A developer-friendly lightweight replacement for the config
module that works with custom config directories, pluggable parsers and with many other handy features.
Notice of change of ownership: Starting version 3.0.0 this package has changed it's owner and goals. The old version (2.0.3) is still available on npm via npm install [email protected]
and on github.com/ksmithut/configly. Thank you.
$ yarn add configly
or
$ npm install --save configly
Original config
module is convenient and easy to start with library, but in the same time being focused that much on "easy" it lacks certain features to be a "developer friendly" library.
This package is addressing those issues, while keeping easy of use and feature set on par with the original module.
Configly provides rich configuration options, as well as set of sane (and battle-tested) defaults, which allow you to get a lot from basic setups.
To simply replace your current config
setup, add following to your files:
var config = require('configly');
console.log(config.my.combined.options);
It will load .js
and .json
files from ./config
folder,
relative to the current working directory (process.cwd()
).
It will cache the result, so files will be read only once per process.
Out of the box configly
supports only two formats (.js
and .json
), but developers can add their own parsers and support for more formats (e.g. .ini
, .yaml
, .cson
).
var config = require('configly/configure');
// more parsers
var ini = require('ini');
// Note: recommended to use `cson-parser` over `cson`
// due to https://github.com/groupon/cson-parser/issues/56
// Update:
// recommended to use `cson` over `cson-parser`
// due to https://github.com/groupon/cson-parser/issues/74
var cson = require('cson');
var yaml = require('js-yaml');
var properties = require('properties');
var json5 = require('json5');
// assemble new parsers list
// order doesn't matter since they
// will be alphabetically sorted
var configObj = config({
parsers: {
ini : ini.parse,
// have it as a wrapper to prevent extra arguments leaking
cson : function(str) { return cson.parse(str); },
yaml : function(str) { return yaml.safeLoad(str); },
// same options as used within `config` module
properties: function(str) { return properties.parse(str, {namespaces: true, variables: true, sections: true}); },
// use json5 instead of `JSON.parse`
json : json5.parse,
// keep the original one
js : config.parsers.js
}
});
Or create new instance with new defaults
var configNew = config.new({
parsers: {
ini : ini.parse,
// have it as a wrapper to prevent extra arguments leaking
cson : function(str) { return cson.parse(str); },
yaml : function(str) { return yaml.safeLoad(str); },
// same options as used within `config` module
properties: function(str) { return properties.parse(str, {namespaces: true, variables: true, sections: true}); },
// use json5 instead of `JSON.parse`
json : json5.parse,
// keep the original one
js : config.parsers.js
}
});
// use it as usual
var configObj = configNew();
You can export newly created instance and reuse it all over your app,
it won't be affected by other instances of the configly
even if it
used in dependencies of your app, or you module is part of the bigger app,
that uses configly
.
To load config files from a custom directory, just specify it as the first argument.
var config = require('configly/configure')({directories: './etc'}); // `require('configly')('etc');` would work the same way
It will load files from the etc
folder relative to the current working directory,
by providing absolute path, you can make sure exact location of the config files,
which is useful to libraries meant to be used within larger applications
and for command line apps that could be invoked from different directories.
var path = require('path');
var config = require('configly/configure')({directories: path.join(__dirname, 'etc')});
Or you can set up new directory as default one
and invoke configly
without custom arguments
from within other files.
// config.js
var path = require('path');
var configly = require('configly/configure');
module.exports = configly.new({
defaults: {
directories: path.join(__dirname, 'etc')
}
});
// app.js
var config = require('./config')();
It is possible to load files from more than one config directory within one application/module.
var path = require('path');
var configly = require('configly/configure');
// "inline"
var oneConfig = configly({directories: [
path.join(__dirname, 'app-config'),
path.join(__dirname, 'rules-config')
]});
Or creating new default
module.exports = configly.new({
defaults: {
directories: [
path.join(__dirname, 'app-config'),
path.join(__dirname, 'rules-config')
]
}
});
Configly scans config directory for number of config files, then loads and merges them in specific order.
It always starts with default
file, as it's the place for all project's defaults, it will try load default
with extensions for provided parsers (out of the box it will be .js
and .json
).
Then it will search for environment specific files, specified via NODE_ENV
(in case it's omitted falls back to development
). After that it searches for files related to the runtime's hostname. It closes off "static" config files with local
file, which usually being git-ignored and meant for local overrides of the checked in configuration options.
After that configly proceeds to search for dynamic/meta config files – custom-include-files
and custom-environment-variables
...
Last file to read would be runtime
file, and meant for deployment target local configuration rather than being checked into the code repository.
Example ordered list of files that will be read, in NODE_ENV=production
environment on dev382.corp.example.com
host:
default.js
default.json
default-production.js
default-production.json
production.js
production.json
dev382.js
dev382.json
dev382-production.js
dev382-production.json
dev382.corp.example.com.js
dev382.corp.example.com.json
dev382.corp.example.com-production.js
dev382.corp.example.com-production.json
local.js
local.json
local-production.js
local-production.json
custom-include-files.js
custom-include-files.json
custom-include-files-production.js
custom-include-files-production.json
custom-environment-variables.js
custom-environment-variables.json
custom-environment-variables-production.js
custom-environment-variables-production.json
runtime.js
runtime.json
runtime-production.js
runtime-production.json
Also configly
can load config data from custom files (along with the default list),
handling them the same way – search for supported extensions and within specified directory(-ies).
var config = configly({
files: configly.files.concat(['custom_file_A', 'custom_file_B'])
});
Following code will completely replace list of filenames.
var config = configly({
directories: '/etc',
files: [
'totally',
'custom',
'list',
'of',
'files'
]
});
For use cases where you need to load config files within the app, but augment it with server/environment specific config you can add absolute path filename to the files list.
var config = configly({
directories: path.join(__dirname, 'config'),
// `configly` will search for `/etc/consul/env.js`, `/etc/consul/env.json`, etc
// after loading default files from local `config` folder
files: configly.files.concat('/etc/consul/env')
});
For bigger apps / more complex configs, combination of multiple directories and custom files would provide needed functionality.
var path = require('path')
, configly = require('configly/configure')
, package = require('./package.json')
;
module.exports = configly.new({
defaults: {
directories: [
// will search local config directory
path.join(__dirname, 'config'),
// and augment with same files
// from environment specific folder
'/etc/consul'
]
},
// also will try to load config files matching current app name
// e.g. 'my-app.js', `my-app.json`, `my-app-production.js`, `my-app-production.json`,
// from both local config folder and `/etc/consul`
files: configly.files.concat(package.name)
});
It allows to combine environment variables within a single entry (e.g. "endpoint": "${REMOTE_HOST}:${REMOTE_PORT}"
), which helps to keep application config consumption simple and to the point, and gives greater control over different environments, like using Docker environment variables for linked containers.
custom-environment-variables.json
:
{
"Customers":
{
"dbPassword" :"ENV_VAR_NAME_AS_STRING",
"dbPassword2" :"${ENV_VAR_NAME_VARIABLE}"
},
"CombinedWithText" : "${VARPART1}:${VARPART2}",
"BothVarsEmptyWillBeSkippedCompletely" : "${NOEXISTING} + another ${EMPTY} var"
}
Since environment variables only passed as string, it creates inconvenience for handling non-string values, like boolean, falsy or complex values.
Modifiers will come handy to "pre-parse" values passed from the environment,
for example to pass false
to your config or if some library requires casted parameters for number value.
custom-environment-variables.json
:
{
"api":
{
"mocksEnabled" :"json FETCH_MOCKS_ENABLED", // 'false' -> false
"timeout" :"json FETCH_TIMEOUT", // '1500' -> 1500
"cutOffDate" : "date DATE_CUTOFF", // '2018-08-08' -> new Date('2018-08-08')
"nullSignal": "json PASSED_NULL" // 'null' -> null
},
"combinedBoolCastedToString" : "${json FETCH_MOCKS_ENABLED}", // 'false' -> false -> 'false'
"combinedNumberCastedToString" : "${json FETCH_TIMEOUT}", // '1500' -> 1500 -> '1500'
"combinedDateCastedToString" : "${date DATE_CUTOFF}", // '2018-08-08' -> new Date('2018-08-08') -> 'Tue Aug 07 2018 17:00:00 GMT-0700 (Pacific Daylight Time)'
"combinedNullSignalSkip": "${json PASSED_NULL}", // 'null' -> null -> 'null'
"nullNoModifierIsString": "PASSED_NULL", // 'null' -> 'null'
"combinedNullNoModifierIsString": "${PASSED_NULL}" // 'null' -> 'null' -> 'null'
}
For better integration with other configuration systems and third-party tools that generate configs, configly
can "mount" custom config files into specified entries, for example if you need to pull webpack manifest files into app's config).
custom-include-files.json
:
{
"MyModule": {
"WillBeIgnored": "NONEXISTENT_FILE",
"and": {
"here": {
"some": {
"nested": {
"structure": "custom_file"
}
}
}
}
},
"AnotherModule": {
"yes": {
"it": {
"should": {
"be": {
"filename": {
"no": {
"extension": "../include_files/another_file"
}
}
}
}
}
}
}
}
In the above example it will check for following custom files, (given NODE_ENV
equals production
):
custom_file.js
custom_file.json
custom_file-production.js
custom_file-production.json
../include_files/another_file.js
../include_files/another_file.json
../include_files/another_file-production.js
../include_files/another_file-production.json
And will attach their content to the respective branches.
Developers can extend configly
's functionality with custom hooks for config files. For example if you need to pull in sensitive into your config and it couldn't be stored on disk among rest of the configuration data, you can add your own hooks to relace placeholders or attach extra values with data pull from the environment or from a secure storage. All hooks should be synchronous.
var customHooks = merge(configly.hooks, {
// will be applied to `local`, `local-development` and other `local*` files
// expects `config` object of the parsed file
local: function(config)
{
iterate(config, function(value, key, node)
{
// increments each value by 4
node[key] = value + 4;
});
// expected to return updated object
return config;
}
});
var config = configly(configDir, {hooks: customHooks});
To fully replicate config
's behavior and provide easy way to include static customized config
in your app files, without resorting to require('../../../config')
, you can create virtual node module,
based on the custom config file within your app.
config/config.js
var path = require('path')
, configly = require('configly/configure')
, ini = require('ini')
, yaml = require('js-yaml')
;
// run configly once with inlined modifiers
// and have it as node-cached module
module.exports = configly({
defaults: {
directories: [
// will search local config directory
'./etc',
// and augment with same files
// from environment specific folder
'/etc/consul'
]
},
// also will try to load config files matching current app name
// e.g. `my-app.ini`, 'my-app.js', `my-app.json`, `my-app.yaml`,
// `my-app-production.ini`, `my-app-production.js`, `my-app-production.json`, `my-app-production.yaml`,
// from both local config folder and `/etc/consul`
files: configly.files.concat('my-app'),
// throw in custom parsers as well
parsers: {
ini : ini.parse,
// have it as a wrapper to prevent extra arguments leaking
yaml : function(str) { return yaml.safeLoad(str); }
}
});
config/package.json
{
"name": "config",
"version": "0.0.0",
"main": "config.js"
}
"dependencies": {
"config": "./config"
},
Now npm will copy ./config/
files into node_modules
and execute ./config/config.js
on first require,
making it's output available for every file of your app, via var config = require('config')
.
This way migration of your app from config
module to configly
will be limited to a few extra lines of code,
while providing more functionality and better separation of concerns out of the box.
app/lib/file.js
var config = require('config');
console.log('value', config.my.data.from.etc.consul.myApp.json);
And similar to the method described above, it could serve as a handy toolkit for your own config module, below you can see simple example of the custom config module that allows for flexible extension on per project basis while keeping standard approach within your organization.
var path = require('path')
, configly = require('configly/configure')
// get application's name for per application custom config file,
// useful for having per environment specific config files
// separate from code base of the application
, appName = process.env['NODE_MY_ORG_CONFIG_APPNAME'] || require(path.resolve('./package.json')).name
// by default, it will search local (CWD) `./config` directory
// augmented with same files from environment specific folder `/etc/consul`,
// and could overridden from outside of the app, for cases when app/module itself
// being used as the base for another service, or in test/integration environments
, directories = process.env['NODE_MY_ORG_CONFIG_PATH'] || './config:/etc/consul'
// use standard path separator `:`
, separator = ':'
;
// run configly once with inlined modifiers
// and have it as node-cached module
module.exports = configly({
defaults: {
directories: directories.split(separator)
},
// also will try to load config files matching current app name
// e.g. 'my-app.js', `my-app.json`,
// `my-app-production.js`, `my-app-production.json`,
// from both local `config` folder and from `/etc/consul`.
// Also `appName` could be composite value, like:
// `my_app:my_group:my_org` for more flexibility
files: configly.files.concat(appName.split(separator))
});
Above code could be published on npm (internal or public), as your organization specific config module @myorg/config
,
and used within all your organization's projects:
var config = require('@myorg/config');
console.log(config.env.specific.value);
For more examples check out test directory.
Main differences between configly
and config
:
- Configly provides deterministic (and controllable) order of the config files it loads from.
- Configly provides deterministic (and controllable) order of the file extensions it loads from.
- Configly provides post-load hooks for config files, (e.g.
custom-environment-variables
andcustom-include-files
work via this mechanism). - Configly provides ability to combine environment variables within one entry (e.g.
"endpoint": "${REMOTE_HOST}:${REMOTE_PORT}"
). - Configly supports built-in and custom modifiers for environment variables overrides (e.g.
"loggingEnabled": "boolean LOGGING_ENABLED_BOOLEAN_AS_STRING"
) - Configly provides ability to "mount" custom config files into specified entries (e.g. useful to pull webpack manifest files into app's config).
- Configly provides access to the underlying functions and defaults, allowing to utilize parts of the functionality for greater flexibility.
- Configly doesn't read/write
NODE_CONFIG
environment variable. - Configly doesn't pollute your logs with warnings of non-existent files, it will either throw (if couldn't read/parse a file) or be silent.
- Configly doesn't provide
get
,has
methods, it always returns pure js (POJO) object. - Configly doesn't auto-strip comments from JSON files, instead use
parsers: {json: (str) => json5.parse(str)}
.
Configly is released under the MIT license.