First steps with Gulp.js

Front end front end developers spend little time coding. Even if we ignore meetings, much of the job involves basic tasks your working day:

Tasks must be repeated every time you make a change. You begin with good intentions but the infallible front end developer will forget to compress an image or two. Over time, tasks become increasingly arduous and time-consuming; you’ll dread the inevitable content and template changes. It’s mind-numbing, repetitive work. Would it be better to spend your time on more profitable jobs?

If so, you need a task runner.

That Sounds Complicated!

Creating a build process will take time. It’s more complex than performing each task manually but, in long-term, you will save hours of effort, reduce human error and save your sanity. Adopt a pragmatic approach:

• automate the most irritating tasks first
• try not to over-complicate your build process; an hour or two is more than enough for the initial set-up
• choose task runner software and stick with it for a while.

Part of the tools may be new to you but take a deep breath and concentrate on one thing at a time.

Task Runners: Gulp

Many issues were addressed in later editions but Gulp had already arrived and offered a number of improvements:

1. Features such as file watching were built-in.
2. Gulp plug-ins were (mostly) designed to do a single job.
3. Gulp used JavaScript configuration code which was less verbose, easier to read, simpler to modify, and provided better flexibility.
4. Gulp was faster because it uses Node.js streams to pass data through a series of piped plug-ins. Files were only written at the end of the task.

Of course, Gulp itself isn’t perfect and new task runners such as Broccoli.js, Brunch and webpack have also been competing for front end developer attention. More recently, npm itself has been touted as a simpler option. All have their pros and cons, but Gulp remains the favorite and is currently used by more than 40% of web developers.

Gulp needs Node.js but, while Part JavaScript knowledge is beneficial, front end developers from all web programming faiths will find it useful.

This guidedescribes how to use Gulp 3 – the most recent release version at the time of writing. Gulp 4 has been in development for Part time but remains a beta product. It’s possible to use or switch to Gulp 4 but I recommend sticking with version 3 until the final release.

Step 1: Node.js

Node.js can be downloaded for Windows, Mac and Linux from nodejs.org/download/. There are different options for installing from binaries, package managers and docker images – full instructions are available.

Note for Windows users: Node.js and Gulp run on Windows but some plug-ins may not install or run if they depend on native Linux binaries such as image compression libraries. One option for Windows 10 users is the new bash command-line; this solves many issues but is a beta product and could introduce alternative problems.

Once installed, open a command prompt and enter:

node -v

to reveal the version number. You’re about to make heavy use of npm – the Node.js package manager which is used to install modules. Examine its version number:

npm -v

Note for Linux users: Node.js modules can be installed globally so they are available throughout your system. However, most users will not have permission to write to the global directories unless npm commands are prefixed with sudo. There are a number of options to fix npm permissions and tools such as nvm can help but I often change the default directory, e.g. on Ubuntu/Debian-based platforms:

cd ~
mkdir .node_modules_global
npm config set prefix=$HOME/.node_modules_global
npm install npm -g

Then add the following line to the end of ~/.bashrc:

export PATH="$HOME/.node_modules_global/bin:$PATH"

and update with:

source ~/.bashrc

Step 2: Install Gulp Globally

Install Gulp command-line interface globally so the gulp command can be run from any project folder:

npm install gulp-cli -g

Verify Gulp has installed with:

gulp -v

Step 3: Configure Your Bulid

Note for Node.js projects: you can skip this step if you already have a package.jsonconfiguration file.

Presume you have a new or pre-existing project in the folder project1. Navigate to this folder and initialize it with npm:

cd project1
npm init

You will be asked a series of questions – enter a value or hit Return to accept defaults. A package.json file will be created on completion which stores your npm configuration settings.

For the remainder of this article we’ll presume your project folder contains the following sub-folders:

src folder: pre-processed source files

This contains further sub-folders:

• html – HTML source files
• images — the original uncompressed images
• js — multiple pre-processed script files
• scss — multiple pre-processed Sass .scss files

build folder: compiled/processed files

Gulp will create files and create sub-folders as necessary:

• html – compiled static HTML files
• images — compressed images
• js — a single concatenated and minified JavaScript file
• css — a single compiled and minified CSS file

Your project will almost certainly be different but this structure is used for the examples below.

Tip: If you’re on a Unix-based system and you just want to follow along with the tutorial, you can recreate the folder structure with the following command:

mkdir -p src/{html,images,js,scss} build/{html,images,js,css}

Step 3a: Install Gulp

You can now install Gulp in your project folder using the command:

npm install gulp --save-dev

This installs Gulp as a development dependency and the "devDependencies" section of package.json is updated accordingly.

Step 4: Create a Gulpfile.js

Create a new gulpfile.js configuration file in the root of your project folder. Add some basic code to get begined:

// Gulp.js configuration
var
  // modules
  gulp = require('gulp'),

  // development mode?
  devBuild = (process.env.NODE_ENV !== 'production'),

  // folders
  folder = {
    src: 'src/',
    build: 'build/'
  }
;

This references the Gulp module, sets a devBuild variable to true when running in development (or non-production mode) and defines the source and build folder locations.

Step 5: Create new Gulp Tasks

On it’s own, Gulp does nothing. You must:

1. install Gulp plug-ins, and
2. write tasks which utilize those plug-ins to do something useful.

It’s possible to write your own plug-ins but, since almost 3,000 are available, it’s unlikely you’ll ever need to. You can search using Gulp’s own directory at gulpjs.com/plugins/, on npmjs.com, or search “gulp something” to harness the mighty power of Google.

Gulp provides three primary task methods:

• gulp.task – defines a new task with a name, optional array of dependencies and a function.
• gulp.src – sets the folder where source files are located.
• gulp.dest – sets the destination folder where build files will be placed.

Any number of plug-in calls are set with pipe between the .src and .dest.

Image Task

This is best demonstrated with an example so let’s create a basic task which compresses images and copies them to the appropriate build folder. Since this process could take time, we’ll only compress new and modified files. Two plug-ins can help us: gulp-newer and gulp-imagemin. Install them from the command-line:

npm install gulp-newer gulp-imagemin --save-dev

We can now reference both modules the top of gulpfile.js:

// Gulp.js configuration

var
  // modules
  gulp = require('gulp'),
  newer = require('gulp-newer'),
  imagemin = require('gulp-imagemin'),

We can now define the image processing task itself as a function at the end of gulpfile.js:

// image processing
gulp.task('images', function() {
  var out = folder.build + 'images/';
  return gulp.src(folder.src + 'images/**/*')
    .pipe(newer(out))
    .pipe(imagemin({ optimizationLevel: 5 }))
    .pipe(gulp.dest(out));
});

All tasks are syntactically similar. This code:

1. Creates a new task named images.
2. Defines a function with a return value which…
3. Defines an out folder where build files will be located.
4. Sets the Gulp src source folder. The /**/* ensures that images in sub-folders are also processed.
5. Pipes all files to the gulp-newer module. Source files that are newer than corresponding destination files are passed through. Everything else is removed.
6. The remaining new or changed files are piped through gulp-imagemin which sets an optional optimizationLevelargument.
7. The compressed images are output to the Gulp dest folder set by out.

Save gulpfile.js and place a few images in your project’s src/images folder before running the task from the command line:

gulp images

All images are compressed accordingly and you will see output such as:

Using file gulpfile.js
Running 'imagemin'...
Finished 'imagemin' in 5.71 ms
gulp-imagemin: image1.png (saved 48.7 kB)
gulp-imagemin: image2.jpg (saved 36.2 kB)
gulp-imagemin: image3.svg (saved 12.8 kB)

Try running gulpimages again and nothing should happen because no newer images exist.

HTML Task

We can now create a similar task which copies files from the source HTML folder. We can safely minify our HTML code to remove unnecessary whitespace and attributes using the gulp-htmlclean plug-in:

npm install gulp-htmlclean --save-dev

which is then referenced at the top of gulpfile.js:

var
  // modules
  gulp = require('gulp'),
  newer = require('gulp-newer'),
  imagemin = require('gulp-imagemin'),
  htmlclean = require('gulp-htmlclean'),

We can now create an html task at the end of gulpfile.js:

// HTML processing
gulp.task('html', ['images'], function() {
  var
    out = folder.build + 'html/',
    page = gulp.src(folder.src + 'html/**/*')
      .pipe(newer(out));

  // minify production code
  if (!devBuild) {
    page = page.pipe(htmlclean());
  }

  return page.pipe(gulp.dest(out));
});

This reuses gulp-newer and introduces a couple of concepts:

1. The [images] argument states that our images task must be run before processing the HTML (the HTML is likely to reference images). Any number of dependent tasks can be listed in this array and all will complete before the task function runs.
2. We only pipe the HTML through gulp-htmlclean if NODE_ENV is set to production. Therefore, the HTML remains uncompressed during development which may be useful for debugging.

Save gulpfile.js and run gulp html from the command line. Both the html and images tasks will run.

JavaScript Task

Too easy for you? Let’s process all our JavaScript files by building a basic module bundler. It will:

1. ensure dependencies are loaded first using the gulp-deporder plug-in. This analyses comments at the top of each script to ensure correct ordering e.g. // needs: defaults.js lib.js.
2. concatenate all script files into a single main.js file using gulp-concat, and
3. remove all console and debugging statements with gulp-strip-debug and minimize code with gulp-uglify. This step will only occur when running in production mode.

Install the plug-ins:

npm install gulp-deporder gulp-concat gulp-strip-debug gulp-uglify --save-dev

Reference them at the top of gulpfile.js:

var
  ...
  concat = require('gulp-concat'),
  deporder = require('gulp-deporder'),
  stripdebug = require('gulp-strip-debug'),
  uglify = require('gulp-uglify'),

Then add a new js task:

// JavaScript processing
gulp.task('js', function() {

  var jsbuild = gulp.src(folder.src + 'js/**/*')
    .pipe(deporder())
    .pipe(concat('main.js'));

  if (!devBuild) {
    jsbuild = jsbuild
      .pipe(stripdebug())
      .pipe(uglify());
  }

  return jsbuild.pipe(gulp.dest(folder.build + 'js/'));

});

Save then run gulp js to watch the magic happen!

CSS Task

Finally, let’s create a CSS task which compiles Sass .scss files to a single .css file using gulp-sass. This is a Gulp plug-in for node-sass which binds to the super-fast LibSass C/C++ port of the Sass engine (you won’t need to install Ruby). We’ll presume your primary Sass file scss/main.scss is responsible for loading all partials.

Our task will also utilize the fabulous PostCSS via the gulp-postcss plug-in. PostCSS needs its own set of plug-ins and we’ll install:

• postcss-assets to manage assets. This allows us to use properties such as background: resolve('image.png'); to resolve file paths or background: inline('image.png'); to inline data-encoded images.
• autoprefixer to automatically add vendor prefixes to CSS properties.
• css-mqpacker to pack multiple references to the same CSS media query into a single rule.
• cssnano to minify the CSS code when running in production mode.

First, install all the modules:

npm install gulp-sass gulp-postcss postcss-assets autoprefixer css-mqpacker cssnano --save-dev

and reference them at the top of gulpfile.js:

var
  ...
  sass = require('gulp-sass'),
  postcss = require('gulp-postcss'),
  assets = require('postcss-assets'),
  autoprefixer = require('autoprefixer'),
  mqpacker = require('css-mqpacker'),
  cssnano = require('cssnano'),

We can now create a new css task at the end of gulpfile.js. Note the images task is set as a dependency because the postcss-assets plug-in can reference images during the build process. In addition, most plug-ins can be passed arguments – refer to their documentation for more information:

// CSS processing
gulp.task('css', ['images'], function() {

  var postCssOpts = [
  assets({ loadPaths: ['images/'] }),
  autoprefixer({ browsers: ['last 2 versions', '> 2%'] }),
  mqpacker
  ];

  if (!devBuild) {
    postCssOpts.push(cssnano);
  }

  return gulp.src(folder.src + 'scss/main.scss')
    .pipe(sass({
      outputStyle: 'nested',
      imagePath: 'images/',
      precision: 3,
      errLogToConsole: true
    }))
    .pipe(postcss(postCssOpts))
    .pipe(gulp.dest(folder.build + 'css/'));

});

Save the file and run the task from the command line:

gulp css

Step 6: Automate You Gulp Tasks

We’ve been running one task at a time. We can run them all in one command by adding a new run task to gulpfile.js:

// run all tasks
gulp.task('run', ['html', 'css', 'js']);

Save and enter gulp run at the command line to execute all tasks. Note I omitted the images task because it’s already set as a dependency for the html and css tasks.

Is this still too much hard work? Gulp offers another method – gulp.watch – which can monitor your source files and run an appropriate task whenever a file is changed. The method is passed a folder and a list of tasks to execute when a change occurs. Let’s create a new watch task at the end of gulpfile.js:

// watch for changes
gulp.task('watch', function() {

  // image changes
  gulp.watch(folder.src + 'images/**/*', ['images']);

  // html changes
  gulp.watch(folder.src + 'html/**/*', ['html']);

  // javascript changes
  gulp.watch(folder.src + 'js/**/*', ['js']);

  // css changes
  gulp.watch(folder.src + 'scss/**/*', ['css']);

});

Rather than running gulp watch immediately, let’s add a default task:

// default task
gulp.task('default', ['run', 'watch']);

Save gulpfile.js and enter gulp at the command line. Your images, HTML, CSS and JavaScript will all be processed then Gulp will remain active watching for updates and re-running tasks as necessary. Hit Ctrl/Cmd + C to abort monitoring and return to the command line.

Step 7: Yupi!

Other plug-ins you may find useful:

• gulp-load-plugins – load all Gulp plug-in modules without require declarations
• gulp-preprocess – a simple HTML and JavaScript preprocessor
• gulp-less – the Less CSS pre-processor plug-in
• gulp-stylus – the Stylus CSS pre-processor plug-in
• gulp-sequence – run a series of gulp tasks in a specific order
• gulp-plumber – error handling which prevents Gulp stopping on failures
• gulp-size – displays file sizes and savings
• gulp-nodemon – uses nodemon to automatically rebegin Node.js applications when changes occur.
• gulp-util – utility functions including logging and color-coding

One useful method in gulp-util is .noop() which passes data straight through without performing any action. This could be used for cleaner development/production processing code, e.g.

var gutil = require('gulp-util');

// HTML processing
gulp.task('html', ['images'], function() {
  var out = folder.src + 'html/**/*';

  return gulp.src(folder.src + 'html/**/*')
    .pipe(newer(out))
    .pipe(devBuild ? gutil.noop() : htmlclean())
    .pipe(gulp.dest(out));

});

Gulp can also call other Node.js modules – they don’t necessarily need to be plug-ins, e.g.

• browser-sync – automatically reload assets or refresh your browser when changes occur
• del – delete files and folders (perhaps clean your build folder at the begin of every run).

Invest a little time and Gulp could save many hours of development frustration. The advantages:

• plug-ins are plentiful
• configuration using pipes is readable and easy to follow
• gulpfile.js can be adapted and reused in other projects
• your total page weight can be reduced to improve performance
• you can simplify your deployment.

Useful links:

• Gulp home page
• Gulp plug-ins
• npm home page