CommOp

Solution for complex cli interfaces.

Installation

$ npm install -g commop

Writing another module

This is not an option parser module. If you writing small script, which handles one simple task, please use another module, there is many option parsers: commander, dashdash, minimist or yargs.

For some of my tasks I had to write complicated cli interface. With global options, options per command, command sets, localized messages, environment as fallback/override and so on. So, I decided to write helper module which can use any option parser module for parsing argv and then build a cli I want.

Example

{
	"options": {
		"verbose":  {"type": "boolean", "global": true, "alias": "v"},
		"arduino":  {"type": "string",  "global": true, "alias": "A"},
		"include":  {"type": "string",  "alias": "I"},
		"define":   {"type": "string",  "alias": "D"},
		"port":     {"type": "string",  "alias": "p"},
		"baudrate": {"type": "number",  "alias": "r"},
	},
	"commands": {
		"compile":  {"options": ["inc", "define"], "run": "compile"},
		"upload":   {"options": ["port", "baudrate"], "run": "upload"},
		"platform": {
			"sub": {
				"add":    {"run": "addPlatform"},
				"remove": {"run": "removePlatform"}
			}
		}
	}
}

var commop = require ('commop');
var config = require ('./commop.json');

var launcher = new commop (config);

function compile (cmd, next) {
	if (cmd.options.inc) {
		// TODO: make something with inc
	}
	// code to compile
}

module.exports.compile = compile;

launcher.start (); // process.argv, module.exports

I had to create cli for arduino supporting commands like compile, upload and platform. Each task should have verbose and arduino options. And compile and upload commands should have sketch and board options. Afterall compile must be configurable via -D, includes via -I options. Upload must support port and baudrate options. platform is actually set of commands, like platform add <> and platform remove <>.

Option parsing

I've tried a lot of option parsing modules, and found out that it is matter of preference which module to use. For my own tasks (almost) any option parser can handle job. But some needs advanced validation, in some cases no requirements is a must, and some loves pirates.

commop uses bundled minimist module and supports yargs and dashdash interfaces.

Interface

There is no fancy interface to add options. Module takes configuration and launches tasks tied to that configuration.

var commop = require ('commop');
var config = require ('./commop.json');

var launcher = new commop (cmdConfig);

// getting command configuration from argv
var cmd;

// now you can command configuration
cmd = launcher.findCommand ();
// or
cmd = launcher.findCommand (process.argv);
// or
cmd = launcher.findCommand (process.argv);

/*
cmd: {
	config: <command config from configuration object>
	branch: [ 'library' ], // actual command branch
	options: { verbose: false, debug: false }, // options, accessible by that command
	failedOptions: {}, // if option is required and missed or some options conflicts
	errors: [] // another errors
}
*/

// launch tasks associated with that command
launcher.start ();

// launch tasks associated with that command
launcher.start (process.argv, require.main, function (cmd, data) {
	// every task is completed
});

Features

Command handler(s)

Using list of tasks with run key

Each command must have a handler, described by run key in command configuration. If handler is an array, then tasks launched one after one. Task laso can be a promise or function, which return a promise.

/**
* task function
* @param {Object}   cmd  command object
* @param {Object}   data shared object between tasks in current handler
* @param {Function} next call next if done
*/
function task (cmd, data, next) {

}

Command objects usually contains config key with associated configuration structure, branch key with list of parsed command names, positional with positional parameters, options — list of applicable options, also failedOptions and errors

Data should be modified and returned via next callback or promise's resolve. Next task will be launched regardless of return status of previous task. If your task rely on data from previous command, assert data section.

Those tasks can be object methods, you just have to provide an origin as parameter to the start call. If origin is not provided, require.main is used instead (it is your main module exports).

var theProgram = new Program ();

launcher.start (null, theProgram, function (cmd, data) {
});

Using external script with script key

If you defined script key for command, this script will run. You can pass options as environment variables. Script's stdout and stderr you'll get via data.scriptStderr and data.scriptStdout. If command configuration contains anyway key, then script error will be written to the data.scriptError and callback will be called.

var testConfig2 = {
	options: {
		BBB: {type: "string"},
		DDD: {type: "string"}
	},
	commands: {
		node: {
			script: nodePath + " -e 'console.log (process.argv)' AAA=${BBB} CCC=${DDD}",
			options: ["BBB", "DDD"]
		}
	}
};

var co = new CommOp (testConfig2);

co.start (null, null, function (cmd, data) {
	console.log (data.scriptStderr, data.scriptStdout)
});

Also, if you need to launch different scripts on different OS'es, use next format:

script: {
	win32:  nodePath + " -e 'console.log (process.argv)' AAA=%BBB% CCC=%DDD%",
	default: nodePath + " -e 'console.log (process.argv)' AAA=${BBB} CCC=${DDD}"
}

Usage/help localization

Each message is localizable,for more information please refer test/04-l10n.js

Option sharing

Option can be global or shared between specified commands.

Implied, conflicting and required options

Examples at test/02-options.js

Usage generator

If your commands and options have properly defined descriptions, commop will generate usage automatically. By default, if there is no command specified, usage will be displayed automatically. But you can override this behaviour.

// usage is displayed automatically unless showUsage is false
launcher.showUsage = false;

// usage will be displayed automatically if there is no commands in argv
launcher.start ([], null, function (cmd, data) {
	// we cannot find any commands
	if ("usage" in cmd) {
		// do something like this:
		if (cmd.branch[0] === 'xxx') {
			launcher.helpForCommand (["xxx"]);
		}
	}

});

// or, you can generate usage

var usageString = launcher.usage();

Help generator for command

Command help displayed automatically if you added a help keyword before actual command. Also, you can provide your own help handler by setting run key for a help command in config.

// command help will be displayed automatically if you prefixed command with help keyword
launcher.start (["help", "cmd"], null, function (cmd, data) {
	if ("usage" in cmd) {
	}
});

// or, you can generate command help programmatically

var helpString = launcher.helpForCommand (["cmd"]);

Configuration

Configuration is a javascript object. Options and commands configurations under options and commands keys.

Configuration for option

"options": {
	"verbose":  {
		"type": "boolean",
		"global": true,
		"alias": "v",
		"description": "be verbose",
		"default": false,
		"env": "VERBOSE"
	},
	"arduino":  {"type": "string",  "global": true, "alias": "A"},
	"include":  {"type": "string",  "alias": "I"},
	"define":   {"type": "string",  "alias": "D"},
	"port":     {"type": "string",  "alias": "p"},
	"baudrate": {"type": "number",  "alias": "r"},
}

Configuration keys:

• type handling depends on option parsing module

• global options available for all commands

• Single alias or list of aliases

• description needed for usage/help generation

• default value

• env keys to use for that option

Configuration for command

"commands": {
	"compile":  {"options": ["inc", "define"], "run": "compile"},
	"upload":   {"options": ["port", "baudrate"], "run": "upload"},
	"platform": {
		"sub": {
			"add":    {"run": "addPlatform"},
			"remove": {"run": "removePlatform"}
		}
	}
}

Configuration keys:

• run task or task set to run command

• sub subcommands

• description needed for usage/help generation

• options is applicable options (implies is WIP):

// simplest way to use:
options: ["port", "baudrate"]

// complete:
options: {
	"port": {"required": true, "conflicts": "board"},
	"board": null,
	"baudrate": {"implies": "port"}
}

Note

Implied/conflicting options works quite badly with booleans. Using boolean options you will allways get value, either true or false even if that option is not present.

Additional flags

• envMode can be override or fallback. In fallback mode every missing option will be filled from environment variable. In override mode environment values takes over precedence over argv options.

• ignoreUnknownCommands allows to ignore unknown commands

See also

https://github.com/freeformsystems/cli-command complex, many deps, don't have option relationship, events, +sections, +pod

License

MIT