mongore

v1.0.1

Published

3 years ago

Creating native JavaScript classes with MongoDB backend.

Downloads

0High
0Medium
0Low

ronenness

mongodb mongo orm models schema

Mongore

Mongore provides an easy way to create native JavaScript classes with MongoDB backend. The idea is to make Mongo pretty much invisible and still enjoy Mongo-powered persistency and fast search.

Fair warning: This project is still undertested, use at your own risk.

Lets take a look at a quick example. First, lets connect to MongoDB (for simplicity, I will ignore callbacks and validations for now):

const MongoClient = require('mongodb').MongoClient
const Mongore = require('mongore')

Mongore.Client.connect(MongoClient, "mongodb://localhost:27017/dogsDb", "dogsDb");

Now lets define a model:

// define a class to represent a dog 
class Dog extends Mongore.Model
{
    // create the dog with name and breed.
    constructor(name, breed) 
    {
        super();
        this.name = name;
        this.breed = breed;
    }

    // bark.
    bark() 
    {
        if (this.breed == "chihuahua") {
            console.log("*angry rat noises*");
        }
        else {
            console.log("Woof!");
        }
    }

    // give dog a bone.
    // note: cooked bones are actually dangerous for dogs, you should never give your dog cooked bones!
    giveBone()
    {
        this.bonesEaten++;
        this.bark();
    }

    // calculate dog's age based on when it was created in DB.
    get age()
    {
        return (new Date()).getYear() - this.mongore.creationTime.getYear());
    }
}

// define the dog fields that will be stored in DB:
Dog.buildModel(
    fields: {
        _id: new Mongore.Fields.StringField({source: "name", maxLength: 16}), 
        breed: new Mongore.Fields.ChoiceField({choices: ["chihuahua", "bulldog", "golden", "lab"]}),
        bonesEaten: new Mongore.Fields.NumericField({default: 0, parser: parseInt})
    }
);

Using the model is super easy!

var myDog = new Dog("Rex", "golden");
myDog.giveBone();
myDog.mongore.save();

Or to load a dog from db..

Dog.mongore.load("Rex", (myDog) => {
    console.log("My dog ate ", myDog.bonesEaten, "bones.");
});

Key Features

Mongore's key features are:

Using DB models like plain JavaScript classes.
Built-in field types with validation and automatic 'cleanup' (also convert to MongoDB validators).
Useful added metadata like creation time, last update time, versioning, ect.
Optimized - will only save changed fields, if anything was changed.
Foreign keys that turn into model instances when loaded.
Automatic collections setup.

Using Mongore

Install

To install via npm:

npm install mongore

Initialize

Before using Mongore, you need to initialize a connection (note: its OK to define your models before initializing, you only need to connect before loading / saving objects):

const Mongore = require('./mongore');
const MongoClient = require('mongodb').MongoClient;

Mongore.Client.connect(MongoClient, Config.db.url, Config.db.name, () => {
    // ready for action!
})

As you can see Mongore don't require MongoDB directly. This means you can use whatever version you want or wrap Mongo's client yourself, as long as the basic API of insertOne, updateOne, findOne is kept.

A Small Example..

Now lets define some Models, ie. persistent JS objects with MongoDB backend. For this example, we'll make a simple forum app.

Step 1: Users Model

We'll start with the User model:

/**
 * Represent a user.
*/
class User extends Mongore.Model
{
    /**
     * Create a new user.
    */
    constructor(username, password, userType) 
    {
        super();
        this.username = username;
        this.password = password;
        this.userType = userType;
    }

    /**
     * Get if admin user.
    */
    get isAdmin()
    {
        return this.userType === "admin";
    }

    /**
     * Create and return a new board (don't save it).
    */
    createBoard(name)
    {
        if (!this.isAdmin) {
            throw new Error("Only admins can open boards!");
        }
        return new Board(name, this);
    }
}
User.buildModel(
    fields: {
        _id: new Mongore.Fields.StringField({source: "username", maxLength: 16}),
        userType: new Mongore.Fields.ChoiceField({choices: ["guest", "registered", "admin"]}),
        password: new Mongore.Fields.StringField()
    }
);

So what's going on above?

First we define a User class that get username, password, and type. It also have a method to create a board (will be defined next) and a getter to check if type is admin.
Next, we define the persistent part of the model, which is the fields that will represent this object in MongoDB.
1. First we define the _id field. We create it as a string with max length of 16 characters, and we make it a proxy of the username property.
2. Next we add a user type choice field, which is a string that will only accept one of the listed values: 'guest', 'registered', or 'admin'.
3. Finally, we add a password field as plain text (horrible when considering security, I know).

Once the fields are defined, you can use them just as you would with any JS property from the instance, as demonstrated in the constructor and in the class's methods.

Create The Collection

As you probably know, MongoDB works with collections, and while you can let MongoDB create collections with default values automatically, its best to define them with validations and other metadata (like indexes). So lets ask Mongore to create a collection for Users:

User.mongore.createCollection(() => {
    console.log("Users collection is now ready!");
});

Obviously you don't have to call createCollection every time your app starts (although there's no harm in that), so its best to create short init-db.js script, and put all the createCollection() calls there. Call this script whenever you change or define a new Model.

Create A User Instance

Now lets create and save a new user:

var user = new User("Billy", "pass1234", "admin");
user.mongore.save(() => {
    console.log("Great success!");
},
(err) => {
    console.log("Error! ", err);
});

That's it! You can now check your MongoDB and see a Users collection with a single object, Billy.

Load A User

So how do we load the user we just created? Easy!

User.mongore.load("Billy", (user) => {
    console.log(`Welcome back, ${user.username}!`);
});

A Word About `mongore` Property

All the fields you define in your model are created as-if they were plain properties under your class. However, as you noticed above, we also have the mongore object under the instance and the class itself.

This object contains the API and metadata of mongore, and represent the bridge between your plain JS object and the Database.

.mongore contains the API related to the Model itself, like the load() and createCollection() methods.
.mongore contains the API related to a specific instance, like the save() method and other useful functions described later.

Step 2: Board Model

Continuing with our example, time to define the Board model to represent a board:

/**
 * Represent a board.
*/
class Board extends Mongore.Model
{
    /**
     * Create a new board.
    */
    constructor(name, creator) 
    {
        super();
        this.name = name;
        this.creator = creator;
    }

    /**
     * Create and return a new message (don't save it).
    */
    createMessage(user, text)
    {
        return new Message(this, user, text);
    }
}
Board.buildModel(
    fields: {
        _id: new Mongore.Fields.StringField({source: "name", maxLength: 16}),
        creator: new Mongore.Fields.ForeignKeyField({model: User, canBeNull: false}),
    }
);

The name field is similar to User.username and will not be discussed here. The new field, creator is interesting.

A ForeignKeyField is a field that expect an instance of the provided Model (in this case, User). In Mongo, it will be stores as the instance's _id field. When an instance of Board is loaded, it will also load all its foreign keys.

Note that the success callback will be called before foreign keys are loaded. Lets take a look at how we load a Board instance, wait for the User foreign key to be ready, then performing an action:

Board.mongore.load("myBoard", (board) => {
    board.creator.mongore.onLoaded((user) => {
        board.createMessage(user, "Hi all, just logged in! - XOXO").mongore.save();
    })
})

Step 3: Message Model

And finally, lets create the message Model:

/**
 * Represent a single text message in one of the forum's boards.
*/
class Message extends Mongore.Model
{
    /**
     * Create a new message.
    */
    constructor(board, author, msg) 
    {
        super();
        this.board = board;
        this.author = author;
        this.msg = msg;
    }

    /**
     * Give this message a thumb up.
    */
    upvote()
    {
        this.score++;
    }

    /**
     * Give this message a thumb down.
    */
    downvote()
    {
        this.score--;
    }
}
Message.buildModel(
    fields: {
        board: new Mongore.Fields.ForeignKeyField({model: Board, canBeNull: false}),
        author: new Mongore.Fields.ForeignKeyField({model: User, canBeNull: false}),
        msg: new Mongore.Fields.StringField(),
        score: new Mongore.Fields.NumericField({parser: parseInt, index: true})
    }
);

The above is pretty similar to the previous models, with a just a minor thing to notice: index: true will set the score field as an index, which means we can quickly search on it.

That's it for the example, now lets explore the API!

Mongore.Model.buildModel()

The main function that turn your JS class into a proper Model is the buildModel() method, as demonstrated in the example above. This method accept the following arguments (as a dictionary):

fields: A dictionary of fields to attach to this Model. Every field will receive a corresponding getter and setter, and will be stored in MongoDB. Field types are described later.
maxSizeInBytes: If defined, will limit this Model's collection to this number of bytes.
maxCount: If defined, will limit number of records in this Model's collection to this amount.
collectionName: Collection name to use for this model. If not defined, will use class name + 's'.

Fields

When you build your Model you specify one or more fields to include in the DB scheme. As briefly mentioned before, every field you define will create a getter and setter in your class with the same name. In addition, every field define validations and cleanup logic to trigger when writing / reading from DB, as well as validators to set in MongoDB itself (there are validations in app level and in Mongo's level).

There are several built-in Field types (its really easy to define new ones) and every field accept different arguments. However, all fields support the following:

default: Default value to set for the field.
source: If provided, will always copy value from a given property in the object before writing to DB. When loaded, will also set the property.
sourceReadonly: If 'source' is provided and 'sourceReadonly' is true, will only copy from source when writing to DB, but won't set the property when loaded.
canBeNull: Available for most types; allow null value.
index: If true, will make this field an index field.
unique: If true and indexed, will make this field unique.

Now lets describe all the field types:

GenericField

A generic field type without validations or cleanup logic.