Dans ce tout premier article consacré à Express, nous répondons aux questions « Qu'est-ce que Node ? » (réponse : une bibliothèque) et « Qu'est-ce que Express? », et vous donnons un aperçu de ce qui fait d'Express un framework web si spécial. Nous décrirons les principales fonctionnalités et vous montrerons quelques-uns des principaux composants d'une application Express (bien que vous ne disposiez pas encore d'un environnement de développement pour le tester).
| Prérequis : | Une culture de base en informatique, une compréhension globale de la programmation côté serveur et, en particulier, les mécanismes d'interactions client-serveur dans un site web. |
|---|---|
| Objectif : | Devenir familier avec ce qu'est Express et comment il s'intégre dans Node, les fonctionnalités qu'il apporte, et les principales étapes pour construire une application Express. |
Node (ou plus formellement Node.js) est un environnement d'exécution open-source, multi-plateforme, qui permet aux développeurs de créer toutes sortes d'applications et d'outils côté serveur en JavaScript. Cet environnement est destiné à être utilisé en dehors du navigateur (il s'exécute directement sur son ordinateur ou dans le système d'exploitation du serveur). Il ignore les API JavaScript liées au navigateur et ajoute du support pour des API plus traditionnelles, comprenant HTTP et des librairies de fichiers système.
Dans la perpective d'un développement serveur web, Node possède de nombreux atouts :
Vous pouvez utiliser Node.js pour créer un simple serveur web en utilisant le paquet Node HTTP.
L'exemple qui suit crée un serveur web qui écoute toutes sortes de requêtes HTTP sur l'URL http://127.0.0.1:8000/. Quand une requête est reçue, le script répond avec la chaine "Salut tout le monde". Si vous avez déjà installé Node, suivez les étapes de l'exemple suivant :
"test-node" et placez-vous dedans en utilisant la commande suivante dans votre console :cd test-node
"hello.js" et collez ce qui suit dedans :// Charge le module HTTP
const http = require("http");
const hostname = "127.0.0.1";
const port = 8000;
// Crée un serveur HTTP
const server = http.createServer((req, res) {
// Configure l'en-tête de la réponse HTTP avec le code du statut et le type de contenu
response.writeHead(200, {'Content-Type': 'text/plain'});
// Envoie le corps de la réponse "Salut tout le monde"
response.end('Salut tout le monde\n');
})
// Démarre le serveur à l'adresse 127.0.0.1 sur le port 8000
// Affiche un message dès que le serveur commence à écouter les requêtes
server.listen(port, hostname, () => {
console.log(`Le serveur tourne à l'adresse http://${hostname}:${port}/`);
})
node hello.js
Puis saisissez l'URL "http://localhost:8000" dans votre navigateur. Vous devriez alors voir "Salut tout le monde" en haut à gauche d'une page web ne contenant rien d'autre que ce texte.
D'autres tâches de développement web ne sont pas directement supportées par Node lui-même. Si vous voulez ajouter différentes manipulations pour divers requêtes HTTP (GET, POST, DELETE, etc.), gérer différemment des requêtes vers plusieurs chemins URL ("routes"), servir des pages statiques ou utiliser des modèles pour créer dynamiquement la réponse, alors vous devrez écrire tout le code vous-même ou, pour éviter de réinventer la roue, vous servir des cadres applicatifs web (frameworks).
Express est le cadre applicatif actuellement le plus populaire dans Node et est la bibliothèque sous-jacente pour un grand nombre d'autres cadres applicatifs web pour Node. Il fournit des mécanismes pour :
Bien qu'Express soit assez minimaliste, les développeurs ont créé des middlewares (fonctions intermédiaires) compatibles pour résoudre quasiment tous les problèmes de développement web. Il existe des librairies pour se servir des cookies, gérer les sessions, la connexion utilisateur, les paramètres de l'URL, les données POST, les entêtes de sécurité et d'autres encore. Vous trouverez une liste des paquets maintenus par l'équipe Express ici : Express Middleware (ainsi que la liste de paquets de tierce-parties populaires).
Note: Cette flexibilité est à double tranchant. Il y a une multitude de paquets pour résoudre chaque problème mais trouver le bon paquet à utiliser peut vite devenir un challenge. Il n'y a pas non plus de "bonne manière" pour structurer une application et beaucoup d'exemples que vous trouverez sur le net ne sont pas optimisés ou montrent seulement une infime partie de ce que vous devez faire pour développer une application web.
À ses débuts en 2009, Node a été publié pour Linux uniquement. Le gestionnaire de paquets NPM est sorti en 2010, et le support natif de Windows fut ajouté en 2012. Aujourd'hui (juin 2020), la version courante ("current release") de Node est la 14.4.0 et la dernière version publiée est la 12.18.1. Ceci est un très court aperçu d'une aventure riche en rebondissements. Allez creuser ça sur Wikipedia si vous voulez en savoir plus.
Express est sorti pour la première fois en Novembre 2010 et est aujourd'hui (juin 2020) à sa version 4.17.1 de l'API. Vous pouvez consulter la liste des modifications pour plus d'informations sur la version courante et GitHub pour plus de détails sur l'historique des publications.
La popularité d'un "framework" web est importante car elle conditionne la maintenance dans le temps et les ressources qu'il est raisonnable de mettre à disposition dans la documentation, les librairies d'extensions et le support technique.
Il n'existe pas d'échelle de mesures définitive et fiable pour l'estimation de la popularité des "frameworks" côté serveur, bien que des sites comme Hot Frameworks essaient d'estimer la popularité par le comptage du nombre de projets Github ou StackOverflow. La question est : " Est-ce que Node et Express sont suffisamment populaires pour pouvoir s'affranchir des plateformes non-populaires ? Continuent-ils à évoluer ? Pouvez-vous avoir de l'aide si besoin ? Existe-til une opportunité pour vous de gagner de l'argent si vous apprenez Express ?".
Si on se réfère à la liste de compagnies utilisant Express, la quantité de gens contribuant au code et le nombre de gens fournissant un support payant ou bien gratuit, alors oui, Express est un framework populaire !
Les cadres logiciels web se décrivent souvent comme étant "opiniâtres" ou non (opinionated en anglais).
Les cadres logiciels opiniâtres sont ceux qui ont un avis arrêté sur la "bonne manière" de gérer certaines tâches. Ils fournissent souvent un cadre permettant de développer rapidement dans un domaine particulier (résolvant des problèmes d'un type particulier) parce que la bonne manière de faire quoi que ce soit est généralement bien comprise et bien documentée. Toutefois, ils peuvent manquer de flexibilité pour la résolution de problèmes hors de de leur portée et tendent à offrir peu de choix concernant les composants et approches qu'ils peuvent utiliser.
Les cadres logiciels non opiniâtres, par contraste, ont beaucoup moins de restrictions sur la meilleure manière d'assembler des composants ensemble pour atteindre un objectif, ou encore sur les composants que vous devriez utiliser. Ils laissent aux développeurs la possibilité d'utiliser les outils les plus adaptés pour achever une tâche particulière, bien que celà nécessite que vous cherchiez et trouviez ces composants par vous-même.
Express n'est pas opiniâtre. Vous pouvez intégrer quasiment n'importe quelle fonction intermédiaire compatible que vous le désirez dans la pile de gestion de requête, dans quasiment n'importe quel ordre. Vous pouvez structurer l'application en un fichier comme en plusieurs, et utiliser n'importe quelle structure de dossiers. Vous pourrez même quelquefois vous sentir perdu par la liberté que vous avez de vous organiser comme vous le souhaitez !
Dans un site web traditionnellement orienté données, une application web attend des requêtes HTTP du navigateur web (ou d'un autre client). Quand une requête est reçue, l'application cherche quelle action est requise en fonction du modèle de l'URL et des possibles informations associés contenues dans les données POST ou GET. Selon ce qui est requis, il pourra alors lire ou écrire des informations dans une une base de données ou effectuer d'autre tâches requises pour satisfaire la requête. L'application va alors retourner une réponse au navigateur web, souvent une page HTML créée dynamiquement pour le navigateur, en intégrant les données récupérées dans un modèle HTML.
Express fournit des méthodes pour spécifier quelle fonction est appelée pour une méthode HTTP particulière (GET, POST, SET, etc.) et un modèle d'URL ("Route"), ainsi que des méthodes pour spécifier quel moteur de rendu de vues ("view") est utilisé, où sont les modèles de vues et quel modèle utiliser pour générer une réponse. Vous pouvez utiliser les fonctions intermédiaires d'Express pour ajouter un support pour les cookies, les sessions, les utilisateurs, obtenir les paramètres POST/GET, etc. Vous pouvez utliser n'importe que système de base données supporté par Node (Express ne définit aucun comportement relatif au bases de données).
Les sections suivantes expliquent quelques choses communes que vous verrez en travaillant avec du code Express et Node.
First lets consider the standard Express Hello World example (we discuss each part of this below, and in the following sections).
Tip: If you have Node and Express already installed (or if you install them as shown in the next article), you can save this code in a text file called app.js and run it in a bash command prompt by calling:
./node ./app.js
var express = require('express');
var app = express();
app.get('/', function(req, res) {
res.send('Hello World!');
});
app.listen(3000, function() {
console.log('Example app listening on port 3000!');
});
The first two lines require() (import) the express module and create an Express application. This object, which is traditionally named app, has methods for routing HTTP requests, configuring middleware, rendering HTML views, registering a template engine, and modifying application settings that control how the application behaves (e.g. the environment mode, whether route definitions are case sensitive, etc.)
The middle part of the code (the three lines starting with app.get) shows a route definition. The app.get() method specifies a callback function that will be invoked whenever there is an HTTP GET request with a path ('/') relative to the site root. The callback function takes a request and a response object as arguments, and simply calls send() on the response to return the string "Hello World!"
The final block starts up the server on port '3000' and prints a log comment to the console. With the server running, you could go to localhost:3000 in your browser to see the example response returned.
A module is a JavaScript library/file that you can import into other code using Node's require() function. Express itself is a module, as are the middleware and database libraries that we use in our Express applications.
The code below shows how we import a module by name, using the Express framework as an example. First we invoke the require() function, specifying the name of the module as a string ('express'), and calling the returned object to create an Express application. We can then access the properties and functions of the application object.
var express = require('express');
var app = express();
You can also create your own modules that can be imported in the same way.
Tip: You will want to create your own modules, because this allows you to organise your code into managable parts — a monolithic single-file application is hard to understand and maintain. Using modules also helps you manage your namespace, because only the variables you explicitly export are imported when you use a module.
To make objects available outside of a module you just need to assign them to the exports object. For example, the square.js module below is a file that exports area() and perimeter() methods:
exports.area = function(width) { return width * width; };
exports.perimeter = function(width) { return 4 * width; };
We can import this module using require(), and then call the exported method(s) as shown:
var square = require('./square'); // Here we require() the name of the file without the (optional) .js file extension
console.log('The area of a square with a width of 4 is ' + square.area(4));
Note: You can also specify an absolute path to the module (or a name, as we did initially).
If you want to export a complete object in one assignment instead of building it one property at a time, assign it to module.exports as shown below (you can also do this to make the root of the exports object a constructor or other function):
module.exports = {
area: function(width) {
return width * width;
},
perimeter: function(width) {
return 4 * width;
}
};
For a lot more information about modules see Modules (Node API docs).
JavaScript code frequently uses asynchronous rather than synchronous APIs for operations that may take some time to complete. A synchronous API is one in which each operation must complete before the next operation can start. For example, the following log functions are synchronous, and will print the text to the console in order (First, Second).
console.log('First');
console.log('Second');
By contrast, an asynchronous API is one in which the API will start an operation and immediately return (before the operation is complete). Once the operation finishes, the API will use some mechanism to perform additional operations. For example, the code below will print out "Second, First" because even though setTimeout() method is called first, and returns immediately, the operation doesn't complete for several seconds.
setTimeout(function() {
console.log('First');
}, 3000);
console.log('Second');
Using non-blocking asynchronous APIs is even more important on Node than in the browser, because Node is a single threaded event-driven execution environment. "single threaded" means that all requests to the server are run on the same thread (rather than being spawned off into separate processes). This model is extremely efficient in terms of speed and server resources, but it does mean that if any of your functions call synchronous methods that take a long time to complete, they will block not just the current request, but every other request being handled by your web application.
There are a number of ways for an asynchronous API to notify your application that it has completed. The most common way is to register a callback function when you invoke the asynchronous API, that will be called back when the operation completes. This is the approach used above.
Tip: Using callbacks can be quite "messy" if you have a sequence of dependent asynchronous operations that must be performed in order, because this results in multiple levels of nested callbacks. This problem is commonly known as "callback hell". This problem can be reduced by good coding practices (see http://callbackhell.com/), using a module like async, or even moving to ES6 features like Promises.
Note: A common convention for Node and Express is to use error-first callbacks. In this convention the first value in your callback functions is an error value, while subsequent arguments contain success data. There is a good explanation of why this approach is useful in this blog: The Node.js Way - Understanding Error-First Callbacks (fredkschott.com).
In our Hello World Express example (see above), we defined a (callback) route handler function for HTTP GET requests to the site root ('/').
app.get('/', function(req, res) {
res.send('Hello World!');
});
The callback function takes a request and a response object as arguments. In this case the method simply calls send() on the response to return the string "Hello World!" There are a number of other response methods for ending the request/response cycle, for example you could call res.json() to send a JSON response or res.sendFile() to send a file.
JavaScript tip: You can use any argument names you like in the callback functions; when the callback is invoked the first argument will always be the request and the second will always be the response. It makes sense to name them such that you can identify the object you're working with in the body of the callback.
The Express application object also provides methods to define route handlers for all the other HTTP verbs, which are mostly used in exactly the same way: post(), put(), delete(), options(), trace(), copy(), lock(), mkcol(), move(), purge(), propfind(), proppatch(), unlock(), report(), mkactivity(), checkout(), merge(), m-search(), notify(), subscribe(), unsubscribe(), patch(), search(), and connect().
There is a special routing method, app.all(), which will be called in response to any HTTP method. This is used for loading middleware functions at a particular path for all request methods. The following example (from the Express documentation) shows a handler that will be executed for requests to /secret irrespective of the HTTP verb used (provided it is supported by the http module).
app.all('/secret', function(req, res, next) {
console.log('Accessing the secret section ...');
next(); // pass control to the next handler
});
Routes allow you to match particular patterns of characters in a URL, and extract some values from the URL and pass them as parameters to the route handler (as attributes of the request object passed as a parameter).
Often it is useful to group route handlers for a particular part of a site together and access them using a common route-prefix (e.g. a site with a Wiki might have all wiki-related routes in one file and have them accessed with a route prefix of /wiki/). In Express this is achieved by using the express.Router object. For example, we can create our wiki route in a module named wiki.js, and then export the Router object, as shown below:
// wiki.js - Wiki route module
var express = require('express');
var router = express.Router();
// Home page route
router.get('/', function(req, res) {
res.send('Wiki home page');
});
// About page route
router.get('/about', function(req, res) {
res.send('About this wiki');
});
module.exports = router;
Note: Adding routes to the Router object is just like adding routes to the app object (as shown previously).
To use the router in our main app file we would then require() the route module (wiki.js), then call use() on the Express application to add the Router to the middleware handling path. The two routes will then be accessible from /wiki/ and /wiki/about/.
var wiki = require('./wiki.js');
// ...
app.use('/wiki', wiki);
We'll show you a lot more about working with routes, and in particular about using the Router, later on in the linked section Routes and controllers .
Middleware is used extensively in Express apps, for tasks from serving static files to error handling, to compressing HTTP responses. Whereas route functions end the HTTP request-response cycle by returning some response to the HTTP client, middleware functions typically perform some operation on the request or response and then call the next function in the "stack", which might be more middleware or a route handler. The order in which middleware is called is up to the app developer.
Note: The middleware can perform any operation, execute any code, make changes to the request and response object, and it can also end the request-response cycle. If it does not end the cycle then it must call next() to pass control to the next middleware function (or the request will be left hanging).
Most apps will use third-party middleware in order to simplify common web development tasks like working with cookies, sessions, user authentication, accessing request POST and JSON data, logging, etc. You can find a list of middleware packages maintained by the Express team (which also includes other popular 3rd party packages). Other Express packages are available on the NPM package manager.
To use third party middleware you first need to install it into your app using NPM. For example, to install the morgan HTTP request logger middleware, you'd do this:
$ npm install morgan
You could then call use() on the Express application object to add the middleware to the stack:
var express = require('express');
var logger = require('morgan');
var app = express();
app.use(logger('dev'));
...
Note: Middleware and routing functions are called in the order that they are declared. For some middleware the order is important (for example if session middleware depends on cookie middleware, then the cookie handler must be added first). It is almost always the case that middleware is called before setting routes, or your route handlers will not have access to functionality added by your middleware.
You can write your own middleware functions, and you are likely to have to do so (if only to create error handling code). The only difference between a middleware function and a route handler callback is that middleware functions have a third argument next, which middleware functions are expected to call if they are not that which completes the request cycle (when the middleware function is called, this contains the next function that must be called).
You can add a middleware function to the processing chain with either app.use() or app.add(), depending on whether you want to apply the middleware to all responses or to responses with a particular HTTP verb (GET, POST, etc). You specify routes the same in both cases, though the route is optional when calling app.use().
The example below shows how you can add the middleware function using both methods, and with/without a route.
var express = require('express');
var app = express();
// An example middleware function
var a_middleware_function = function(req, res, next) {
// ... perform some operations
next(); // Call next() so Express will call the next middleware function in the chain.
}
// Function added with use() for all routes and verbs
app.use(a_middleware_function);
// Function added with use() for a specific route
app.use('/someroute', a_middleware_function);
// A middleware function added for a specific HTTP verb and route
app.get('/', a_middleware_function);
app.listen(3000);
JavaScript Tip: Above we declare the middleware function separately and then set it as the callback. In our previous route handler function we declared the callback function when it was used. In JavaScript, either approach is valid.
The Express documentation has a lot more excellent documentation about using and writing Express middleware.
You can use the express.static middleware to serve static files, including your images, CSS and JavaScript (static() is the only middleware function that is actually part of Express). For example, you would use the line below to serve images, CSS files, and JavaScript files from a directory named 'public' at the same level as where you call node:
app.use(express.static('public'));
Any files in the public directory are served by adding their filename (relative to the base "public" directory) to the base URL. So for example:
http://localhost:3000/images/dog.jpg
http://localhost:3000/css/style.css
http://localhost:3000/js/app.js
http://localhost:3000/about.html
You can call static() multiple times to serve multiple directories. If a file cannot be found by one middleware function then it will simply be passed on to the subsequent middleware (the order that middleware is called is based on your declaration order).
app.use(express.static('public'));
app.use(express.static('media'));
You can also create a virtual prefix for your static URLs, rather than having the files added to the base URL. For example, here we specify a mount path so that the files are loaded with the prefix "/media":
app.use('/media', express.static('public'));
Now, you can load the files that are in the public directory from the /media path prefix.
http://localhost:3000/media/images/dog.jpg
http://localhost:3000/media/video/cat.mp4
http://localhost:3000/media/cry.mp3
For more information, see Serving static files in Express.
Errors are handled by one or more special middleware functions that have four arguments, instead of the usual three: (err, req, res, next). For example:
app.use(function(err, req, res, next) {
console.error(err.stack);
res.status(500).send('Something broke!');
});
These can return any content required, but must be called after all other app.use() and routes calls so that they are the last middleware in the request handling process!
Express comes with a built-in error handler, which takes care of any remaining errors that might be encountered in the app. This default error-handling middleware function is added at the end of the middleware function stack. If you pass an error to next() and you do not handle it in an error handler, it will be handled by the built-in error handler; the error will be written to the client with the stack trace.
Note: The stack trace is not included in the production environment. To run it in production mode you need to set the the environment variable NODE_ENV to 'production'.
Note: HTTP404 and other "error" status codes are not treated as errors. If you want to handle these, you can add a middleware function to do so. For more information see the FAQ.
For more information see Error handling (Express docs).
Express apps can use any database mechanism supported by Node (Express itself doesn't define any specific additional behaviour/requirements for database management). There are many options, including PostgreSQL, MySQL, Redis, SQLite, MongoDB, etc.
In order to use these you have to first install the database driver using NPM. For example, to install the driver for the popular NoSQL MongoDB you would use the command:
$ npm install mongodb
The database itself can be installed locally or on a cloud server. In your Express code you require the driver, connect to the database, and then perform create, read, update, and delete (CRUD) operations. The example below (from the Express documentation) shows how you can find "mammal" records using MongoDB.
//this works with older versions of mongodb version ~ 2.2.33
var MongoClient = require('mongodb').MongoClient;
MongoClient.connect('mongodb://localhost:27017/animals', function(err, db) {
if (err) throw err;
db.collection('mammals').find().toArray(function (err, result) {
if (err) throw err;
console.log(result);
});
});
//for mongodb version 3.0 and up
let MongoClient = require('mongodb').MongoClient;
MongoClient.connect('mongodb://localhost:27017/animals', function(err, client){
if(err) throw err;
let db = client.db('animals');
db.collection('mammals').find().toArray(function(err, result){
if(err) throw err;
console.log(result);
client.close();
});
}
Another popular approach is to access your database indirectly, via an Object Relational Mapper ("ORM"). In this approach you define your data as "objects" or "models" and the ORM maps these through to the underlying database format. This approach has the benefit that as a developer you can continue to think in terms of JavaScript objects rather than database semantics, and that there is an obvious place to perform validation and checking of incoming data. We'll talk more about databases in a later article.
For more information see Database integration (Express docs).
Template engines (referred to as "view engines" by Express) allow you to specify the structure of an output document in a template, using placeholders for data that will be filled in when a page is generated. Templates are often used to create HTML, but can also create other types of documents. Express has support for a number of template engines, and there is a useful comparison of the more popular engines here: Comparing JavaScript Templating Engines: Jade, Mustache, Dust and More.
In your application settings code you set the template engine to use and the location where Express should look for templates using the 'views' and 'view engines' settings, as shown below (you will also have to install the package containing your template library too!)
var express = require('express');
var app = express();
// Set directory to contain the templates ('views')
app.set('views', path.join(__dirname, 'views'));
// Set view engine to use, in this case 'some_template_engine_name'
app.set('view engine', 'some_template_engine_name');
The appearance of the template will depend on what engine you use. Assuming that you have a template file named "index.<template_extension>" that contains placeholders for data variables named 'title' and "message", you would call Response.render() in a route handler function to create and send the HTML response:
app.get('/', function(req, res) {
res.render('index', { title: 'About dogs', message: 'Dogs rock!' });
});
For more information see Using template engines with Express (Express docs).
Express makes no assumptions in terms of structure or what components you use. Routes, views, static files, and other application-specific logic can live in any number of files with any directory structure. While it is perfectly possible to have the whole Express application in one file, typically it makes sense to split your application into files based on function (e.g. account management, blogs, discussion boards) and architectural problem domain (e.g. model, view or controller if you happen to be using an MVC architecture).
In a later topic we'll use the Express Application Generator, which creates a modular app skeleton that we can easily extend for creating web applications.
Congratulations, you've completed the first step in your Express/Node journey! You should now understand Express and Node's main benefits, and roughly what the main parts of an Express app might look like (routes, middleware, error handling, and template code). You should also understand that with Express being an unopinionated framework, the way you pull these parts together and the libraries that you use are largely up to you!
Of course Express is deliberately a very lightweight web application framework, so much of its benefit and potential comes from third party libraries and features. We'll look at those in more detail in the following articles. In our next article we're going to look at setting up a Node development environment, so that you can start seeing some Express code in action.