dotenv-flow

dotenv is a zero-dependency npm module that loads environment variables from a .env file into process.env.

dotenv-flow extends dotenv, adding support of NODE_ENV-specific .env* files like .env.development, .env.test, .env.stage, and .env.production, and the appropriate .env*.local overrides.

It allows your app to have multiple environments (like "development", "test", "stage", and "production" respectively) with selectively-adjusted environment variable setups and load them dynamically depending on the current NODE_ENV.

In addition to that, .env*.local overrides add the ability to overwrite variables locally for development, testing, and debugging purposes (note that the appropriate .env*.local entry should be added to your .gitignore).

🌱 Inspired by Ruby's dotenv (a.k.a. dotenv-rails) gem, CreateReactApp's storing configs in .env* files approach, the Twelve-Factor App methodology in general, and its store config in the environment section in particular.

Installation

Using NPM:

1$ npm install dotenv-flow --save

Using Yarn:

1$ yarn add dotenv-flow

Using PNPM:

1$ pnpm add dotenv-flow

Usage

As early as possible in your Node.js app, initialize dotenv-flow:

1require('dotenv-flow').config();

It will allow you to configure and use dotenv-flow from your code programmatically.

If you're using TypeScript or ES Modules:

1import dotenvFlow from 'dotenv-flow';
2dotenvFlow.config();

Alternatively, you can use the default config entry point that allows you to configure dotenv-flow using command switch flags or predefined environment variables:

1require('dotenv-flow/config');

Or even make dotenv-flow load environment variables for your app without adding it to the code using preload technique:

1$ node -r "dotenv-flow/config" your_app.js

It works with ts-node as well:

1$ ts-node -r "dotenv-flow/config" your_app.ts

How it works

Once dotenv-flow is initialized (using .config or any other method above), environment variables defined in your .env* files are loaded and become accessible in your Node.js app via process.env.*.

For example, let's suppose that you have the following .env* files in your project:

1# .env
2
3DATABASE_HOST=127.0.0.1
4DATABASE_PORT=27017
5DATABASE_USER=default
6DATABASE_PASS=
7DATABASE_NAME=my_app

1# .env.local
2
3DATABASE_USER=local-user
4DATABASE_PASS=super-secret

1# .env.development
2
3DATABASE_NAME=my_app_dev

1# .env.test
2
3DATABASE_NAME=my_app_test

1# .env.production
2
3DATABASE_NAME=my_app_prod

1# .env.production.local
2
3DATABASE_HOST=10.0.0.32
4DATABASE_PORT=27017
5DATABASE_USER=devops
6DATABASE_PASS=1qa2ws3ed4rf5tg6yh
7DATABASE_NAME=application_storage

1// your_script.js
2
3require('dotenv-flow').config();
4
5console.log('database host:', process.env.DATABASE_HOST);
6console.log('database port:', process.env.DATABASE_PORT);
7console.log('database user:', process.env.DATABASE_USER);
8console.log('database pass:', process.env.DATABASE_PASS);
9console.log('database name:', process.env.DATABASE_NAME);

And if you run your_script.js in the development environment, like:

1$ NODE_ENV=development node your_script.js

you'll get the following output:

1database host: 127.0.0.1
2database port: 27017
3database user: local-user
4database pass: super-secret
5database name: my_app_dev

Or if you run the same script in the production environment:

1$ NODE_ENV=production node your_script.js

you'll get the following:

1database host: 10.0.0.32
2database port: 27017
3database user: devops
4database pass: 1qa2ws3ed4rf5tg6yh
5database name: application_storage

Note that the .env*.local files should be ignored by your version control system (refer the Files under version control section below to learn more), and you should have the .env.production.local file only on your production deployment machine.

NODE_ENV-specific env files

Actually, dotenv-flow doesn't have any predefined environment names, so you may use whatever names you want. However, it's a good practice to use the world's universally recognized environment names like development, test, production, as well as frequently used qa or stage.

The naming convention for NODE_ENV-specific files is simply as .env.${NODE_ENV}[.local] (i.e. .env.development, .env.test, .env.production, .env.development.local, .env.production.local, etc.).

To activate specific environment run your application with predefined NODE_ENV environment variable, like:

1$ export NODE_ENV=production
2$ node your_script.js

or:

1$ NODE_ENV=production node your_script.js

If you are on Windows:

1> SET NODE_ENV=production
2> node your_script.js

Or even better, use cross-env to make it work independent of platform:

1$ cross-env NODE_ENV=production node your_script.js

The --node-env switch is also supported:

1$ node your_script.js --node-env=production

Preload

Alternatively, you can preload dotenv-flow using node's -r (--require) command line option.

1$ NODE_ENV=production node -r dotenv-flow/config your_script.js

or:

1$ node -r dotenv-flow/config your_script.js --node-env=production

You can also use environment variables to set configuration options when preloading the dotenv-flow/config:

1$ DOTENV_FLOW_PATH=/path/to/env-files-dir node -r dotenv-flow/config your_script.js

Refer to the dotenv-flow/config options section below to see all available options.

Files under version control

The main point here is not to commit production database passwords, API keys and other sensitive things to your source code repository, but it's still nice to have default database connections, ports, hosts, etc., for development and testing purposes to keep your code clean and simple.

Understanding the above, we have the following approach:

You can keep all the fallback values in the default .env file, that (if exists) will always be loaded by default. Also, it is a good place to have all the application used environment variables there, thus having a reference of environment variables that are used by your application on the whole. So it is a good reason to share the .env file with other developers in your team, but keep all the sensitive data on your own (or production) machine locally in the .env*.local files.

It is not necessary, but also a good practice to use NODE_ENV to control the environment to run your application in. And if you follow this practice you can keep the NODE_ENV-specific defaults in your .env.development, .env.test, .env.production files sharing them with your team as well. Any NODE_ENV-specific .env.* file's values can also be overwritten in the appropriate .env.*.local file (i.e. .env.development.local, .env.test.local, .env.production.local).

Summarizing the above, you can have the following .env* files in your project:

• .env – for default (fallback) values, tracked by VCS
• .env.development – for development environment, tracked by VCS
• .env.test – for test environment, tracked by VCS
• .env.production – for production environment, tracked by VCS
• .env.local – for individual default values, ignored by VCS
• .env.development.local – for individual development environment values, ignored by VCS
• .env.test.local – for individual test environment values, ignored by VCS
• .env.production.local – for production environment values (DB passwords, API keys, etc.), ignored by VCS

Note that .env.* file names may vary in your project depending on your own needs/preferences, just keep in mind that .env*.local files should be untracked (ignored) by your version control system.

Here is an example of the .gitignore (or .hgignore) file entry to keep it clear:

1# local .env* files
2.env.local
3.env.*.local

Variables overwriting/priority

Since multiple .env* files are loaded simultaneously, all the variables defined in these files are merged in the following order:

1. The .env file has the lowest priority. Keep the most default (fallback) values there;
2. The .env.local file has a priority over the .env (except when NODE_ENV=test, in which case this file is not loaded). Create it if you want to overwrite the default values for your own environment-specific needs;
3. NODE_ENV-specific env files (like .env.development, .env.test, etc.) have a priority over the default .env and .env.local files. Keep NODE_ENV-specific environment variables there;
4. NODE_ENV-specific local env files (.env.development.local, .env.production.local, etc.) have the highest priority over all the env files. As with .env.local, create them only if you need to overwrite NODE_ENV-specific values for your own environment-specific needs;
5. Environment variables that are already set will not be overwritten, that means that the command line variables have a higher priority over all those defined in env files;

Alternative defaults: .env.defaults

In addition to .env, you may also use .env.defaults to store default (fallback) values.

This may come handy e.g. when migrating from dotenv (where it is strongly advised against committing .env file to VCS) and you already have .env file used to store your local values.

In such case, you may prefer to keep using your existing .env (ignored by VCS) as your local config and create additional .env.defaults (tracked by VCS) file which will be loaded before .env.

Then at every place .env is mentioned in the docs, read it as: ".env.defaults first, then .env".

dotenv-flow/config options

The following configuration options can be used when:

• a) preloading dotenv-flow using Node's -r ([ts-]node --require) switch, or…
• b) requireing the dotenv-flow/config entry point (using require('dotenv-flow/config');).

Environment variables

• NODE_ENV => options.node_env;
• DEFAULT_NODE_ENV => options.default_node_env;
• DOTENV_FLOW_PATH => options.path;
• DOTENV_FLOW_PATTERN => options.pattern;
• DOTENV_FLOW_ENCODING => options.encoding;
• DOTENV_FLOW_PURGE_DOTENV => options.purge_dotenv;
• DOTENV_FLOW_DEBUG => options.debug;
• DOTENV_FLOW_SILENT => options.silent;

for example:

1$ NODE_ENV=production DOTENV_FLOW_PATH=/path/to/env-files-dir node -r dotenv-flow/config your_script.js

Command line switches

• --node-env => options.node_env;
• --default-node-env => options.default_node_env;
• --dotenv-flow-path => options.path;
• --dotenv-flow-pattern => options.pattern;
• --dotenv-flow-encoding => options.encoding;
• --dotenv-flow-purge-dotenv => options.purge_dotenv;
• --dotenv-flow-debug => options.debug;
• --dotenv-flow-silent => options.silent;

Make sure that dotenv-flow/config-specific CLI switches are separated from Node's by -- (double dash) since they are not recognized by Node.js.

for example:

1$ node --require dotenv-flow/config your_script.js -- --dotenv-flow-path=/path/to/project --dotenv-flow-encoding=base64

API reference

.config([options]) => object

"dotenv-flow" initialization function (API entry point).

Allows configuring dotenv-flow programmatically.

Also, like the original module (dotenv), it returns an object with .parsed property containing the resulting varname => values pairs or .error property if the initialization is failed.

options.node_env

Type: string

Default: process.env.NODE_ENV

With the node_env option you can force the module to use your custom environment value independent of process.env.NODE_ENV:

1require('dotenv-flow').config({
2  node_env: process.argv[2] || 'development'
3});

options.default_node_env

Type: string

Default: undefined

If the NODE_ENV environment variable is not set, the module doesn't load/parse any NODE_ENV-specific files at all. Therefore, you may want to use "development" as a default environment, like:

1require('dotenv-flow').config({
2  default_node_env: 'development'
3});

To be clear, just make a note that all the following initialization examples are also equivalent:

1process.env.NODE_ENV = process.env.NODE_ENV || 'development';
2
3require('dotenv-flow').config();

1require('dotenv-flow').config({
2  node_env: process.env.NODE_ENV || 'development'
3});

1require('dotenv-flow').config({
2  node_env: process.env.NODE_ENV,
3  default_node_env: 'development'
4});

All the examples above, considers the value of process.env.NODE_ENV at first, and if it is not set, uses "development" as the value by default. You can just choose one that looks prettier for you.

options.path

Type: string

Default: process.cwd() (current working directory)

With the path initialization option you can specify a path to .env* files directory:

1require('dotenv-flow').config({
2  path: '/path/to/env-files-dir'
3});

If the option is not provided, the current working directory is used.

options.pattern

Type: string

Default: ".env[.node_env][.local]"

Allows you to change the default .env* files' naming convention if you want to have a specific file naming structure for maintaining your environment variables' files.

Default Value

The default value ".env[.node_env][.local]" makes dotenv-flow look up and load the following files in order:

1. .env
2. .env.local
3. .env.${NODE_ENV}
4. .env.${NODE_ENV}.local

For example, when the proess.env.NODE_ENV (or options.node_env) is set to "development", dotenv-flow will be looking for and parsing (if found) the following files:

1. .env
2. .env.local
3. .env.development
4. .env.development.local

Custom Pattern

Here is a couple of examples of customizing the .env* files naming convention:

For example, if you set the pattern to ".env/[local/]env[.node_env]", dotenv-flow will look for these files instead:

1. .env/env
2. .env/local/env
3. .env/env.development
4. .env/local/env.development

… or if you set the pattern to ".env/[.node_env/].env[.node_env][.local]", dotenv-flow will try to find and parse:

1. .env/.env
2. .env/.env.local
3. .env/development/.env.development
4. .env/development/.env.development.local

› Please refer to .listFiles([options]) to dive deeper.

options.files

Type: string[]

Allows explicitly specifying a list (and the order) of .env* files to load.

Note that options like node_env, default_node_env, and pattern are ignored in this case.

1require('dotenv-flow').config({
2  files: [
3    '.env',
4    '.env.local',
5    `.env.${process.env.NODE_ENV}`, // '.env.development'
6    `.env.${process.env.NODE_ENV}.local` // '.env.development.local'
7  ]
8});

options.encoding

Type: string

Default: "utf8"

You can specify the encoding for reading your files containing environment variables.

1require('dotenv-flow').config({
2  encoding: 'base64'
3});

options.purge_dotenv

Type: boolean

Default: false

In some cases the original "dotenv" library can be used by one of the dependent npm modules. It causes calling the original dotenv.config() that loads the .env file from your project before you can call dotenv-flow.config().

Such cases breaks .env* files priority because the previously loaded environment variables are treated as shell-defined thus having a higher priority.

Setting the purge_dotenv option to true can gracefully fix this issue.

1require('dotenv-flow').config({
2  purge_dotenv: true
3});

options.debug

Type: boolean

Default: false

Enables detailed logging to debug why certain variables are not being set as you expect.

options.silent

Type: boolean

Default: false

Suppresses all kinds of warnings including ".env*" files' loading errors.

1require('dotenv-flow').config({
2  silent: true
3});

---

The following API is considered as internal, although it's exposed for programmatic use of dotenv-flow for your own project-specific needs.

.listFiles([options]) => string[]

Returns a list of existing .env* filenames depending on the given options.

The resulting list is ordered by the env files' variables overwriting priority from lowest to highest.

This can also be referenced as "env files' environment cascade" or "order of ascending priority."

⚠️ Note that the .env.local file is not listed for "test" environment, since normally you expect tests to produce the same results for everyone.

Parameters:

options.node_env

Type: string

Default: undefined

The node environment (a.k.a. process.env.NODE_ENV).

The conventionally used values are "development", "test" (or "staging") and "production", also commonly used are "qa", "uat", "ci".

options.pattern

Type: string

Default: ".env[.node_env][.local]"

.env* files' naming convention pattern.

The default one, (".env[.node_env][.local]") without options.node_env given, produces the following list of filenames:

• .env
• .env.local

When options.node_env is set, for example to "development", it appends "node_env-specific" filenames, that will make .listFiles to return:

• .env
• .env.local
• .env.development
• .env.development.local

Another example might be the pattern ".env/[local/]env[.node_env]", that without options.node_env will produce:

• .env/env
• .env/local/env

… and if options.node_env is set to (for example) "development", will append "node_env-specific" files producing the following:

• .env/env
• .env/local/env
• .env/env.development
• .env/local/development

Also, note that if [node_env] placeholders is missing in the pattern, none of the "node_env-specific" files fill be listed. For example, a pattern like ".env[.local]", independently of whether the options.node_env is set, will always produce:

• .env
• .env.local

… except the case when options.node_env is set to "test", which (as mentioned above) will exclude .env.local producing just a single:

• .env

… since normally we expect tests to produce the same results for everyone.

options.debug

Type: boolean

Default: false

Enables debug messages.

Returns:

Type: string[]

A list of .env* filenames.

Example:

1const dotenvFlow = require('dotenv-flow');
2
3const filenames = dotenvFlow.listFiles({ node_env: 'development' });
4
5console.log(filenames); // will output the following:
6// > [ '/path/to/project/.env.defaults',
7// >   '/path/to/project/.env',
8// >   '/path/to/project/.env.local',
9// >   '/path/to/project/.env.development',
10// >   '/path/to/project/.env.development.local' ]

.parse(filenames, [options]) => object

Parses the content of a given file(s) to use the result programmatically. Accepts a filename or a list of filenames and returns a map of the parsed key/values as an object.

When several filenames are given, the parsed variables are merged into a single object using the "overwrite" strategy.

Parameters:

filenames

Type: string|string[]

A filename or a list of filenames to parse.

options.encoding

Type: string

Default: "utf8"

An optional encoding for reading files.

options.debug

Type: boolean

Default: false

Enables debug messages.

Returns:

Type: object

The resulting map of { env_var: value } as an object.

Example:

1# .env
2
3FOO=bar
4BAZ=bar

1# .env.local
2
3BAZ=qux

1const dotenvFlow = require('dotenv-flow');
2
3const variables = dotenvFlow.parse([
4  '/path/to/project/.env',
5  '/path/to/project/.env.local'
6]);
7
8console.log(typeof variables, variables); // > object { FOO: 'bar', BAZ: 'qux' }

.load(filenames, [options]) => object

Loads variables defined in a given file(s) into process.env.

When several filenames are given, parsed environment variables are merged using the "overwrite" strategy since it utilizes .parse() for doing this. But eventually, assigning the parsed environment variables to process.env is done using the "append" strategy, thus giving a higher priority to the environment variables predefined by the shell.

Parameters:

filenames

Type: string|string[]

A filename or a list of filenames to load.

options.encoding

Type: string

Default: "utf8"

An optional encoding for reading files.

options.debug

Type: boolean

Default: false

Optionally, turn on debug messages.

options.silent

Type: boolean

Default: false

If enabled, suppresses all kinds of warnings including ".env*" files' loading errors.

Returns:

Type: object

The same as .config(), the returning object contains .parsed property with a parsed content of the given file(s), or if parsing is failed the .error property with a reference to the reasoning error.

Example:

1# .env
2
3FOO=bar
4BAZ=bar

1# .env.local
2
3BAZ=qux

1const dotenvFlow = require('dotenv-flow');
2
3process.env.BAZ = 'Yay!';
4
5const result = dotenvFlow.load([
6  '/path/to/project/.env',
7  '/path/to/project/.env.local'
8]);
9
10console.log(typeof result, result); // > object { parsed: { FOO: 'bar', BAZ: 'qux' } }
11
12console.log(process.env.FOO); // > 'bar'
13console.log(process.env.BAZ); // > 'Yay!'

.unload(filenames, [options]) => void

Unloads variables defined in a given file(s) from process.env.

The environment variables that are predefined (i.e. by the shell) will not be unloaded.

Parameters:

filenames

Type: string|string[]

A filename or a list of filenames to unload.

options.encoding

Type: string

Default: "utf8"

An optional encoding for reading files.

Example:

1# .env
2
3FOO=bar
4BAZ=bar

1# .env.local
2
3BAZ=qux

1const dotenvFlow = require('dotenv-flow');
2
3process.env.BAZ = 'Yay!';
4
5dotenvFlow.load([  '/path/to/project/.env',  '/path/to/project/.env.local']);
6
7console.log(process.env.FOO); // > 'bar'
8console.log(process.env.BAZ); // > 'Yay!'
9
10dotenvFlow.unload([  '/path/to/project/.env',  '/path/to/project/.env.local']);
11
12console.log(process.env.FOO); // > undefined
13console.log(process.env.BAZ); // > 'Yay!'

---

Related packages

• @types/dotenv-flow – type definitions for using dotenv-flow with TypeScript
• dotenv-flow-webpack – a webpack plugin for using dotenv-flow in web applications
• dotenv-flow-cli – CLI executable that preloads environment variables using dotenv-flow
• dotenv-expand – environment variables expansion (originally designed for dotenv, but also compatible with dotenv-flow)

Contributing

Feel free to dive in! Open an issue or submit PRs.

Running tests

Using NPM:

1$ npm test

Using Yarn:

1$ yarn test

License

Licensed under MIT © 2018-2023 Dan Kerimdzhanov