Mongo-type-gen NPM

mongo-type-gen is a lib for unifing your types. Define once, reuse everywhere!

Simple: Short mtg command allows for easy generating, uploading, and downloading of types.
Performant: Lightweight with very few deps and a small bundle size.

Takes a collection definition, ie **.collection.js;

module.exports {
  indexes: [
    {
      key: { name: 1 },
      unique: true,
    },
  ],
  validator: {
    $jsonSchema: {
      bsonType: 'object',
      properties: {
        gpa: {
          bsonType: ['double'],
          description: "'gpa' must be a double if field exists",
        },
        name: {
          bsonType: 'string',
          description: "'name' must be a string and is required",
        },
        year: {
          bsonType: 'int',
          description: "'year' must be an integer in [  2017,  3017  ] and is required",
          maximum: 3017,
          minimum: 2017,
        },
      },
      required: ['major', 'name', 'year'],
      title: 'Student Object Validation',
    },
  },
};

And generates a mongo.types.ts and mongo.sdls.ts typing files for your Typescript types and GraphQL SDLs;

/* This file is generated by mongo-type-gen.  Do not edit */

import { ObjectId } from 'mongodb';

export enum CollectionEnum {
  students = 'students',
}

export type StudentDoc = {
  gpa?: number;
  name: string;
  year: number;
};

/* This file is generated by mongo-type-gen.  Do not edit */

import { gql } from 'graphql-tag';

export default gql`
  enum CollectionEnum {
    students
  }

  type StudentDoc {
    gpa: Float
    name: String!
    year: Int!
  }
`;

Motivation

Working with typescript can involve some busy work. Creating all the different types associated with your data by hand is tedious and time consuming. From GraphQL SDLs, to Typescript types, to Mongo JSON schemas or Mongoose Schemas. This takes time and introduces tech debt if data is added or modified.

Let's dry this up and only define our types once. Let mongo-type-gen handle the busy work of producing the different types needed for your GraphQL app writen in Typescript.

First, let's identify the best source for types. What syntax allows us to be the most expressive 🤔. This is where Mongo's `$jsonSchema validators come into play. They allow for expression of meta data that cannot be expressed in Typescript, GraphQL, or Mongoose alone. This lib uses validators as the source.

Whether you have your validators defined in your project or they live in MongoDB, we'll grab em' and generate your types and SDLs.

Usage

To get started, add the lib to your project. You'll only need it for local dev so install it as a dev dep.

npm i -D mongo-type-gen

Then create some ease-of-use scripts in your package.json.

{
  "scripts": {
    "start": "mtg -w & ts-node-dev ./src/index.ts"
  }
}

With mongo-type-gen, you'll need Mongo JSONSchema flavor validators. You can either define these in the project and ref them in a mtg.config.ts config file or have mtg download them from mongo.

To see this in action, create a mtg.config.js config file.

import configMtg from 'mongo-type-gen';

export default configMtg({
  db: 'mongo-type-gen', // name of your db, required
  input: '**/*.collection.*s', // glob pattern of your collection files, if not present mtg will fetch your collection from Mongo
  output: {
    collections: 'examples/collections', // output directory to place your collection files if mtg fetches them from Mongo
    sdls: 'examples/sdls', // output dir to place your typescript types
    types: 'examples/types', // output dir to place your SDLs
  },
  uri: 'mongodb://localhost:27017',
});

This let's mtg learn where it should download or upload your collections and where it should place your types.

There is also a watcher available by passing -w so any changes you make to your collection's is reflected in your types and Mongo.

When downloading your collection's from Mongo, mtg will add an isGenerated boolean field to the data. This lets mtg know how to manage your collection data. If the bool is true, mtg will not upload your data to Mongo as it is expecting Mongo to be the source of truth. This is useful for apps that need Mongo's types but are not responsibile to setting validators and indexes. You can switch the role by simiply removing the isGenerated flag from the data.