command-line-interface

command-line-interface is a foundation for CLI applications.

Status

Category	Status
Version
Dependencies
Dev dependencies
Build
License

Installation

$ npm install command-line-interface

Quick Start

First you need to add a reference to command-line-interface to your application:

const { runCli, Command } = require('command-line-interface');

If you use TypeScript, use the following code instead:

import { runCli, Command } from 'command-line-interface';

Before being able to run a CLI, you must create a command. A command represents a piece of logic callable from the command line. In that sense, a command has a name, a description, a list of optionDefinitions, and a handle function. Optionally, you may also specify remarks to show in the command's help:

const hello = {
  name: 'hello',
  description: 'Say hello on the command line.',
  remarks: `
    If you don't specify a name, 'Jane' will be used as default.
  `,

  optionDefinitions: [
    {
      name: 'name',
      description: 'The name to use.',
      type: 'string',
      alias: 'n',
      defaultValue: 'Jane'
    }
  ],

  handle ({ options }) {
    console.log(`Hello ${options.name}!`);
  }
};

Then you can use the runCli function to run your CLI application. For that, hand over the command and provide the command line parameters, usually taken from process.argv:

await runCli({ rootCommand: hello, argv: process.argv });

Now you can run your CLI application and use it to say hello:

$ node app.js
Hello Jane!

Since you configured a name option, you may adjust the name by providing the --name or -n flag:

$ node app.js --name Jenny
Hello Jenny!

You may also ask for help by using the --help flag, which is automatically available:

$ node.js app.js --help

The handle function may be synchronous or asynchronous, so you may use async depending on your needs. If you throw an error from within that function, the CLI application will end with exit code 1, and print the exception's message as well as its stack trace to the terminal.

Writing commands using TypeScript

If you are using TypeScript, you may want to use types for the command and its options. First, define an interface for the command's options, such as:

export interface HelloOptions {
  name: string;
}

Please note that it is highly recommended to put the interface for a command's options into a file of its own instead of putting it into the same file as the command it belongs to. This make it easier to extend and re-use interfaces when implementing sub-commands.

Additionally, provide the correct type for the variable that contains the command:

const hello: Command<HelloOptions> = {
  // ...
};

Everything else stays the same, but now you will have type support.

Defining options

As you have already seen, you can define options for commands. An option needs to have at least a name and a type, with the following types being supported and verified:

• boolean
• number
• string

Additionally, you may provide a description and an alias. While the former is used when printing the usage, the latter is used to give a single-character alias for an otherwise lengthy option. You have seen this with the alias n for the option name in the examples above.

Sometimes it make sense to allow providing an option more than once. For that, add the multiple option to the option definitions, and set it to true. This lets you provide the appropriate option value multiple times on the command line:

$ node.js app.js --name Jane --name Jenny

If an option is mandatory, set the isRequired property to true. For optional options, it usually makes sense to specify a default value. For that, use the defaultValue property and set it to the desired value. This can be seen in the example above as well.

If you want to give a dedicated name to the value, you can set it using the parameterName property. This sometimes makes sense, to e.g. show off that a parameter is not just a string, but a url or another domain-specific concept.

You can also define whether an option is the default option of a command by setting the defaultOption property to true. In this case you can skip the option's flag, and just provide its value.

Last but not least, you may specify a validate function for an option definition. Inside this function you are free to do whatever you need to do to ensure that the option's given value is valid. However, if you throw an exception from within this function, command-line-interface aborts the command's execution, and shows an error message:

const hello = {
  name: 'hello',
  description: 'Say hello on the command line.',

  optionDefinitions: [
    {
      name: 'name',
      description: 'The name to use.',
      type: 'string',
      alias: 'n',
      defaultValue: 'Jane',
      validate (value) {
        if (value.length > 20) {
          throw new Error('The name must be less than 20 characters.');
        }
      }
    }
  ],

  handle ({ options }) {
    console.log(`Hello ${options.name}!`);
  }
};

Implementing sub-commands

For more complex applications, you might want to setup a variety of commands, and let the user decide which one to run. Typical CLI applications that make use of this concept are git and docker. The command that you hand over to runCli is the so-called root command.

To define additional commands, you need to define them as sub-commands. Actually, you can also define sub-command on sub-commands, and nest them arbitrarily deep (although this doesn't make too much sense). To define a sub-command, first define a command as already known, and then add them using the subcommands property of an already defined command:

const hello = {
  name: 'hello',
  description: 'Say hello on the command line.',

  optionDefinitions: [
    {
      name: 'name',
      description: 'The name to use.',
      type: 'string',
      alias: 'n',
      defaultValue: 'Jane'
    }
  ],

  handle ({ options }) {
    console.log(`Hello ${options.name}!`);
  },

  subcommands: {
    anotherCommand,
    anotherOtherCommand
  }
};

All the flags that are given on the command line to the parent command are handed over to the handle function of the sub-command.

Please note that if you are using TypeScript, make sure to derive the sub-commands options interface from the parent commands' one.

Printing the usage

By default, all commands automatically come with a --help flag. Additionally, every application automatically gets a dedicated help command that you can use to show help for each command.

Sometimes you may want to show the usage manually from within a command. For that, add the parameters getUsage and ancestors to your handle function, run the getUsage function and hand over an object with the property commandPath containing an array with the path to the name of the current command. You may use the ancestors array to get the list of names of the parent commands:

const hello = {
  name: 'hello',
  description: 'Say hello on the command line.',

  optionDefinitions: [
    // ...
  ],

  handle ({ options, getUsage, ancestors }) {
    console.log(getUsage(
      { commandPath: [ ...ancestors, 'hello' ] }
    ));
  }
};

Customizing handling errors

By default, command-line-interface takes care of handling any errors that occur. However, sometimes you may want to customize handling errors, e.g. to format them before displaying them. For that, hand over the optional handlers parameter to the runCli function, and provide functions for the error types you want to customize:

await runCli({
  rootCommand: hello,
  argv: process.argv,
  handlers: {
    commandFailed ({ ex }) {
      // ...
    },

    commandUnknown ({ unknownCommandName, recommendedCommandName, ancestors }) {
      // ...
    },

    optionInvalid ({ optionDefinition, reason }) {
      // ...
    },

    optionMissing ({ optionDefinition }) {
      // ...
    },

    optionUnknown ({ optionName }) {
      // ...
    }
  }
});

Please note that if you do not provide all handlers, the remaining ones stick to the default behavior.

Running quality assurance

To run quality assurance for this module use roboter:

$ npx roboter