Metafig is a library for loading configuration values from external services like AWS Parameter Store, Environment Variables, S3 objects, etc. All of this is configurable from a single place, making it easy to synthesize different values from different locations.

We had a bunch of lambda functions with a bunch of different configuration methods. Some used S3 before lambda had encryption helpers for environment variables. Some used environment variables. Some used the Parameter Store.

While we migrated most of them to the Parameter store, we wanted to:

1. Make it easier to mix-and-match
   • We had a few things too big for the Parameter Store
   • There are a few things that make more sense as environment variables.
2. Make it easier to migrate in the future if we change our mind.

npm install --save @cimpresscloud/metafig

./config.json

{
  "plugins": {
    "database": {
      "awsParam": {
        "path": "/production/myapp/database",
        "decryption": true
      }
    },
    "api": {
      "awsParam": {
        "path": "/production/myapp/api",
        "decryption": true
      },
      "environment": {
        "baseUrl": "API_BASE_URL"
      }  
    }
  }
}

./index.js

const metafig = require('metafig');
metafig(require('./config.json'), 'plugins')
  .then(config => {
    var db = new Db(config.plugins.database.connectionString);
    return db.query('SELECT * FROM MyTable');
  });

In this case, the config object looks something like:

{
  plugins: {
    database: {
      connectionString: "sql://mydatabase.example.com:5678/db",
      username: "myApplication",
      password: "secret"
    },
    api: {
      token: "123456",
      baseUrl: "https://myapi.example.com"
    }
  }
}

As you can see, it pulled some items from environment variables, and some from the AWS SSM Parameter Store.

Metafig comes with 3 built-in configuration providers:

The awsParam provider can pull down items from the AWS SSM Parameter Store. It can take 2 parameters:

"awsParam": {
  "path": "/app", // the path to the hierarchy in SSM to pull down
  "decryption": true // whether or not to decrypt SecureStrings
}

It will pull down every parameter listed under the specified path and put them into the configuration object, nesting objects if there are deeper configurations.

The awsS3 provider can pull down JSON configuration snippets as objects from S3. It can accept any parameter for the getObject API action:

"awsS3": {
  "Bucket": "metafig-integration-test",
  "Key": "test.json"
}

It will pull down the specified object and run it through the JSON parser.

The environment provider pulls values out of environment variables and maps them to configuration items.

"environment": {
  "myValue": "USERNAME"
}

Take the USERNAME environment variable and puts it into the myValue property of your configuration.

The literal provider just uses the literal value passed to it:

"literal": {
  "username": "nmaclennan"
}