The modern way to handle and validate your application configuration with any standard-schema-compliant validation library.
Figue is a modern configuration management library for Node.js. It is designed to be easy to use, flexible, it can used in any environment, and can be used with any standard-schema-compliant validation library, like zod or valibot.
Think of it as a modern version of convict but simpler, cross env and using battle tested validation libraries.
- Environment variables support
- Validation with any standard-schema-compliant validation library
- Flat object support
- Multiple sources of configuration
- Type-safe configuration
- Composable configuration
- No runtime dependencies
Install package:
# pnpm
pnpm install figue
# npm
npm install figue
# yarn
yarn install figue
Import:
import { defineConfig } from 'figue';Use the defineConfig function to define your configuration, here with valibot:
import { defineConfig } from 'figue';
import * as v from 'valibot';
const { config } = defineConfig(
{
env: {
doc: 'Application current environment',
default: 'development',
schema: v.picklist(['development', 'production', 'test']),
env: 'NODE_ENV',
},
port: {
doc: 'Application port to listen',
schema: v.pipe(v.union([v.number(), v.string()]), v.transform(Number)),
default: 3000,
env: 'PORT',
},
db: {
host: {
doc: 'Database server url',
schema: v.pipe(v.string(), v.url()),
default: 'http://localhost:5432',
env: 'APP_DB_HOST',
},
username: {
doc: 'Database server username',
schema: v.string(),
default: 'pg',
env: 'APP_DB_USERNAME',
},
password: {
doc: 'Database server password',
schema: v.string(),
default: '',
env: 'APP_DB_PASSWORD',
},
},
},
{
envSource: process.env,
},
);
console.log(config);
// {
// env: "development",
// port: 3000,
// db: {
// url: "https://localhost",
// username: "pg",
// password: "",
// },
// }You can see more examples in the demo folder.
- Figue with zod: demo/figue-zod.ts
- Figue with valibot: demo/figue-valibot.ts
Use the envSource key of the second argument of defineConfig to specify the source of the environment variables:
const { config } = defineConfig(
{
/* ... */
},
{
envSource: process.env,
},
);In some case you don't have access to a process.env variable, like with vite, just simply load what stores your env variables :
const { config } = defineConfig(
{
/* ... */
},
{
envSource: import.meta.env,
},
);You can even specify you custom environment storage as long as it's a simple flat object map, for example:
const { config } = defineConfig(
{
env: {
doc: 'Application current environment',
default: 'development',
schema: z.enum(['development', 'production', 'test']),
env: 'NODE_ENV',
},
/* ... */
},
{
envSource: {
NODE_ENV: 'development',
PORT: '3000',
APP_DB_HOST: 'localhost',
APP_DB_USERNAME: 'pg',
APP_DB_PASSWORD: '',
},
},
);If, for some reason, you have multiple sources of environment variables, you can use the envSources key of the second argument of defineConfig to specify an array of sources:
const { config } = defineConfig(
{
/* ... */
},
{
envSource: [import.meta.env, myEnvs],
},
);You can specify multiple environment variable names for a single configuration field by providing an array of strings. Figue will use the first environment variable that is found in your environment sources.
This feature is particularly useful when:
- Supporting multiple deployment environments with different environment variable naming conventions
- Migrating from legacy environment variable names while maintaining backward compatibility
- Providing fallback options for missing environment variables
const { config } = defineConfig(
{
port: {
doc: 'Application port to listen',
schema: z.coerce.number(),
default: 3000,
// Will use PORT if available, otherwise APP_PORT, otherwise SERVER_PORT
env: ['PORT', 'APP_PORT', 'SERVER_PORT'],
},
database: {
host: {
doc: 'Database host',
schema: z.string(),
default: 'localhost',
// Useful for supporting legacy environment variable names
env: ['DATABASE_HOST', 'DB_HOST', 'LEGACY_DB_URL'],
},
},
workerId: {
doc: 'Worker identifier',
schema: z.string().optional(),
// Or using some plateform specific environment variable
env: ['WORKER_ID', 'HEROKU_DYNO_ID', 'RENDER_INSTANCE_ID'],
},
},
{
envSource: process.env,
},
);Some caveats:
- If none of the specified environment variables are found, Figue will fall back to the default as expected when no
envkey is present. - If a variable is found but its value is nullish or falsy (like an empty string, or undefined), Figue will still consider it as set and use that value. If you want to ignore such values, you should handle that in your schema validation.
- Ensure that the order of environment variables in the array reflects their priority, as Figue will use the first one it finds.
You can use the getDefaults key of the second argument of defineConfig to specify a function that will be called to get some defaults:
const { config } = defineConfig(
{
env: {
doc: 'Application current environment',
default: 'development',
schema: z.enum(['development', 'production', 'test']),
env: 'NODE_ENV',
},
port: {
doc: 'Application port to listen',
schema: z.coerce.number().int().positive(),
default: 3000,
env: 'PORT',
},
},
{
envSource: {
PORT: 3001,
},
// The config argument is build from the config definition defaults and the envSources
// Typically you will use it to override some defaults based the config
getDefaults: ({ config }) => ({
port: config.env === 'test' ? 4444 : config.port,
}),
},
);You can also use the defaults property of the second argument of defineConfig to specify some static defaults (for example taken from a json file):
const { config } = defineConfig(
{
/* ... */
},
{
// Either an array of config partial...
defaults: [
{
port: 4444,
},
],
// ... or a single config partial
defaults: {
port: 4444,
},
},
);Convict is meant to be used in node based environnement, it needs to have access to global variables that may may not be present in some environnement (like process, global), and it also imports fs.
Figue is the french for fig -> con-fig.
- Clone this repository
- Install dependencies using
pnpm install - Run interactive tests using
pnpm dev
This project is crafted with ❤️ by Corentin Thomasset. If you find this project helpful, please consider supporting my work.
Fig icons created by Freepik - Flaticon
This project is under the MIT license.