< Back to Blog

Improving Your Documentation and Communication by Bringing Together Swagger and Node.js

Node.js and Swagger

Let's connect

Thank you! We'll get in touch soon

What is any developer’s favorite activity? Coding, of course. However, if a developer does nothing but write code, the situation starts looking pretty sad. In order to achieve optimal results, software engineers need to communicate with each other. In the software development process, documentation is one of the best ways you can support such communication.

One of the best examples of documentation — improving the lives of frontend and backend developers as well as testers — comes from Swagger. The platform’s features include descriptions of input and output models, request parameters, the ability to change environments, and implication and expansion of models. However, as Benjen Stark once said, “You know, my brother once told me that nothing someone says before the word ‘BUT’ really counts.” In this context, our “BUT” is the combination of Swagger and Node.js. Swagger and Node.js exist as separate entities, and — in our non-magic-enabled world — combining them isn’t as smooth a process as developers would like.

There are several different ways to approach the integration of Swagger and Node.js.

Using the Swagger-node Module

The first method is to use the swagger-node module. This module is really easy to work with:

$> npm install -g swagger
$> swagger project create hello-world
$> swagger project edit
… editing controllers, adding descriptions …
$> swagger project start

However, this method has one significant disadvantage. In non-experimental, real-life situations, information required to create documentation (input and output parameters and filters) must be described by the controller, making its code incredibly heavy, hard to read, and difficult to understand. Accordingly, this isn’t the best method for a serious project.

Using Swagger UI

The second method — a method more applicable to large-scale projects — is Swagger UI. It can be used as a module of an existing application, or launched in a separate Docker container:

$> npm install -g swagger-ui swagger-dist
// app.js
const express = require('express')
const pathToSwaggerUi = require('swagger-ui').absolutePath()
const app = express()

Creating the Docker Container

To create the Docker container, you can use a readymade image, launching the container and indicating the path to the documentation file. I recommend using a different approach, however — one which allows more precise adjusting and tuning:

  1. Wrap Swagger UI and NGINX into a docker-container.

  2. Indicate the endpoint for the app in order to create proper requests.

  3. Add a link to the documentation file.

At this point, you understand that you need to have a united configuration file. However, working with a giant file that contains the entire project’s documentation is a highly uncomfortable experience. The only way to work with any project bigger than “Hello, World!” is to divide everything into modules.

Dividing Documentation into Modules

There are two ways to divide documentation into modules: You can use third-party libraries, or you can solve the problem on your own. The most popular library makes it possible to use the internal reference tool

$ref: ‘../path/to/some/file’

, which is obviously comfortable. However, this approach limits you with the links to the description of the models within the same file.

You may think that other proprietary solutions will be less sophisticated, but let’s not jump at conclusions. Swagger documentation consists of several big sections described in JSON format:


Collecting and Generating Documentation

After looking at this kind of representation, you may conclude that one of the most convenient ways to collect documentation from files is via simple concatenation. In this approach, you describe each model and path in the file.
The documentation generator can be also divided into several logical parts. The first one is preparatory: You are going to connect all required libraries and a “header” for Swagger documentation.

const _ = require(’lodash’);
const fs = require(’fs’);
const yaml = require(’js-yaml’);
const header = {
	swagger: ’2.0′,
	info: {
	version: ’1.0.0′,
	title: ’Some Title’,
	description: ’Some funny description’
	schemes: [’http’, ’https’],
	basePath: ’/api’

The next important part is a function that will open the path sent as an argument, read the list of the files, process them, and combine them in the final documentation file.

/* object */ function readDocs(dir /* string */) {
    const files = fs.readdirSync(dir);
    const docs = {};

    files.forEach(file => {
        if (! /\.yaml$/.test(file)) {
            /* do nothing */
        const fullpath = path.join(dir, file)

        const data = yaml.load(fs.readFileSync(fullpath, 'utf-8').toString());
        _.merge(docs, data);

    return docs;

The addGeneratedFields function (shown above) adds all the announced documentation endpoints along with their required fields, such as the Authorization header and variables to access various elements of the collections. In the long term, you will save a lot of time and code strings because you won’t need to copy many parts, and all such parameters will be described on a consistent basis. Furthermore, the process of adding new required parameters will become much easier and faster.

/* void */ function addGeneratedFields(data /* object */) {
    const authParam = { 
      in: 'header',
      name: 'Authorization',
      description: 'Authorization token using bearer schema',
      required: true, type: 'string'
    const pathRequiredParam = (name /* string */) => ({
      in: 'path',
      name: name,
      required: true,
      type: 'string',
      description: 'Path parameter'

    if (! ('paths' in data)) {

    for (const path in data.paths) {
        for (const method in data.paths[path]) {
            const parameters = data.paths[path][method].parameters || [];

            const pathParams = path.match(/{([\w\s\|]+)}/gi);
            if (pathParams) {
                /* path params exists */
                for (const pathParam of pathParams.reverse()) {
                    const paramName = pathParam.replace('{', '').replace('}', '')

            data.paths[path][method].parameters = parameters;

Saving Your Documentation to a File

The last required part is to save everything into a file. To do that, it’s recommended to use the fs.createWriteStream tool and stream.write from Node.js kernel.

module.exports = function GenerateDocs(dir /* string */, outFile /* string */) {
	const stream = fs.createWriteStream(outFile, { encoding: ’utf-8′, flags: ’w’ });

	stream.once(’open’, fd => {
	stream.write(’# This file is generated\n’);

Fewer than 100 strings are needed to make your documentation modular. Undoubtedly, they can be improved and expanded by:

  • Adding segmentation by subdirectories

  • Adding the name of the file to the model name in case the documentation is fairly voluminous and you are afraid of potential name crosscutting

  • Implementing Swagger’s $allOf, $oneOf, or $anyOf model inheritance mechanisms

  • Using limit / offset / page for list routes

Big-picture Benefits That Are Worth the Effort

After looking at the final result, you may be tempted to think, “Hey, aren’t we trying to reinvent the wheel here?” You are right and wrong at the same time. Self-written pieces of code can help make the process of team documentation writing more precise (and the process can be totally different for you). You can call it a necessary evil that makes the process of documentation writing easier and more convenient. In addition, later, your “reinvented wheel” may become a flexible library ready to be used across the board.

Swagger is a great tool that improves communication across the software development team. However, it’s important to remember that documentation is a much bigger term that includes not only Swagger but also descriptions of processes, preservation of agreements, and step-by-step instructions. Writing good documentation is as difficult as writing a good code, BUT the final result is absolutely worth it. And as Benjen Stark would remind us, nothing that comes before the “but” in that sentence really matters.

Want to learn more about how Distillery’s developers find solutions that help us streamline the product development process? Let us know!