Cómo construir una API GraphQL ultrarrápida con Node.js, MongoDB y Fastify

Este tutorial es la segunda parte de una serie de cuatro partes, cuyo objetivo es llevarlo desde cero hasta implementar una aplicación de pila completa completamente funcional.

  • Parte 1: Cómo crear API REST ultrarrápidas con Node.js, MongoDB, Fastify y Swagger
  • Parte 2: ¡Cómo construir una API GraphQL ultrarrápida con Node.js, MongoDB, Fastify y GraphQL! (Estás aquí.)
  • Parte 3: Acoplamiento de Vue.js con una API GraphQL .
  • Parte 4: Implementación de una API GraphQL y Vue.js aplicación frontend .

La primera parte de la serie está disponible aquí y el código fuente de la aplicación se puede encontrar aquí.

En esta parte revisaremos los modelos , controladores y rutas de la parte uno y luego integraremos GraphQL en la aplicación. Como beneficio adicional también usaremos Faker.jspara crear algunos datos falsos y sembrar la base de datos .

Introducción:

GraphQL es un lenguaje de consulta para API y un tiempo de ejecución para completar esas consultas con sus datos existentes.

Cada consulta GraphQL pasa por tres fases: las consultas se analizan, validan y ejecutan.

GraphQL proporciona una descripción completa y comprensible de los datos en su API, brinda a los clientes el poder de pedir exactamente lo que necesitan, facilita la evolución de las API con el tiempo y habilita poderosas herramientas de desarrollo. Aprende más.

Requisitos previos ...

Si ha completado la primera parte de esta serie, debe estar al día con los conocimientos de JavaScript para principiantes / intermedios , Node.js, Fastify.JS y MongoDB (Mongoose).

Para seguir adelante, deberá completar la primera parte de esta serie o tomar el código de Git, aunque recomendaría al menos leer la primera parte.

¡Empecemos!

Clone el repositorio para la parte uno (omita este paso si siguió la parte uno y continúa con su propio código) abriendo su terminal, navegando al directorio de su proyecto yejecutando cada una de las siguientes líneas de código:

git clone //github.com/siegfriedgrimbeek/fastify-api.git cd fastify-api

Entonces, ahora que tenemos una copia del código base, actualizaremos nuestros paquetes y package.jsonarchivos ejecutando el siguiente código:

sudo npm i -g npm-check-updates ncu -u npm install

Primero instalamos globalmente el paquete npm “ npm-check-updates ” y luego usamos este paquete para actualizar automáticamente nuestro package.jsonarchivo con las últimas versiones del paquete y luego instalamos / actualizamos todos nuestros módulos npm ejecutándose npm install.

Esto se hace para garantizar que todos los que completen el tutorial estén trabajando con las mismas versiones de paquete.

¡Refactorice nuestro servidor e inicie la aplicación!

Como sucede con todas las soluciones de software, a medida que la solución crece, los desarrolladores a menudo necesitan revisar y refactorizar el código.

En el srcdirectorio crearemos un nuevo archivo llamado server.js:

cd src touch server.js

Agregue el siguiente código de código al server.jsarchivo:

// Require the fastify framework and instantiate it const fastify = require('fastify')({ logger: true }) // Require external modules const mongoose = require('mongoose') // Connect to DB mongoose .connect('mongodb://localhost/mycargarage') .then(() => console.log('MongoDB connected...')) .catch(err => console.log(err)) module.exports = fastify

Ahora hemos extraído la lógica que inicia el servidor en el server.jsarchivo, lo que nos permite reutilizar este código en todo el proyecto.

A continuación, necesitamos actualizar nuestro index.jsarchivo en el srcdirectorio:

 // Import Server const fastify = require('./server.js') // Import Routes const routes = require('./routes') // Import Swagger Options const swagger = require('./config/swagger') // Register Swagger fastify.register(require('fastify-swagger'), swagger.options) // Loop over each route routes.forEach((route, index) => { fastify.route(route) }) // Run the server! const start = async () => { try { await fastify.listen(3000, '0.0.0.0') fastify.swagger() fastify.log.info(`server listening on ${fastify.server.address().port}`) } catch (err) { fastify.log.error(err) process.exit(1) } } start()

Volveremos a visitar el index.jsarchivo, una vez que configuremos GraphQL.

Inicie el servidor Fastify ejecutando el siguiente código en su terminal :

npm start

Tenga en cuenta que no hay una configuración de ruta predeterminada, por lo que, por ahora, navegar a // localhost: 3000 / dará como resultado que el servidor devuelva un error 404, que es correcto.

Inicie MongoDB y actualice los modelos

Extendamos el modelo existente para incluir también Servicios y Propietarios. El siguiente diagrama a continuación demuestra las relaciones entre las colecciones:

  • Un auto puede tener un dueño.
  • Un propietario puede tener muchos coches.
  • Un coche puede tener muchos servicios.

Vuelva a visitar el Car.jsarchivo en el modelsdirectorio y actualícelo de la siguiente manera:

// External Dependancies const mongoose = require("mongoose") const ObjectId = mongoose.Schema.Types.ObjectId const carSchema = new mongoose.Schema({ title: String, brand: String, price: String, age: Number, owner_id: ObjectId }) module.exports = mongoose.model("Car", carSchema)

Crear dos nuevos archivos en el modelsdirectorio, Owner.js y Service.jsy añadir el siguiente código a los archivos respectivamente:

Owner.js

// External Dependancies const mongoose = require('mongoose') const ownerSchema = new mongoose.Schema({ firstName: String, lastName: String, email: String }) module.exports = mongoose.model('Owner', ownerSchema)

Service.js

// External Dependancies const mongoose = require("mongoose") const ObjectId = mongoose.Schema.Types.ObjectId const serviceSchema = new mongoose.Schema({ car_id: ObjectId, name: String, date: String }) module.exports = mongoose.model("Service", serviceSchema) view rawService.js hosted with ❤ by GitHub

No se utilizan conceptos nuevos en el código anterior. Acabamos de crear esquemas estándar de Mongoose, como con el Car.jsmodelo.

Revisit the Car Controller and create the additional controllers

There are some slight changes to the carController.js so navigate to the controllers directory and update your file as per below:

// External Dependancies const boom = require('boom') // Get Data Models const Car = require('../models/Car') // Get all cars exports.getCars = async () => { try { const cars = await Car.find() return cars } catch (err) { throw boom.boomify(err) } } // Get single car by ID exports.getSingleCar = async req => { try { const id = req.params === undefined ? req.id : req.params.id const car = await Car.findById(id) return car } catch (err) { throw boom.boomify(err) } } // Add a new car exports.addCar = async req => { try { const car = new Car(req) const newCar = await car.save() return newCar } catch (err) { throw boom.boomify(err) } } // Update an existing car exports.updateCar = async req => { try { const id = req.params === undefined ? req.id : req.params.id const updateData = req.params === undefined ? req : req.params const update = await Car.findByIdAndUpdate(id, updateData, { new: true }) return update } catch (err) { throw boom.boomify(err) } } // Delete a car exports.deleteCar = async req => { try { const id = req.params === undefined ? req.id : req.params.id const car = await Car.findByIdAndRemove(id) return car } catch (err) { throw boom.boomify(err) } }

Create two new files in the controllers directory, serviceController.js and ownerController.js, and add the following code to the files respectively:

serviceController.js

// External Dependancies const boom = require('boom') // Get Data Models const Service = require('../models/Service') // Get single service ID exports.getSingleService = async req => { try { const id = req.params === undefined ? req.id : req.params.id const service = await Service.findById(id) return service } catch (err) { throw boom.boomify(err) } } // Get single car's services exports.getCarsServices = async req => { try { const id = req.params === undefined ? req.id : req.params.id const services = await Service.find({ car_id: id }) return services } catch (err) { throw boom.boomify(err) } }

ownerController.js

// External Dependancies const boom = require('boom') // Get Data Models const Owner = require('../models/Owner') const Car = require('../models/Car') // Get all owners exports.getOwner = async () => { try { const owners = await Owner.find() return owners } catch (err) { throw boom.boomify(err) } } // Get single owner by ID exports.getSingleOwner = async req => { try { const id = req.params === undefined ? req.id : req.params.id const owner = await Owner.findById(id) return owner } catch (err) { throw boom.boomify(err) } } // Get single owner's cars exports.getOwnersCars = async req => { try { const id = req.params === undefined ? req.id : req.params.id const cars = await Car.find({ owner_id: id }) return cars } catch (err) { throw boom.boomify(err) } }

The biggest change to the controllers is how we get the parameters:

const id = req.params === undefined ? req.id : req.params.id const updateData = req.params === undefined ? req : req.params

The above code is called a “conditional (ternary) operatorand is used as shorthand for the following if statement:

let id if (req.params === undefined) { id = req.id } else { id = req.params.id }

We are using the ternary operator to accommodate requests from both the REST API and the GraphQL API, as they have a slightly different implementation.

Time to seed the database with some fake data!

In the src directory let’s create a new directory and file by running the following code:

mkdir helpers touch seed.js

Add the following code to the seed.js file:

 // Import external dependancies const faker = require('faker') const boom = require('boom') // Import internal dependancies const fastify = require('../server.js') // Fake data const cars = [ { name: 'Tesla', models: ['S', 'E', 'X', 'Y'] }, { name: 'Mercedes', models: ['GLA', 'GLC', 'GLE', 'GLS'] }, { name: 'BMW', models: ['X4', 'Z3', 'M2', '7'] }, { name: 'Audi', models: ['A1', 'A3', 'A4', 'A5'] }, { name: 'Ford', models: ['Fiesta', 'Focus', 'Fusion', 'Mustang'] } ] const serviceGarages = ['A++ Auto Services', "Gary's Garage", 'Super Service', 'iGarage', 'Best Service'] // Get Data Models const Car = require('../models/Car') const Owner = require('../models/Owner') const Service = require('../models/Service') // Fake data generation functions const generateOwnerData = () => { let ownerData = [] let i = 0 while (i  { let carData = [] let i = 0 while (i  { let serviceData = [] let i = 0 while (i  { try { const owners = await Owner.insertMany(generateOwnerData()) const ownersIds = owners.map(x => x._id) const cars = await Car.insertMany(generateCarData(ownersIds)) const carsIds = cars.map(x => x._id) const services = await Service.insertMany(generateServiceData(carsIds)) console.log(` Data successfully added: - ${owners.length} owners added. - ${cars.length} cars added. - ${services.length} services added. `) } catch (err) { throw boom.boomify(err) } process.exit() }, err => { console.log('An error occured: ', err) process.exit() } )

Let’s break down this mountain of code:

First we import two external libraries, Faker.jswhich is used to generate fake data and Boom, which is used to throw http friendly error objects.

Then we import the server.js file which will spin up an instance of our server allowing us to interact with the models.

We then declare two arrays with fake data, cars and serviceGarages.

Then we import the models and declare three functions (generateOwnerData, generateCarData, generateServiceData) which each return an array of objects with the owner, car and service data respectively.

Once the Fastify.js instance is ready we use the Mongoose insertMany() function to insert the generated arrays into the database. The function then returns an array of objects containing the original object data and ids of the each record.

We use the JavaScript Map function to create an array of idsowners and cars arrays. We use the ownersIDs array for when generating car data and we use the carsIds array when generating service data, they are passed into the respective functions and then values are randomly selected from them.

Lastly we need to install the Faker.js package and add the seed task to our package.json file.

We can add the Faker.js package by navigating to the root directory and running the following code:

npm i faker -D

We then add the following to the package.json file:

... "scripts": { ... "seed": "node ./src/helpers/seed.js" }, ...

That’s it! We can now run our seeding script from the project root directory with the following code:

npm run seed

If you are using MongoDB Compass (you should), you will see the data in your database:

GraphQL installation, setup and testing

Let’s get started by navigating to the root directory and running the following code:

npm i fastify-gql graphql

The above installs GraphQL and the Fastify barebone GraphQL adapter.

Navigate to the src directory and run the following code:

mkdir schema cd shema touch index.js

Navigate to the src directory update the index.js file with the following:

// Import Server const fastify = require('./server.js') // Import external dependancies const gql = require('fastify-gql') // Import GraphQL Schema const schema = require('./schema') // Register Fastify GraphQL fastify.register(gql, { schema, graphiql: true }) ... end here // Import Routes const routes = require('./routes')

With the above code we require the Fastify GraphQL Adapter, import the schema and register the GraphQl Adapter with Fastify.

We register the schema and enable GraphiQL, an in-browser IDE for exploring GraphQL.

Navigate to the schema directory and open the index.js file and add the following boilerplate code:

// Import External Dependancies const graphql = require('graphql') // Destructure GraphQL functions const { GraphQLSchema, GraphQLObjectType, GraphQLString, GraphQLInt, GraphQLID, GraphQLList, GraphQLNonNull } = graphql // Import Controllers const carController = require('../controllers/carController') const ownerController = require('../controllers/ownerController') const serviceController = require('../controllers/serviceController') // Define Object Types const carType = new GraphQLObjectType({ name: 'Car', fields: () => ({}) }) const ownerType = new GraphQLObjectType({ name: 'Owner', fields: () => ({}) }) const serviceType = new GraphQLObjectType({ name: 'Service', fields: () => ({}) }) // Define Root Query const RootQuery = new GraphQLObjectType({ name: 'RootQueryType', fields: { car: {}, cars: {}, owner: {}, service: {} } }) // Define Mutations const Mutations = new GraphQLObjectType({ name: 'Mutations', fields: { addCar: { type: carType, args: {}, async resolve(args) { return '' } }, editCar: { type: carType, args: {}, async resolve(args) { return '' } }, deleteCar: { type: carType, args: {}, async resolve(args) { return '' } } } }) // Export the schema module.exports = new GraphQLSchema({ query: RootQuery, mutation: Mutations })

Let’s run through the above code:

We require the main GraphQL package and use JavaScript Destructuring to get the necessary GraphQL functions(GraphQLSchema, GraphQLObjectType, GraphQLString, GraphQLInt, GraphQLID, GraphQLList and GraphQLNonNull).

We import our three controllers (carController, ownerController and serviceController).

We declare the carType, ownerType and serviceTypeGraphQL Object Types, which are functions that accept an object as a parameter, with a name and a fields key.

These functions are used to define our GraphQL schema, similar to the Mongoose models defined earlier.

The fields can return a particular type, and methods that take arguments. Learn More about Object Types.

Then we declare the RootQuery which is also a GraphQL Object Type and is found at the top level of every GraphQL server. It represents all of the possible entry points into the GraphQL API. Learn More about root fields and resolvers.

We then declare our Mutations, which are used to change data. Although any query could be implemented to change data, operations that cause changes should be sent explicitly via a mutation. Learn More about Mutations.

Lastly we export the GraphQLSchema.

Now that we have our template setup we can start populating the Object Types, Root Query and Mutations.

Note that there are Mongoose to GraphQL schema generators available, but for the tutorial purposes we will manually create the schema.

Let’s update the carTypeObject Type as follows:

const carType = new GraphQLObjectType({ name: 'Car', fields: () => ({ _id: { type: GraphQLID }, title: { type: GraphQLString }, brand: { type: GraphQLString }, price: { type: GraphQLString }, age: { type: GraphQLInt }, owner_id: { type: GraphQLID }, owner: { type: ownerType, async resolve(parent, args) { return await ownerController.getSingleOwner({ id: parent.owner_id }) } }, services: { type: new GraphQLList(serviceType), async resolve(parent, args) { return await serviceController.getCarsServices({ id: parent._id }) } } }) })

Let’s dive deeper into the GraphQL functions, starting with the Scalars types in GraphQL:

GraphQL comes with a set of default scalar types out of the box:

  • Int: A signed 32‐bit integer. GraphQLInt
  • Float: A signed double-precision floating-point value. GraphQLFloat
  • String: A UTF‐8 character sequence. GraphQLString
  • Boolean: true or false. GraphQLBoolean
  • ID: The ID scalar type represents a unique identifier, often used to refetch an object or as the key for a cache. The ID type is serialised in the same way as a String; however, defining it as an ID signifies that it is not intended to be human‐readable. GraphQLID

The owner and service fields are where it gets interesting. These fields are not defined as Scalar types like the rest — instead, their type is referencing the ownerType and serviceType that we have created and are yet to populate.

El segundo argumento que pasamos a los campos ownery serviceson las funciones de resolución.

Las funciones o métodos de resolución son funciones que resuelven un valor para un tipo o campo en un esquema

¡Los resolutores también pueden ser asincrónicos! Pueden resolver valores de otra API REST, base de datos, caché, constante, etc.

Puede pensar en cada campo en una consulta GraphQL como una función o método del tipo anterior que devuelve el siguiente tipo. De hecho, así es exactamente como funciona GraphQL. Cada campo de cada tipo está respaldado por una función llamada resolver que es proporcionada por el desarrollador del servidor GraphQL. Cuando se ejecuta un campo, se llama al resolutor correspondiente para producir el siguiente valor.

If a field produces a scalar value like a string or number, then the execution completes. However if a field produces an object value then the query will contain another selection of fields which apply to that object. This continues until scalar values are reached. GraphQL queries always end at scalar values.

In order to create the relationship between the different types we pass the _id and the owner_id values into the respective controller functions.

So essentially we are requesting the owner details along with the car details:

return await userController.getSingleOwner({ id: parent.owner_id })

and the details of all the services related to the car:

return await serviceController.getCarsServices({ id: parent._id })

To return a list or array from with GraphQL, we use the GraphQLList. Here is a great in depth tutorial about using arrays in GraphQL Schema, but it is really simple: whenever we need an array we will use the GraphQLList function.

Let’s update the ownerType and serviceType with the following code:

ownerType

const ownerType = new GraphQLObjectType({ name: 'Owner', fields: () => ({ _id: { type: GraphQLID }, firstName: { type: GraphQLString }, lastName: { type: GraphQLString }, email: { type: GraphQLString }, cars: { type: new GraphQLList(carType), async resolve(parent, args) { return await ownerController.getOwnersCars({ id: parent._id }) } } }) })

serviceType

const serviceType = new GraphQLObjectType({ name: 'Service', fields: () => ({ _id: { type: GraphQLID }, car_id: { type: GraphQLID }, name: { type: GraphQLString }, date: { type: GraphQLString }, car: { type: carType, async resolve(parent, args) { return await carController.getSingleCar({ id: parent.car_id }) } } }) })

The above two Object Types are very similar to the carType. You can notice a pattern between the different Object Types and their relationships.

We can now populate the RootQuery root with the following code:

const RootQuery = new GraphQLObjectType({ name: 'RootQueryType', fields: { car: { type: carType, args: { id: { type: GraphQLID } }, async resolve(parent, args) { return await carController.getSingleCar(args) } }, cars: { type: new GraphQLList(carType), async resolve(parent, args) { return await carController.getCars() } }, owner: { type: ownerType, args: { id: { type: GraphQLID } }, async resolve(parent, args) { return await ownerController.getSingleOwner(args) } }, service: { type: serviceType, args: { id: { type: GraphQLID } }, async resolve(parent, args) { return await serviceController.getSingleService(args) } } } })

There are no new concepts in the above code, but keep in mind that the RootQuery query is the entry point to all queries on the GraphQL API. So from the above we can see that we can run the following queries directly:

  • Get all the Cars
  • Get a single Car
  • Get a single Owner
  • Get a single Service

Let’s open the GraphiQL user interface and build some queries: //localhost:3000/graphiql.html

Queries are entered on the left, results are in the middle, and the documentation explorer is on the right.

The documentation explorer can be used to explore the entire graph down to Scalar level. This is very helpful when building queries.

The language used to build the queries resembles JSON. This cheat sheet is a great a reference.

Below demonstrates why GraphQL is so awesome:

In the above example, we are using the cars root query to display a list of all the cars, their owners, and their services.

We have one final topic to address, and that is mutations. Let’s update the mutations with the following code:

const Mutations = new GraphQLObjectType({ name: 'Mutations', fields: { addCar: { type: carType, args: { title: { type: new GraphQLNonNull(GraphQLString) }, brand: { type: new GraphQLNonNull(GraphQLString) }, price: { type: GraphQLString }, age: { type: GraphQLInt }, owner_id: { type: GraphQLID } }, async resolve(parent, args) { const data = await carController.addCar(args) return data } }, editCar: { type: carType, args: { id: { type: new GraphQLNonNull(GraphQLID) }, title: { type: new GraphQLNonNull(GraphQLString) }, brand: { type: new GraphQLNonNull(GraphQLString) }, price: { type: new GraphQLNonNull(GraphQLString) }, age: { type: new GraphQLNonNull(GraphQLInt) }, owner_id: { type: GraphQLID } }, async resolve(parent, args) { const data = await carController.updateCar(args) return data } }, deleteCar: { type: carType, args: { id: { type: new GraphQLNonNull(GraphQLID) } }, async resolve(parent, args) { const data = await carController.deleteCar(args) return data } } } })

As before, we declare our Object Type, specify the name and the fields.

A mutation consists of the the type, args and the async resolve function. The resolve function passes the args to the controller, which returns the result of the mutation.

You have now coded a fully functional REST API and a fully functional GraphQL API.

There are no rules stating that one should use exclusively REST or exclusively GraphQL. In some projects, the best solution may be a mix of both. This is really determined on a project-to-project basis.

You can download the source code form Git here.

What is Next?

In the next tutorial, we will consume our GraphQLAPI with a Vue.js frontend as a single page application!