newly created bare command config.

Running teletype new [app-name] will bootstrap an entire project file structure based on the bundler gem command setup enhanced by additional files and folders related to command application development.

For example, to create a new command line application called app do:

$ teletype new app

The output will contain all the files that have been created during setup:

Creating gem 'app'
    create app/Gemfile
    create app/.gitignore
    create app/lib/app.rb
    create app/lib/app/version.rb
    ...

In turn, the following files and directories will be generated in the app folder familiar to anyone who has created a gem beforehand:

▾ app/
├── ▾ exe/
│   └── app
├── ▾ lib/
│   ├── ▾ app/
│   │   ├── ▸ commands/
│   │   ├── ▸ templates/
│   │   ├── cli.rb
│   │   ├── command.rb
│   │   └── version.rb
│   └── app.rb
├── CODE_OF_CONDUCT.md
├── Gemfile
├── LICENSE.txt
├── README.md
├── Rakefile
└── app.gemspec

By convention the file lib/app/cli.rb provides the main entry point to your command line application:

module App
  class CLI < Thor
    # Error raised by this runner
    Error = Class.new(StandardError)

    desc 'version', 'app version'
    def version
      require_relative 'version'
      puts "v#{App::VERSION}"
    end
    map %w(--version -v) => :version
  end
end

This is where all your application commands and subcommands will be defined.

Teletype uses Thor as an option parsing library by directly inheriting from it.

And also by convention the start method is used to parse the command line arguments inside the app executable:

App::CLI.start

Run the new command with --help or -h flag to see all available options:

$ teletype new --help
$ teletype new -h

Execute teletype to see all available commands.

The teletype generator can inject name into documentation for you:

$ teletype new app --author 'Piotr Murach'

To specify that teletype should create a binary executable (as exe/GEM_NAME) in the generated project use the --ext flag. This binary will also be included in the GEM_NAME.gemspec manifest. This is disabled by default, to enable do:

$ teletype new app --ext

The teletype generator comes prepackaged with most popular open source licenses: agplv3, apache, bsd2, bsd3, gplv2, gplv3, lgplv3, mit, mplv2, custom. By default the mit license is used. To change that do:

$ teletype new app --license bsd3

The teletype comes configured to work with rspec and minitest frameworks which are the only two acceptable values. The GEM_NAME.gemspec will be configured and appropriate testing directory setup. By default the RSpec framework is used.

$ teletype new app --test=minitest
$ teletype new app -t=minitest

Once application has been initialized, you can create additional command by using teletype add [command-name] task:

$ teletype add config
$ teletype add create

This will add create.rb and config.rb commands to the CLI client:

▾ app/
├── ▾ commands/
│   ├── config.rb
│   └── create.rb
├── ▸ templates/
│   ├── ▸ config/
│   └── ▸ create/
├── command.rb
├── cli.rb
└── version.rb

Then you will be able to call the new commands like so:

$ app config
$ app create

The commands require you to specify the actual logic in their execute methods.

Please note that command names should be provided as camelCase or snake_case. For example:

$ teletype add addConfigCommand   # => correct
$ teletype add add_config_command # => correct
$ teletype add add-config-command # => incorrect

You can specify that teletype should add a command with a variable number of arguments using the --args flag. The --args flag accepts space delimited variable names. To specify required argument use a string name, for an optional argument pass name = nil enclosed in quote marks and any variable number of arguments needs to be preceded by asterisk:

$ teletype add config --args name           # required argument
$ teletype add config --args "name = nil"   # optional argument
$ teletype add config --args *names         # variadic argument

For more in-depth usage see 2.4 Arguments.

Every generated command will have a default description 'Command description...', however whilst generating a command you can and should specify a custom description to provide more context with --desc flag:

$ teletype add config --desc 'Set and get configuration options'

For more in-depth usage see 2.5 Description.

If you wish to overwrite currently implemented command use --force flag:

$ teletype add config --force

Running

teletype add config

a new command config will be added to commands folder creating the following files structure inside the lib folder:

▾ app/
├── ▾ commands/
│   └── config.rb
├── ▾ templates/
│   └── ▸ config/
├── cli.rb
├── command.rb
└── version.rb

The lib/app/cli.rb file will contain generated command entry which handles the case where the user asks for the config command help or invokes the actual command:

module App
  class CLI < Thor
    desc 'config', 'Command description...'
    def config(*)
      if options[:help]
        invoke :help, ['config']
      else
        require_relative 'commands/config'
        App::Commands::Config.new(options).execute
      end
    end
  end
end

And the lib/app/commands/config.rb will allow you to specify all the command logic. In the Config class which by convention matches the command name, the execute method provides a place to implement the command logic:

module App
  module Commands
    class Config < App::Command
      def initialize(options)
        @options = options
      end

      def execute
        # Command logic goes here ...
      end
    end
  end
end

Notice that Config inherits from App::Cmd class which you have full access to. This class is meant to provide all the convenience methods to lay foundation for any command development. It will lazy load many tty components inside helper methods which you have access to by opening up the lib/app/command.rb file.

For example in the lib/app/command.rb file, you have access to prompt helper for gathering user input:

# The interactive prompt
#
# @see http://www.rubydoc.info/gems/tty-prompt
#
# @api public
def prompt(**options)
  require 'tty-prompt'
  TTY::Prompt.new(options)
end

or a command helper for running external commands:

# The external commands runner
#
# @see http://www.rubydoc.info/gems/tty-command
#
# @api public
def command(**options)
  require 'tty-command'
  TTY::Command.new(options)
end

You have full control of the file, so you can use only the tty components that you require. Please bear in mind that all the components are added by default in your app.gemspec which you can change to suite your needs and pick only tty components that fit your case.

A command may accept a variable number of arguments.

For example, if we wish to have a config command that accepts a location of configuration file, then we can run teletype add command passing --args flag:

$ teletype add config --args file

which will include the required file as an argument to the config method:

module App
  class CLI < Thor
    desc 'config FILE', 'Set and get configuration options'
    def config(file)
      ...
    end
  end
end

Similarly, if we want to generate command with two required arguments, we run teletype add command with --args flag that can accept variable names delimited by space character:

$ teletype add set --args name value

will generate the following:

module App
  class CLI < Thor
    desc 'set NAME VALUE', 'Set configuration option'
    def set(name, value)
      ...
    end
  end
end

If we want to have a command that has an optional argument, for example, the file argument is an optional argument in the config command, then you need to enclose --args argument in parentheses:

$ teletype add config --args 'file = nil'

In well behaved command line application, any optional argument in a command will be enclosed in square brackets:

module App
  class CLI < Thor
    desc 'config [FILE]', 'Set and get configuration options'
    def config(file = nil)
      ...
    end
  end
end

If you intend for your command to accept any number of arguments, you need to prefix such argument with an asterisk. For example, if we wish to accept many configuration names:

$ teletype add get --args *names

which will append ... to the argument description:

module App
  class CLI < Thor
    desc 'get NAMES...', 'Get configuration options'
    def get(*names)
      ...
    end
  end
end

You can mix and match all the above styles of arguments definitions:

$ teletype add config --args file *names

Use the desc method call to describe your command when displayed in terminal. There are two arguments to this method. First, specifies the command name and the actual positional arguments it will accept. The second argument is an actual text description of what the command does.

For example, given the command config generated in add command section, we can add description like so:

module App
  class CLI < Thor
    desc 'config [FILE]', 'Set and get configuration options'
    def config(file = nil)
      ...
    end
  end
end

Running app executable will include the new description:

Commands:
  app config [FILE]  # Set and get configuration options

To provide long form description of your command use long_desc method.

module App
  class CLI < Thor
    desc 'config [FILE]', 'Set and get configuration options'
    long_desc <<-DESC
      You can query/set/replace/unset options with this command.

      The name is an optional key separated by a dot, and the value will be escaped.

      This command will fail with non-zero status upon error.
    DESC
    def config(file = nil)
      ...
    end
  end
end

Running app config --help will produce the following output:

Usage:
  app config

You can query/set/replace/unset options with this command.

The name is an optional key separated by a dot, and the value will be escaped.

This command will fail with non-zero status upon error.

Flags and options allow to customize how particular command is invoked and provide additional configuration.

To specify individual flag or option use method_option before the command method. All the flags and options can be accessed inside method body via the options hash.

Available metadata for an option are:

• :aliases - A list of aliases for this option.
• :banner — A description of the value if the option accepts one.
• :default - The default value of this option if it is not provided.
• :lazy_default — A default that is only passed if the cli option is passed without a value.
• :desc - The short description of the option, printed out in the usage description.
• :required — Indicates that an option is required.
• :type - :string, :hash, :array, :numeric, :boolean
• :enum — A list of allowed values for this option.

The values for :type option are:

• :boolean is parsed as --option
• :string is parsed as --option=VALUE or --option VALUE
• :numeric is parsed as --option=N or --option N
• :array is parsed as --option=one two three or --option one two three
• :hash is parsed as --option=name:string age:integer

For example, you wish to add an option that allows you to add a new line to a configuration file for a given key with a value thus being able to run app config --add name value. To do this, you would need to specify :array type for accepting more than one value and :banner to provide meaningful description of values:

method_option :add, type: :array, banner: "name value", desc: "Adds a new line the config file. "

The above option would be included in the config method like so:

module App
  class CLI < Thor
    desc 'config [<file>]', 'Set and get configuration options'
    method_option :add, type: :array, banner: "name value",
                        desc: "Adds a new line the config file. "
    def config(*)
      ...
    end
  end
end

Running app help config will output new option:

Usage:
  app config [<file>]

  Options:
    [--add=name value]  # Adds a new line the config file.

You can also specify an option as a flag without an associated value. Let us assume you want to be able to open a configuration file in your system editor when running app config --edit or app config -e. This can be achieved by adding the following option:

method_option :edit, type: :boolean, aliases: ['-e'],
                     desc: "Opens an editor to modify the specified config file."

And adding it to the config method:

module App
  class CLI < Thor
    desc 'config [<file>]', 'Set and get configuration options'
    method_option :edit, type: :boolean, aliases: ['-e'],
                         desc: "Opens an editor to modify the specified config file."
    def config(*)
      ...
    end
  end
end

Next, running app help config will produce:

Usage:
  app config [<file>]

Options:
      [--add=name value]     # Adds a new line the config file.
  -e, [--edit], [--no-edit]  # Opens an editor to modify the specified config file.

You can use method_options as a shorthand for specifying multiple options at once.

method_options %w(list -l) => :boolean, :system => :boolean, :local => :boolean

Once all the command options and flags have been setup, you can access them via options hash in command file lib/app/commands/config.rb:

module App
  module Commands
    class Config < App::Command
      def initialize(options)
        @options = options
      end

      def execute
        if options[:edit]
          editor.open('path/to/config/file')
        end
      end
    end
  end
end

You can specify an option or a flag that is applicable to all commands and subcommands within a given class by using the class_option method. This method takes exactly the same parameters as method_option for an individual command. The options hash in a given command will always include a global level flag information.

For example, if you want a global flag debug that is visible to all commands in your tool then you need to add it to your CLI class like so:

module App
  class CLI < Thor
    class_option :debug, type: :boolean, default: false, desc: 'Run in debug mode'

    ...
  end
end

If your tool grows in complexity you may want to add more refined behaviour for each individual command, a subcommand is a great choice to accomplish this. For example, git utility and its git remote command have various subcommands add, rename, remove, set-url, prune and so on that themselves accept many options and arguments.

The teletype executable allows you to easily create new subcommands by issuing the same add command that is also used for generating commands. The only difference is that you need to provide a command name together with a subcommand name. For example, let's say we want the config with a set subcommand with a description and two positional arguments name and value:

$ teletype add config set --desc 'Set configuration option' --args name value

This will add set.rb command to the commands/config folder:

▾ app/
├── ▾ commands/
│   ├── ▾ config/
│   │   └── set.rb
│   └── config.rb
├── ▾ templates/
│   └── ▾ config/
│       └── ▸ set/
├── cli.rb
├── command.rb
└── version.rb

The lib/app/cli.rb will contain code that registers config namespace with our CLI root application:

module App
  class CLI < Thor
    require_relative 'commands/config'
    register App::Commands::Config, 'config', 'config [SUBCOMMAND]', 'Set configuration option'
  end
end

The lib/app/commands/config.rb will contain code that handles dispatching subcommands to the Config instance:

# frozen_string_literal: true

require 'thor'

module App
  module Commands
    class Config < Thor

      namespace :config

      desc 'set NAME VALUE', 'Set configuration option'
      def set(name, value)
        if options[:help]
          invoke :help, ['set']
        else
          require_relative 'config/set'
          App::Commands::Config::Set.new(name, value, options).execute
        end
      end
    end
  end
end

And finally, the lib/app/commands/config/set.rb will contain the actual set command implementation:

# frozen_string_literal: true

require_relative '../../command'

module App
  module Commands
    class Config
      class Set < App::Command
        def initialize(name, value, options)
          @name = name
          @value = value
          @options = options
        end

        def execute
          # Command logic goes here ...
        end
      end
    end
  end
end

You can now run your command in terminal:

bundle exec app config set debug true

Note that it is not possible to add subcommands to an existing command. Attempting to do so will currently cause teletype to crash. The reason why it is not possible to add subcommands to existing commands is that it is impossible for tty to distinguish between normal arguments to a command, and subcommands for that command. However, you may very well add multiple subcommands one after another.

The TTY allows you to mix & match any components you need to get your job done. The command line applications generated with teletype executable references all of the below components.

You can contribute by posting feature requests, evaluating the APIs or simply by hacking on TTY components:

1. Fork it
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Add some feature')
4. Push to the branch (git push origin my-new-feature)
5. Create new Pull Request

This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.

This project uses EditorConfig to maintain consistent tabbing and file formats. Consider installing the plugin for your editor to help maintain proper code formatting.

Copyright (c) 2012 Piotr Murach. See LICENSE.txt for further details.