Categories
Express MongoDB Nodejs Vue 3

Create a Full Stack Web App with the MEVN Stack

The MEVN stack is a set of technologies for full-stack apps.

MEVN stands for MongoDB, Express, Vue.js, and Node.js

In this article, we’ll take a look at how to create a simple todo app with authentication with the MEVN stack.

Setting Up the Project

The first step to create a full-stack MEVN app is to set up the project.

First, we create a project folder, then add the backend and frontend folders inside it.

backend has the Express app.

frontend has the Vue.js app. We’ll use Vue 3 for the front end.

Next, to create our Express app, we run the Express Generator to create the files.

To do this, go into the backend and run:

npx express-generator

We may need admin privileges to do this.

Next, we go into the frontend folder.

We install the latest version of the Vue CLI by running:

npm install -g @vue/cli

Vue CLI 4.5 or later can create Vue 3 projects.

Then in the frontend folder, we run:

vue create .

Then we choose the ‘Vue 3 (default)’ option.

Next, we have to install the packages.

We stay in the frontend folder and run:

npm i axios vue-router@4.0.0-beta.12

to install the Axios HTTP client and Vue Router 4.x, which is compatible with Vue 3.

In the backend folder, we run:

npm i cors momgoose jsonwebtoken

to install the CORS and Mongoose packages to enable cross-domain communication and work with MongoDB.

jsonwebtoken lets us add JSON web token authentication into our app.

Create the Back End

Now we’re ready to create the back end app to work with MongoDB and add authentication.

First, we create db.js in the backend folder and add:

const { Schema, createConnection } = require('mongoose');
const connection = createConnection('mongodb://localhost:27017/mevn-example', { useNewUrlParser: true });

const userSchema = new Schema({
  name: String,
  password: String
});

const User = connection.model('User', userSchema);

const todoSchema = new Schema({
  name: String,
  done: Boolean,
  user: { type: Schema.Types.ObjectId, ref: 'User' },
});

const Todo = connection.model('Todo', todoSchema);

module.exports = {
  User,
  Todo
}

We connect to the MongoDB database with Mongoose and create the models.

To create the models, we use the Schema constructor to create the schema.

We add the properties and their data types.

The user property references the Object ID from the User model so that we can link the todo entry to the user.

Then we call connection.model to use the schema to create the model.

Then, we create constants.js in the backend folder and write:

module.exports = {
  SECRET: 'secret'
}

We put the secret for the JSON web token into this file and use them in the route files.

Next, we go into the routes folder and add the files for the API routes.

We create todos.js in the routes folder and write:

var express = require('express');
var router = express.Router();
const { Todo } = require('../db');
const jwt = require('jsonwebtoken');
const { SECRET } = require('../constants');

const verifyToken = (req, res, next) => {
  try {
    req.user = jwt.verify(req.headers.authorization, SECRET);
    return next();
  } catch (err) {
    console.log(err)
    return res.status(401);
  }
}

router.get('/', verifyToken, async (req, res) => {
  const { _id } = req.user;
  const todos = await Todo.find({ user: _id })
  res.json(todos);
});

router.get('/:id', verifyToken, async (req, res) => {
  const { _id } = req.user;
  const { id } = req.params;
  const todo = await Todo.findOne({ _id: id, user: _id })
  res.json(todo);
});

router.post('/', verifyToken, async (req, res) => {
  const { name } = req.body;
  const { _id } = req.user;
  const todo = new Todo({ name, done: false, user: _id })
  await todo.save()
  res.json(todo);
});

router.put('/:id', verifyToken, async (req, res) => {
  const { name, done } = req.body;
  const { id } = req.params;
  const todo = await Todo.findOneAndUpdate({ _id: id }, { name, done })
  await todo.save();
  res.json(todo);
});

router.delete('/:id', verifyToken, async (req, res) => {
  const { id } = req.params;
  await Todo.deleteOne({ _id: id })
  res.status(200).send();
});

module.exports = router;

The verifyToken middleware has the jwt.verify method to verify the token with our SECRET key.

If it’s valid, we set the req.user to the decoded token.

Then we call next to call the router handlers.

Then we have the GET / route to get the user data from req.user and then call Todo.find to find the todo entries for the user.

Todo is the model we created earlier.

The GET /:id route gets the Todo entry by the ID.

findOne gets the first result in the todos collection.

req.params gets the URL parameters from the URL parameter.

Then POST / route gets the name from the req.body property, which has the request body daya.

And we get the _id property from the req.user property to get the user data.

We use both to create the todo entry.

user has the Object ID for the user.

The PUT /:id route is used to update the user.

We get the name and done from req.body , which has the request body.

req.params has the id property which has the ID of the todo entry.

We use the id to find the todo entry.

And update the name and done property to update the entry.

Then we call todo.save() to save the data.

The DELETE /:id route lets us delete a todo entry by its ID.

We call deleteOne with the _id to do that.

The verifyToken will make sure the token is valid before we run the route handlers.

req.headers.authorization has the auth token. req.headers has the HTTP request headers.

Next, we create users.js in the routes folder and write:

var express = require('express');
var router = express.Router();
const bcrypt = require('bcrypt');
const { User } = require('../db');
const { SECRET } = require('../constants');
const jwt = require('jsonwebtoken');
const saltRounds = 10;

router.post('/register', async (req, res) => {
  const { name, password } = req.body;
  const existingUser = await User.findOne({ name });
  if (existingUser) {
    return res.json({ err: 'user already exists' }).status(401);
  }
  const hashedPassword = await bcrypt.hash(password, saltRounds);
  const user = new User({
    name,
    password: hashedPassword
  })
  await user.save();
  res.json(user).status(201);
});

router.post('/login', async (req, res) => {
  const { name, password } = req.body;
  const { _id, password: userPassword } = await User.findOne({ name });
  const match = await bcrypt.compare(password, userPassword);
  if (match) {
    const token = await jwt.sign({ name, _id }, SECRET);
    return res.json({ token });
  }
  res.status(401);
});

module.exports = router;

It’s similar to the todos.js .

We have the POST /register route to get the name and password properties from the JSON request body.

Then we check if the user with the given name exists.

Then we get the bcrypt.hash to hash the password with the secret.

And then we save the user with the name and then hashedPassword.

The POST /login route gets the name and password from the JSON request body.

We use the bcrypt.compare method to compare the password.

Then we create the token and return it in the response if validation is successful.

Otherwise, we send a 401 response.

Finally, we bring everything together in app.js in the backend folder.

We replace what’s there with:

var createError = require('http-errors');
var express = require('express');
var path = require('path');
var cookieParser = require('cookie-parser');
var logger = require('morgan');
var cors = require('cors')

var todorouter = require('./routes/todo');
var usersRouter = require('./routes/users');

var app = express();
app.use(cors())
// view engine setup
app.set('views', path.join(__dirname, 'views'));
app.set('view engine', 'jade');

app.use(logger('dev'));
app.use(express.json());
app.use(express.urlencoded({ extended: false }));
app.use(cookieParser());
app.use(express.static(path.join(__dirname, 'public')));

app.use('/todos', todorouter);
app.use('/users', usersRouter);

// catch 404 and forward to error handler
app.use(function(req, res, next) {
  next(createError(404));
});

// error handler
app.use(function(err, req, res, next) {
  // set locals, only providing error in development
  res.locals.message = err.message;
  res.locals.error = req.app.get('env') === 'development' ? err : {};

// render the error page
  res.status(err.status || 500);
  res.render('error');
});

module.exports = app;

We add the cors middleware so that we can communicate with the front end.

We have the todorouter and usersRouter to add the routes to our app.

Vue 3 Front End

Now that we finished the back end, we work on the front end.

First, in the components folder, we create TodoForm.vue and write:

<template>
  <div>
    <h1>{{ edit ? "Edit" : "Add" }} Todo</h1>
    <form @submit.prevent="submit">
      <div class="form-field">
        <label>Name</label>
        <br />
        <input v-model="form.name" />
      </div>
      <div>
        <label>Done</label>
        <input type="checkbox" v-model="form.done" />
      </div>
      <div>
        <input type="submit" value="Submit" />
      </div>
    </form>
  </div>
</template>

<script>
import axios from "axios";
import { APIURL } from "../constants";

export default {
  name: "TodoForm",
  data() {
    return {
      form: { name: "", done: false },
    };
  },
  props: {
    edit: Boolean,
    id: String,
  },
  methods: {
    async submit() {
      const { name, done } = this.form;
      if (!name) {
        return alert("Name is required");
      }
      if (this.edit) {
        await axios.put(`${APIURL}/todos/${this.id}`, { name, done });
      } else {
        await axios.post(`${APIURL}/todos`, { name, done });
      }
      this.$router.push("/todos");
    },
    async getTodo() {
      const { data } = await axios.get(`${APIURL}/todos/${this.id}`);
      this.form = data;
    },
  },
  beforeMount() {
    if (this.edit) {
      this.getTodo();
    }
  },
};
</script>

We have a form to let us add or edit to todo entries.

It takes the edit prop to indicate whether we’re editing a todo or not.

We use v-model to bind the inputs to the reactive properties.

Then we check for the this.edit value.

If it’s true , we make a PUT request with the id of the todo entry.

Otherwise, we make a POST request to create a new todo entry.

The getTodo method lets us get the todo by the ID when we’re editing.

Next, in the src folder, we create the views folder to add the route components.

We create the AddTodoForm.vue file and add:

<template>
  <div>
    <TodoForm></TodoForm>
  </div>
</template>

<script>
import TodoForm from "@/components/TodoForm";

export default {
  components: {
    TodoForm,
  },
};
</script>

We register the TodoForm component and render it in the template.

Next, we create the EditTodoForm.vue in the views folder and add:

<template>
  <div>
    <TodoForm edit :id='$route.params.id'></TodoForm>
  </div>
</template>

<script>
import TodoForm from "@/components/TodoForm";

export default {
  components: {
    TodoForm,
  },
};
</script>

We pass the edit and id props into the TodoForm so it can get todo entry by ID.

$route.params.id has the Object ID of the todo entry.

Next, we create the Login.vue file in the views folder to add a login form.

In this file, we add:

<template>
  <div>
    <h1>Login</h1>
    <form @submit.prevent="login">
      <div class="form-field">
        <label>Username</label>
        <br />
        <input v-model="form.name" type="text" />
      </div>
      <div class="form-field">
        <label>Password</label>
        <br />
        <input v-model="form.password" type="password" />
      </div>
      <div>
        <input type="submit" value="Log in" />
        <button type="button" @click="$router.push('/register')">
          Register
        </button>
      </div>
    </form>
  </div>
</template>

<script>
import axios from "axios";
import { APIURL } from "../constants";

export default {
  data() {
    return {
      form: { name: "", password: "" },
    };
  },
  methods: {
    async login() {
      const { name, password } = this.form;
      if (!name || !password) {
        alert("Username and password are required");
      }
      try {
        const {
          data: { token },
        } = await axios.post(`${APIURL}/users/login`, {
          name,
          password,
        });
        localStorage.setItem("token", token);
        this.$router.push("/todos");
      } catch (error) {
        alert("Invalid username or password.");
      }
    },
  },
};
</script>

We have a form that takes the username and password and let the user log in.

When we submit the form, the login method is called.

In the method, we check if the name and password have values.

If they both are nonempty, we proceed with making the login request.

If it succeeds, we go to the /todos route.

Otherwise, we should an error message.

Similarly, we create Register.vue component for the registration form.

Then we fill it with the following code:

<template>
  <div>
    <h1>Register</h1>
    <form @submit.prevent="register">
      <div class="form-field">
        <label>Username</label>
        <br />
        <input v-model="form.name" type="text" />
      </div>
      <div class="form-field">
        <label>Password</label>
        <br />
        <input v-model="form.password" type="password" />
      </div>
      <div>
        <input type="submit" value="Register" />
        <button type="button" @click="$router.push('/')">Login</button>
      </div>
    </form>
  </div>
</template>

<script>
import axios from "axios";
import { APIURL } from "../constants";

export default {
  data() {
    return {
      form: { name: "", password: "" },
    };
  },
  methods: {
    async register() {
      const { name, password } = this.form;
      if (!name || !password) {
        alert("Username and password are required");
      }
      try {
        await axios.post(`${APIURL}/users/register`, {
          name,
          password,
        });
        alert("Registration successful");
      } catch (error) {
        alert("Registration failed.");
      }
    },
  },
};
</script>

We get the username and password the same way.

If they’re both non-empty, then we make a request to the register route.

Once that succeeds, we show a success message.

Otherwise, we show a failure message, which is in the catch block.

Next, we have a component to show the todo items.

We create Todo.vue in the views folder and write:

<template>
  <div>
    <h1>Todos</h1>
    <button @click="$router.push(`/add-todo`)">Add Todo</button>
    <button @click="logOut">Logout</button>
    <table>
      <thead>
        <tr>
          <th>Name</th>
          <th>Done</th>
          <th>Edit</th>
          <th>Delete</th>
        </tr>
      </thead>
      <tbody>
        <tr v-for="t of todos" :key="t._id">
          <td>{{ t.name }}</td>
          <td>{{ t.done }}</td>
          <td>
            <button @click="$router.push(`/edit-todo/${t._id}`)">Edit</button>
          </td>
          <td>
            <button @click="deleteTodo(t._id)">Delete</button>
          </td>
        </tr>
      </tbody>
    </table>
  </div>
</template>

<script>
import axios from "axios";
import { APIURL } from "../constants";

export default {
  data() {
    return {
      todos: [],
    };
  },
  methods: {
    async getTodos() {
      const { data: todos } = await axios.get(`${APIURL}/todos`);
      this.todos = todos;
    },
    async deleteTodo(id) {
      await axios.delete(`${APIURL}/todos/${id}`);
      this.getTodos();
    },
    logOut() {
      localStorage.clear();
      this.$router.push("/");
    },
  },
  beforeMount() {
    this.getTodos();
  },
};
</script>

<style scoped>
th:first-child,
td:first-child {
  width: 60%;
}

th {
  text-align: left;
}
</style>

We have the Add Todo button that goes to the add-todo route when we click it.

This will be mapped to the AddTodoForm.vue component.

The Logout button calls the logOut method when it’s clicked.

It just clears local storage and redirect the user to the login page.

The getTodos method gets the todo entries for the user.

The user identity is determined from the decoded token.

The v-for directive renders the items and display them.

The Edit button goes to the edit form.

And the Delete button calls deleteTodo when we click it.

deleteTodo makes a DELETE request to the todos route.

We call getTodos after deleting a todo entry or when we load the page.

Next in src/App.vue , we add our router-view to show the route:

<template>
  <router-view></router-view>
</template>

<script>
export default {
  name: "App",
};
</script>

<style>
.form-field input {
  width: 100%;
}

#app {
  margin: 0 auto;
  width: 70vw;
}
</style>

Then we create constants.js in the src folder and add:

export const APIURL = 'http://localhost:3000'

to let us use the APIURL everywhere so that we don’t have to repeat the base URL for the back end API when we make requests.

Finally, in main.js , we write:

import { createApp } from 'vue'
import App from './App.vue'
import { createRouter, createWebHashHistory } from 'vue-router';
import Login from '@/views/Login';
import Todo from '@/views/Todo';
import Register from '@/views/Register';
import AddTodoForm from '@/views/AddTodoForm';
import EditTodoForm from '@/views/EditTodoForm';
import axios from 'axios';

axios.interceptors.request.use((config) => {
  if (config.url.includes('login') || config.url.includes('register')) {
    return config;
  }
  return {
    ...config, headers: {
      Authorization: localStorage.getItem("token"),
    }
  }
}, (error) => {
  return Promise.reject(error);
});

const beforeEnter = (to, from, next) => {
  const token = localStorage.getItem('token');
  if (token) {
    next()
    return true;
  }
  next({ path: '/' });
  return false
}

const routes = [
  { path: '/', component: Login },
  { path: '/register', component: Register },
  { path: '/todos', component: Todo, beforeEnter },
  { path: '/add-todo', component: AddTodoForm, beforeEnter },
  { path: '/edit-todo/:id', component: EditTodoForm, beforeEnter },
]

const router = createRouter({
  history: createWebHashHistory(),
  routes
})

const app = createApp(App);
app.use(router);
app.mount('#app')

to add the vue-router and the Axios request interceptor to add the auth token for authenticated routes.

beforeEnter is a route guard that lets us restrict access for authenticated routes.

We check if the token is present before we redirect the user to the page they’re going to.

Otherwise, we redirect to the login page.

routes has all the routes. path has the route URLs an component has the component we want to load.

beforeEnter is the route guard that loads before the route loads.

createRouter creates the router object.

createWebHashHistory lets us use hash mode for URLs.

So URLs will have the # sign between the base URL and the rest of the URL segments.

Then we call app.use(router) to add the router and make the this.$route and thuis.$router properties available in components.

Now we can run npm run serve from the frontend folder and node bin/www to start the back end.

Conclusion

We can create a simple MEVN stack app with the latest JavaScript frameworks and features without much trouble.

Categories
MongoDB

Using MongoDB with Mongoose — Populate Virtuals and Count

To make MongoDB database manipulation easy, we can use the Mongoose NPM package to make working with MongoDB databases easier.

In this article, we’ll look at how to use Mongoose to manipulate our MongoDB database.

Populate Virtuals: The Count Option

We can add the count option when we create a populate virtual to get the count of the number of children we retrieved.

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const PersonSchema = new Schema({
    name: String,
    band: String
  });

  const BandSchema = new Schema({
    name: String
  }, { toJSON: { virtuals: true } });

  BandSchema.virtual('numMembers', {
    ref: 'Person',
    localField: 'name',
    foreignField: 'band',
    count: true
  });

  const Person = db.model('Person', PersonSchema);
  const Band = db.model('Band', BandSchema);
  const person = new Person({ name: 'james', band: 'superband' });
  await person.save();
  const band = new Band({ name: 'superband' });
  await band.save();
  const doc = await Band.findOne({ name: 'superband' })
    .populate('numMembers');
  console.log(doc.numMembers);
}
run();

We created the PersonSchema and BandSchema schema objects.

We call the virtual method on the BandSchema schema with several options.

The ref property is the model that the Band model is referencing.

localField is the field in the BandSchema we want to join with PeronSchema .

foreignField is the field in the ForeignSchema we want to join with PersonSchema .

count set to true means that we get the count. numMembers is the field name that we get the count from.

Then we save a Person and Band document with the same name.

Then we call populate with the numMembers field to get the number of members.

And finally, we get the numMembers field from the retrieved result to get how many Person children entries are in the Band .

Populate in Middleware

We can add pre or post hooks to populate operations.

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const PersonSchema = new Schema({
    name: String,
    band: String
  });

  const BandSchema = new Schema({
    name: String
  }, { toJSON: { virtuals: true } });

  BandSchema.virtual('numMembers', {
    ref: 'Person',
    localField: 'name',
    foreignField: 'band',
    count: true
  });

  BandSchema.pre('find', function () {
   this.populate('person');
  });

  BandSchema.post('find', async (docs) => {
    for (const doc of docs) {
      await doc.populate('person').execPopulate();
    }
  });

  const Person = db.model('Person', PersonSchema);
  const Band = db.model('Band', BandSchema);
  const person = new Person({ name: 'james', band: 'superband' });
  await person.save();
  const band = new Band({ name: 'superband' });
  await band.save();
  const doc = await Band.findOne({ name: 'superband' })
    .populate('numMembers');
  console.log(doc.numMembers);
}
run();

to add the pre and post hooks to listen to the find operation.

We should always call populate with the given field to do the population.

Conclusion

We can add the count property to add the count and also add pre and post hooks for find operations with when we do the population.

Categories
MongoDB

Using MongoDB with Mongoose — Populate Virtuals

To make MongoDB database manipulation easy, we can use the Mongoose NPM package to make working with MongoDB databases easier.

In this article, we’ll look at how to use Mongoose to manipulate our MongoDB database.

Populate Virtuals

We can control how 2 models are joined together.

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const PersonSchema = new Schema({
    name: String,
    band: String
  });

  const BandSchema = new Schema({
    name: String
  });

  BandSchema.virtual('members', {
    ref: 'Person',
    localField: 'name',
    foreignField: 'band',
    justOne: false,
    options: { sort: { name: -1 }, limit: 5 }
  });

  const Person = db.model('Person', PersonSchema);
  const Band = db.model('Band', BandSchema);
  const person = new Person({ name: 'james', band: 'superband' });
  await person.save();
  const band = new Band({ name: 'superband' });
  await band.save();
  const bands = await Band.find({}).populate('members').exec();
  console.log(bands[0].members);
}
run();

We create the PersonSchema as usual, but the BandSchema is different.

We call the virtual method with the join field call members to get the persons with the band name set to a given name.

The ref property is the name of the model we want to join.

localField is the field of the BandSchema that we want to join with PersonSchema .

The foreignField is the field of PersonSchema that we want to join with the BandSchema .

justOne means we only return the first entry of the join.

options has the options for querying.

Virtuals aren’t included in the toJSON() output by default.

If we want populate virtual to show when using functions that rely on JSON.stringify() , then add the virtuals option and set it to true .

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const PersonSchema = new Schema({
    name: String,
    band: String
  });

  const BandSchema = new Schema({
    name: String
  }, { toJSON: { virtuals: true } });

  BandSchema.virtual('members', {
    ref: 'Person',
    localField: 'name',
    foreignField: 'band',
    justOne: false,
    options: { sort: { name: -1 }, limit: 5 }
  });

  const Person = db.model('Person', PersonSchema);
  const Band = db.model('Band', BandSchema);
  const person = new Person({ name: 'james', band: 'superband' });
  await person.save();
  const band = new Band({ name: 'superband' });
  await band.save();
  const bands = await Band.find({}).populate('members').exec();
  console.log(bands[0].members);
}
run();

to add the virtuals option to the BandSchema .

If we use populate projections, then foreignField should be included in the projection:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const PersonSchema = new Schema({
    name: String,
    band: String
  });

  const BandSchema = new Schema({
    name: String
  }, { toJSON: { virtuals: true } });

  BandSchema.virtual('members', {
    ref: 'Person',
    localField: 'name',
    foreignField: 'band',
    justOne: false,
    options: { sort: { name: -1 }, limit: 5 }
  });

  const Person = db.model('Person', PersonSchema);
  const Band = db.model('Band', BandSchema);
  const person = new Person({ name: 'james', band: 'superband' });
  await person.save();
  const band = new Band({ name: 'superband' });
  await band.save();
  const bands = await Band.find({}).populate({ path: 'members', select: 'name band' }).exec();
  console.log(bands[0].members);
}
run();

We call populate with an object with the path to get the virtual field and the select property has a string with the field names that we want to get separated by a space.

Conclusion

We can use Mongoose populate virtual feature to join 2 models by a column other than an ID column.

Categories
MongoDB

Using MongoDB with Mongoose — Discriminators

To make MongoDB database manipulation easy, we can use the Mongoose NPM package to make working with MongoDB databases easier.

In this article, we’ll look at how to use Mongoose to manipulate our MongoDB database.

Discriminators

Discriminators are a schema inheritance mechanism.

They let us enable multiple models to have overlapping schemas on top of the same MongoDB collection.

For example, we can use them as follows:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const options = { discriminatorKey: 'kind' };

  const eventSchema = new Schema({ time: Date }, options);
  const Event = db.model('Event', eventSchema);
  const ClickedLinkEvent = Event.discriminator('ClickedLink',
    new Schema({ url: String }, options));
  const genericEvent = new Event({ time: Date.now(), url: 'mongodb.com' });
  console.log(genericEvent)

  const clickedEvent =
    new ClickedLinkEvent({ time: Date.now(), url: 'mongodb.com' });
  console.log(clickedEvent)
}
run();

We created an Event model from the eventSchema .

It has the discriminatorKey so that we get can discriminate between the 2 documents we create later.

To create the ClickedLinkEvent model, we call Event.discriminator to create a model by inheriting from the Event schema.

We add the url field to the ClickedLinkEvent model.

Then when we add the url to the Event document and the ClickedLinkEvent document, only the clickedEvent object has the url property.

We get:

{ _id: 5f6f78f17f83ca22408eb627, time: 2020-09-26T17:22:57.690Z }

as the value of genericEvent and:

{
  _id: 5f6f78f17f83ca22408eb628,
  kind: 'ClickedLink',
  time: 2020-09-26T17:22:57.697Z,
  url: 'mongodb.com'
}

as the value of clickedEvent .

Discriminators Save to the Model’s Collection

We can save different kinds of events all at once.

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const options = { discriminatorKey: 'kind' };

  const eventSchema = new Schema({ time: Date }, options);
  const Event = db.model('Event', eventSchema);

  const ClickedLinkEvent = Event.discriminator('ClickedLink',
    new Schema({ url: String }, options));

  const SignedUpEvent = Event.discriminator('SignedUp',
    new Schema({ user: String }, options));

  const event1 = new Event({ time: Date.now() });
  const event2 = new ClickedLinkEvent({ time: Date.now(), url: 'mongodb.com' });
  const event3 = new SignedUpEvent({ time: Date.now(), user: 'mongodbuser' });
  await Promise.all([event1.save(), event2.save(), event3.save()]);
  const count = await Event.countDocuments();
  console.log(count);
}
run();

We created the eventSchema as an ordinary schema.

And the rest of the models are created from the Event.discriminator method.

Then we created the models and saved them all.

And finally, we called Event.countDocuments to get the number of items saved under the Event model.

Then count should be 3 since ClickedLinkEvent and SignedUpEvent both inherit from Event itself.

Discriminator Keys

We can tell the difference between each type of models with the __t property by default.

For instance, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');

  const eventSchema = new Schema({ time: Date });
  const Event = db.model('Event', eventSchema);

  const ClickedLinkEvent = Event.discriminator('ClickedLink',
    new Schema({ url: String }));

  const SignedUpEvent = Event.discriminator('SignedUp',
    new Schema({ user: String }));

  const event1 = new Event({ time: Date.now() });
  const event2 = new ClickedLinkEvent({ time: Date.now(), url: 'mongodb.com' });
  const event3 = new SignedUpEvent({ time: Date.now(), user: 'mongodbuser' });
  await Promise.all([event1.save(), event2.save(), event3.save()]);
  console.log(event1.__t);
  console.log(event2.__t);
  console.log(event3.__t);
}
run();

to get the type of data that’s saved from the console logs. We should get:

undefined
ClickedLink
SignedUp

logged.

We can add the discriminatorKey in the options to change the discriminator key.

So we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const options = { discriminatorKey: 'kind' };
  const eventSchema = new Schema({ time: Date }, options);
  const Event = db.model('Event', eventSchema);

const ClickedLinkEvent = Event.discriminator('ClickedLink',
    new Schema({ url: String }, options));

const SignedUpEvent = Event.discriminator('SignedUp',
    new Schema({ user: String }, options));

const event1 = new Event({ time: Date.now() });
  const event2 = new ClickedLinkEvent({ time: Date.now(), url: 'mongodb.com' });
  const event3 = new SignedUpEvent({ time: Date.now(), user: 'mongodbuser' });
  await Promise.all([event1.save(), event2.save(), event3.save()]);
  console.log(event1.kind);
  console.log(event2.kind);
  console.log(event3.kind);
}
run();

to set the options for each model and then access the kind property instead of __t and get the same result as before.

Conclusion

We can use discriminators to create new models by inheriting one model from another.

Categories
MongoDB

Using MongoDB with Mongoose — Discriminators and Hooks

To make MongoDB database manipulation easy, we can use the Mongoose NPM package to make working with MongoDB databases easier.

In this article, we’ll look at how to use Mongoose to manipulate our MongoDB database.

Discriminators and Queries

Queries are smart enough to take into account discriminators.

For example, if we have:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const options = { discriminatorKey: 'kind' };
  const eventSchema = new Schema({ time: Date }, options);
  const Event = db.model('Event', eventSchema);

  const ClickedLinkEvent = Event.discriminator('ClickedLink',
    new Schema({ url: String }, options));

  const SignedUpEvent = Event.discriminator('SignedUp',
    new Schema({ user: String }, options));

  const event1 = new Event({ time: Date.now() });
  const event2 = new ClickedLinkEvent({ time: Date.now(), url: 'mongodb.com' });
  const event3 = new SignedUpEvent({ time: Date.now(), user: 'mongodbuser' });
  await Promise.all([event1.save(), event2.save(), event3.save()]);
  const clickedLinkEvent =  await ClickedLinkEvent.find({});
  console.log(clickedLinkEvent);
}
run();

We called the ClickedLinkEvent.find method.

Therefore, we’ll get all the ClickedLinkEvent instances.

Discriminators Pre and Post Hooks

We can add pre and post hooks to schemas created with discriminators.

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const options = { discriminatorKey: 'kind' };
  const eventSchema = new Schema({ time: Date }, options);
  const Event = db.model('Event', eventSchema);

  const clickedLinkSchema = new Schema({ url: String }, options)
  clickedLinkSchema.pre('validate', (next) => {
    console.log('validate click link');
    next();
  });
  const ClickedLinkEvent = Event.discriminator('ClickedLink',
    clickedLinkSchema);

  const event1 = new Event({ time: Date.now() });
  const event2 = new ClickedLinkEvent({ time: Date.now(), url: 'mongodb.com' });
  await event2.validate();
}
run();

to add a pre hook for the validate operation to the clickedLinkSchema .

Handling Custom _id Fields

If an _id field is set on the base schema, then it’ll always override the discriminator’s _id field.

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const options = { discriminatorKey: 'kind' };

  const eventSchema = new Schema({ _id: String, time: Date },
    options);
  const Event = db.model('BaseEvent', eventSchema);

  const clickedLinkSchema = new Schema({
    url: String,
    time: String
  }, options);

  const ClickedLinkEvent = Event.discriminator('ChildEventBad',
    clickedLinkSchema);

  const event1 = new ClickedLinkEvent({ _id: 'custom id', time: '12am' });
  console.log(typeof event1._id);
  console.log(typeof event1.time);
}
run();

And from the console log, we can see that both the _id and time fields of event1 are strings.

So the _id field is the same one as the eventSchema , but the ClickedLinkEvent field has the same type as the one in clickedLinkSchema .

Using Discriminators with Model.create()

We can use discriminators with the Model.create method.

For example, we can write:

async function run() {
  const { createConnection, Types, Schema } = require('mongoose');
  const db = createConnection('mongodb://localhost:27017/test');
  const shapeSchema = new Schema({
    name: String
  }, { discriminatorKey: 'kind' });

  const Shape = db.model('Shape', shapeSchema);

  const Circle = Shape.discriminator('Circle',
    new Schema({ radius: Number }));
  const Square = Shape.discriminator('Square',
    new Schema({ side: Number }));

  const shapes = [
    { name: 'Test' },
    { kind: 'Circle', radius: 5 },
    { kind: 'Square', side: 10 }
  ];
  const [shape1, shape2, shape3] = await Shape.create(shapes);
  console.log(shape1 instanceof Shape);
  console.log(shape2 instanceof Circle);
  console.log(shape3 instanceof Square);
}
run();

We created 3 schemas for shapes with the discriminator method.

Then we called Shape.create with an array of different shape objects.

We specified the type with the kind property since we set that as the discriminator key.

Then in the console log, they should all log true since we specified the type of each entry.

If it’s not specified, then it has the base type.

Conclusion

We can add hooks to schemas created from discriminators.

_id fields are handled differently from other discriminator fields.

And we can use the create method with discriminators.