Grunt Tricks: Part One

October 16th 2013

Here are a few solutions to common problems you may encounter while using Grunt.

Configuring Tasks Dynamically

A good Gruntfile is declarative but sometimes explicitly declaring your config can be cumbersome or impossible.

For example we have some tests:

| - tests/
| --- apple_test.js
| --- peach_test.js
| --- tomato_test.js

I could configure my Gruntfile to run all these tests with:

grunt.initConfig({
  nodeunit: {
    tests: ['tests/*_test.js'],
  },
});

Now I would like to run an individual test, I could simply modify my src to ['tests/apple_test.js'] but that is lame. I could create a target for each test:

grunt.initConfig({
  nodeunit: {
    // This is lame
    apple: ['tests/apple_test.js'],
    peach: ['tests/peach_test.js'],
    tomato: ['tests/tomato_test.js'],
  },
});

Now I can run grunt nodeunit:apple to run individual tests but this way is also lame.
Lame, lame, lame.

Let's use a dynamic alias task instead:

grunt.initConfig({
  nodeunit: {
    all: ['tests/*_test.js'],
  },
});
grunt.registerTask('test', function(file) {
  if (file) grunt.config('nodeunit.all', 'tests/' + file + '_test.js');
  grunt.task.run('nodeunit');
});

Now by default if we run grunt test it will run all our tests. We can now run individual tests with grunt test:apple or grunt test:peach.

Let's extend this task and make it run any number of test groups:

grunt.initConfig({
  nodeunit: {
    all: ['tests/*_test.js'],
  },
});
grunt.registerTask('test', function() {
  var tests = Array.prototype.slice.call(arguments, 0).map(function(test) {
    return 'tests/' + test + '_test.js';
  });
  if (tests.length > 0) grunt.config('nodeunit.all', tests);
  grunt.task.run('nodeunit');
});

Now we can do grunt test:apple:tomato to only run the apple and tomato tests.

Pretty neat. Just be sure to use when necessary. Make sure your config stands on it's own without the alias task and avoid implementing too much logic when dynamically setting your config. Gruntfiles should be declarative and people should be able to quickly understand how your Gruntfile is configured without tracing through too many function calls.

Customize File Input: filter

Grunt accommodates a wide variety of ways to match files through glob and minimatch.

But sometimes you need to further select files or programmatically filter files. This is where the filter property of Grunt becomes useful.

Say we have the following file tree:

| - lib/
| --- tacos.js
| --- burritos.js
| --- tamales.js

When making our delicious Mexican food app we hope that Douglas Crockford will use it. So let's make sure we don't forget any semicolons and lint our code with the more friendly derivative of Crockford's library, JSHint.

Our Gruntfile is configured as such to lint our files:

grunt.initConfig({
  jshint: {
    all: ['lib/*.js'],
  },
});

Now each time we run grunt jshint it will lint all of our js files within lib/. As the items on our menu increases, linting all of the files every time may get slower and slower.

We are really only interested in linting the files we've edited in the last day. Say hello to the filter property. It accepts a string of a known fs.Stats method name or more useful a anonymous function that returns true or false whether to feed the file to the task:

grunt.initConfig({
  jshint: {
    all: {
      src: ['lib/*.js'],
      filter: function(filepath) {
        // When the file was modified
        var filemod = (require('fs').statSync(filepath)).mtime;

        // One day ago
        var dayago = (new Date()).setDate((new Date()).getDate()-1);

        // If the file was modified in the last day, give to the task
        // otherwise filter it out
        return (filemod > dayago);
      },
    },
  },
});

You can get more advanced with this by saving the time you last ran your task and comparing to that date. That way you'll only get the files you've recently edited rather than in the last day. There are other methods to handle this as well.

I prefer the above approach as when you're working on a team that is simultaneously running a Gruntfile. Having the last modified interval set statically will work the same with multiple people editing at the same time (albeit sometimes a little slower as it might match files you're not entirely interested in).

Always remember Gruntfiles are javascript! Utilize these functional helpers often and try to name them more semantically for re-usability across your tasks:

function lastModified(minutes) {
  return function(filepath) {
    var filemod = (require('fs').statSync(filepath)).mtime;
    var timeago = (new Date()).setDate((new Date()).getMinutes() - minutes);
    return (filemod > timeago);
  }
}
grunt.initConfig({
  jshint: {
    all: {
      src: ['lib/*.js'],
      filter: lastModified(24 * 60) // one day ago
    },
  },
});

Customize File Output: rename

Where the filter property is useful for modifying the file input, rename is useful for modifying the file output or the destination of the your files.

Let's say you have a bunch of components you wish to copy. Your file tree looks like such:

| - components/
| --- penne/
| ----- js/
| ------- index.js
| --- gnocchi/
| ----- js/
| ------- index.js
| --- malfatti/
| ----- js/
| ------- main.js
| - dist/
| --- js/

and you would like to copy each of the different component javascript files individually to your dist/js/ folder for packaging:

grunt.initConfig({
  copy: {
    components: {
      expand: true,
      flatten: true,
      cwd: 'components',
      src: ['*/js/*.js'],
      dest: 'dist/js',
    },
  },
});

Which will copy every javascript file within each js/ folder of each component into the dist/js/ folder.

We have a problem though. Some of the file names conflict and will overwrite each other.

rename to the rescue!

grunt.initConfig({
  copy: {
    components: {
      expand: true,
      //flatten: true, // dont need this now as we'll flatten using rename
      cwd: 'components',
      src: ['*/js/*.js'],
      dest: 'dist/js',
      rename: function(dest, src) {
        var path = require('path');

        // Get the name of the component folder (or first folder in src path)
        var component = src.split(path.sep).slice(0, 1)[0];

        // Prefix each javascript file with the
        // component folder name into the destination
        return path.join(dest, component + '_' + path.basename(src));
      },
    },
  },
});

With this our dist/js/ folder will look like:

| - dist/
| --- js/
| ----- penne_index.js
| ----- gnocchi_index.js
| ----- malfatti_main.js

Conclusion

The "Building the files object dynamically" section of the "Configuring Tasks" page in the Grunt docs is by far the most useful section of the docs. I highly encourage reading it and keeping it bookmarked as reference: http://gruntjs.com/configuring-tasks#building-the-files-object-dynamically.

Stay tuned for more Grunt tricks!