/configue

⚙️ Configue All the Things.js 🔧

Primary LanguageJavaScriptOtherNOASSERTION

Configue

CONFIGUE ALL THE THINGS \o/

npm version Build Status Coverage Status License: MIT

Configue is a node.js config library to easily customize your app from argv, env, files and more.

It defines a conventional workflow to load a config from environment variables, command line arguments, files, that you can easily configure and extend.

Table of Content

About Configue

Configue builds up on nconf and its Hierarchical configuration system.

It defines a list of configuration steps that are executed in order. Every property defined on a steps will shadow the same key in the following steps.

Quoting nconf:

"The order in which you attach these configuration sources determines their priority in the hierarchy"

Here are the standard steps Configue does define:

  • overrides : properties that would take precedence on all other steps
  • argv : command line options
  • configueFile : file specified by --configue in argv if any.
  • env : environment variables
  • file : config files
  • defaults : default objects

The plugin loads the various configurations in order using predefined steps.

It starts by parsing argv then goes through the env and the files options and finishes by loading the default config objects if any. Hence why every option defined as an argument commandline will override defaults and environment variables.

If --configue option is specified, the config the specified file holds would be loaded after the argv and before the env. This is to enable you to save many options in many files, and specify at launch with options you want to use.

Installation

Just add configue has a dependency installing it with npm, or with yarn.

npm install --save configue

yarn add configue

Usage

How To

To use Configue you need first to create an instance passing the option to the Configue(opts) constructor. Resolving of the config is now done synchronously and automatically unless you specify the defer: true option, or if you opt in for an async resolve. In case of a problem with the configue options it will throw an Error.

See the following examples for concrete presentation.

Basic usage without customization

Basic Configue

const Configue = require('configue');

const configue = new Configue();

const who = configue.get('who', 'World');
console.log(`Hello ${who}`);

Basic Async Configue

const Configue = require('configue');

const configue = new Configue({async: true});
configue.resolve().then(() => {
  const who = configue.get('who', 'World');
  console.log(`Hello ${who}`);
});

Async resolve is necessary for some advanced features like async hooks and shortstop protocols.

Passing By Options

You can specify the who configue in different manners. Here are some:

node basic.js --who=Woman
# configue through Env
export who=Man ; node basic.js
who=Human node basic.js
node basic.js --configue=my-who-conf.json

The full example is available in the examples folder.

Retrieving values

You can retrieve values from the store in different manner, get is the most simple one.

Simple get

To retrieve a configue value, use the get method on the config holder. It takes has argument the key of the argument. For nested value you need to use : or . to deep access to value, or you can an array of keys.

It's also possible to specify a default value in case key is undefined.

configue.get('defined4sure');
configue.get('some:nested:value');
configue.get('some.other.nested.value');
configue.get(['yet', 'another', 'nested', 'value']);
configue.get('mightBeUndefined', 'default');

You can also retrieve a list of value with getAll, or the first non undefined value from a list with getFirst

configue.getAll('defined4sure', 'some:nested:value');
configue.getAll(['defined4sure', 'some:nested:value']);
configue.getAll(['some.other.nested.value', ['yet', 'another', 'nested', 'value']]);

configue.getFirst('defined4sure', 'some:nested:value');
configue.getFirst(['defined4sure', 'some:nested:value'], optionalDefaultValue);

Retrieving Specified Object

When you can to retrieve several values in the same time you can forge object so that they have structure you need.

Load and getObject for punctual retrieval

The two main methods are load and getObject

  • load by default return the whole merged config as an object. But you can give him a model that would be used to craft an object. The model is a object whose leaves are configue keys, or array of configue key: ex: {serverConfig: {host: 'app:server:host', port: 'PORT'}, ...}
  • getObject that takes a list of key, and return an object formed by key / values
const {serverConfig} = configue.load({
  serverConfig: {host: 'app:server:host', port: 'PORT'},
  extraOptions: '...'
});

const {file, prefix} = configue.getObject('file', 'prefix');
Models for frequent usage

There are case where the forged object are to be used several times, and you dont want to query them over and over. To do that you can predefined models in the configuration. These would be populated once during automatic resolved, and they would be made accessible under the _ key.

const configue = new Configue({
  models: {
    serverConfig: {host: 'app:server:host', port: 'PORT'},
    otherModel: {a: 'a', b: '...'}
  }
});
//...
console.log(configue._.serverConfig); // => host: ..., port: ...

Template string

One last way you can get config value is via the configue.template (aliased to configue.t). This is a template function you can prefix a template string. The interpolated values will be keys of the configue and then replaced by their value:

console.log(configue.t`I will say ${'salute'} to ${'who'}`);
// => I will say Hello to You
// (supposing called with --salute=Hello --who=You)

You can defined default values by passing a default object to the template method:

console.log(
  configue.t({times: 2, who: 'World'})`I will say ${'salute'} to ${'who'} ${'times'} times`
);
// => I will say Hello to You 2 times

Argv / Env direct access

For ease of the the argv and env can be directly accessible from the configue instance:

console.log(configue.argv.host);
console.log(configue.env.HOME);

Note that values are neither parsed nor transformed.

Usage with customization of the configuration workflow

Specifying Files

The files key can contain a single object or an array of objects containing a file key containing the path to the config file. The object can also reference a nconf plugin tasked with the formatting using the key format.

Starting from 1.0 the formatter to use can be automatically deduced for standard files. Supported extensions are json, yaml/yml but also properties/ini and json5 In that case you just need to specify the name of the file.

const Configue = require('configue');

const configueOptions = {
  disable: {argv: true},
  files: [
    {file: './config.json'},
    {
      file: './config.yaml',
      format: require('nconf-yaml')
    },
    'my-own.properties'
  ]
};

const configue = new Configue(configueOptions);

Note that if only one file is needed, its path can be directly given as options.

Using Protocalls(Shortstop protocols)

Protocall(originally Shortstop) is a library that help transform json values by interpreting their content. Quoting documentation:

it enables the use of protocols and handlers to enable identification and special handling of json values.

For instance value with the standard env and file protocol, env:MY_ENV_VARIABLE will be replaced with the value of MY_ENV_VARIABLE while value file:/some/path will be resolved with the content of the given file. For more details refer to the protocall project.

To enable it, you just need to have {async: true, protocall: true} in your Configue config object.

By default Configue comes empowered with the protocols from shortstop-handlers: env, file, path, exec, base64, require.

You can customize behavior of protocall by passing a config object as option:

  • You can add extra protocols via the protocols options, by passing an object {$protocolName: $handler}
  • You can also precise the baseDir for file, path, exec, require default handler. (default being current working directory)
  • If you don't want the default protocols, use the noDefaultProtocols option.
  • By default the Buffer are stringified by Configue, but you can choose to preserve them with the preserveBuffer option.

For an example of configuration refer to the following examples using this json file as part of the config.

Passing options to nconf to configure argv and env

You can provide options arguments to argv (yargsunderneath), and env in order to customize the behavior around command line argument and environment variables. For more in depth readings see nconf options here

const Configue = require('configue');

const configueOptions = {
  argv: {
    f: {
      alias: 'file',
      demandOption: true,
      default: '/etc/passwd',
      describe: 'x marks the spot',
      type: 'string'
    }
  },
  env: ['HOME', 'PWD'] // whitelist
};

const configue = new Configue(configueOptions);

Joint options of argv and env to process the values

It is possible to process the raw values you can get from the argv and env step, with the parse, separator ignorePrefix, normalize and transform options.

First you can specify to parse values with the parse option. Argv and Env value will be then parse, which is convenient to pass simple json from the command line.

A separator option is there to indicate the token that will be used to split a key and consider it as a nested value. This affects both the argv and env step. The value can be either a string, or a regexp such as '__' or /__|--/

With ignorePrefix you can list of prefix for argv and env variable you want to be removed. This is particulary useful with environment variables that are prefixed with your app name.

Also a normalize option enables you to make the variable names uniform with the same case, while using the idiomatic case for the argv flag name and env variable. (for instance --my-var and MY_VAR). This option accept as config the name of case function of lodash, the most useful being camelCase which will transform our both variable into myVar as we would like name the javascript variable. (other options are kebabCase, startCase, snakeCase,upperCase, lowerCase)

If you have more complex processing of the env/arg variable name or value, you can use the transform option, which accept a function ({key, value}) => ({key:someKey, value:someValue}) that will be passed to nconf. (cf nconf doc) This will happen after the ignore prefix, and before the case normalisation.

Disabling Steps

The argv and env steps can be skipped using the disable object in options.

const configue = new Configue({disable: {argv: true}});
// ...

There is no disabling for overrides, files and default; you just have to don't provide the matching option.

Step hooks

Every step (overrides, argv, env, files, defaults) has a post hook available. Those can be defined using the postHooks key and accept a function that take nconf. In async mode those hooks can be asynchronous by returning a Promise.

The special hooks first enables you to respectively apply a function on nconf at the very beginning.

const configue = new Configue({
  postHooks: {
    first: function first(nconf) {
      // Your code here
    },
    overrides: function postOverrides(nconf) {
      // Your code here
    },
    argv: function postArgv(nconf) {
      // Your code here
    }
  }
});
// Your code here

Custom Workflow

If needed you can have your full custom configuration workflow, simply by providing an object with the single key customWorkflow attached to a function taking the nconf object, and a done callback.

const configueOptions = {
  customWorkflow(nconf, done) {
    // my own config setting
  }
};

const configue = new Configue(configueOptions);

If you use a yargs instance, you can assign it to nconf._yargs so that argv is directly accessible from `configue.argv

Loading into Hapi

Thought Configue is usable without hapi, (it was originally just a _Hapi_ plugin), it can be easily loaded in hapi` to have the configue being easily accessible from the server, or the request.

To do this, you need to register the plugin. It takes care to resolve the config if was not done due to defer.

const Hapi = require('hapi');
const Configue = require('configue');

const configue = new Configue({some: 'complex config with a model connexion'});

const server = new Hapi.Server();
server.connection(configue._.connexion); // note usage of the model connexion with port in it.
server.register({register: configue.plugin()}, err => {
  // starting the server or else

  // access to the config
  const config = server.configue('some'); // => 'config'
  const configGet = server.configue.get('some'); // => 'config'
  // Any other call to server.configue.getAsync/getFirst/getAll/getObject/template/load
  // ...
});

A more complete example is available in examples folder.

Note it's possible to provide to configue.plugin() a decorateName so that you use a custom accessor on server or request.

Warning: the original plugin is made for the pre 17 version of hapi. If you are using hapi@17 or beyond, please retrive the plugin with the plugin17()` method as you can see in the example server.

Loading into express

Configue can also be loaded into express via it's middleware you can obtain by configue.middleware() you just have to feed to app.use()

A example is available in the examples folder.

Configuration Recap

Configue can be configured in two different way. Either using a config object or using a fluent builder.

Configuration Object

Here is a recap of how the configuration should look like. All options are optional:

  • customWorkflow: a function manipulating the nconf. This option is exclusive of all others
  • argv: Config object for yargv, a map of config key with an object values (alias, demandOption, default,describe, type)
  • env: The options for the nconf.env method that can be:
    • an array of string, the whitelisted env method
    • an object with key: match, `whitelist
  • disable: A object with key argv and/or env attach to a boolean indicated whether the step should be disable.
  • files: file or list of files. (object file, format)
  • defaults: Object of key mapped to default values. (or array of them)
  • overrides: Object of key mapped to overrides values.
  • required: list of key that are required one way or another
  • postHooks: an object of (step: function hook) step being one of first, overrides, argv, env, files defaults
  • parse: boolean to request parsing of argv/env value
  • transform: a function to process argv and env values
  • normalize: the case name in which you want keys to be converted
  • ignorePrefix: a prefix or list of them you want to be remove from key name
  • protocall/shortstop: to activate and customize the protocall/shortstop protocols. (prefer protocall, shortstop is to be deprecated)
  • async: to activate async mode which will defer resolve
  • defer: to defer automatic resolve in sync mode

For more details you can see the internals.schema in the configue-core.js file around the line 60

Fluent builder

Instead to use the configuration object provided to the Configue constructor, you can use the fluent builder. This consist in chaining a list of configuration methods before to retrieve the instance to a get method or via a resolve method.

Here is a simple example:

const configue = Configue.defaults({a: 1, b: '2'})
  .parse(true)
  .normalize('camelCase')
  .get();

You can provide a portion of option with the withOption method as you can see in this example using resolve:

Configue.defaults({a: 1, b: '2'})
  .withOptions({parse: true, normalize: 'camelCase'})
  .protocall(true)
  .resolve(configue => {
    // here goes your code
  });

Here is the builder function list, the function name being the name of the key in he object config (except the postHooks function and withOptions): argv, async, customWorkflow, defaults, overrides, disable, env, files, required, transform, parse, normalize, separator, protocall and firstHook, overridesHook, argvHook, envHook, filesHook, defaultsHook, withOptions