As stated before in my last article New Site this site is written in Jekyll. It contains five pages:

Jekyll is a static site generator. I keep all content in simple text files and jekyll then reads each file and transforms it to HTML. Each bit of content is a file composed of two parts: metadata stored in YAML and the content which is written in HTML.

Sometimes I make mistakes, and missing or invalid metadata can cause layout quirks or unexpected errors. To catch these problems earlier, I decided to treat this site like I would any other code by testing it. Since this site is basically one big blog anyway I now automated unit tests after every push.

An example YAML section of a blog post looks like this:

type: post
tags: testing, newsite, jekyll, codeship, ci 
published: true
title: I unit test my blog 
layout: post
is_featured: false
image-large: "/img/blog/header/unit_test.jpg"
image-small: "/img/blog/unit_test.jpg"

Basically the unit tests just parse the yaml and look for obvious mistakes. These include include 'type', 'layout', 'published', 'status', 'image-large' and 'image-small'. If any of these are missing the tests will fail.

Something that is new for the new site are images on each blog article. A few times I have written something and for some reason my gulp command to take an image and resize it to the appropriate dimensions and save the various files to the correct place would fail. I wanted to ensure that it wouldn't happen again, so I wrote a test:

    it('ensure that the header and list graphics exist', function(){
      var large_filepath = path.resolve(__dirname) + "/.." + metadata['image-large'];
      var small_filepath = path.resolve(__dirname) + "/.." + metadata['image-small'];

No doubt there is a smarter way to do this but this is my own site and I like to tinker with things.

Another thing I wanted to do was check the internal links on the articles. I've had a site in some form since 2000 so most links now just don't simply work so I only really care about links from 2012 onwards. I removed the blog/yyyy/mm/dd/title URI to simply be /title. It made more sense. So I wrote a test to ensure that any old links to the blog/yyyy/mm/dd/title URI would break. I could have easily configured nginx to remap these. However if I move back to Github Pages then those links would start to fail again.

  it('all blog urls should no longer have the date inside them', function(){
    var urls = file.match(/\/blog\/\d{4}\/\d{2}\/\d{2}\//g);
    assert.ok(!urls, filename + ' has some articles with old an old blog uri');

Again there is probably a smarter way but this works for me. Feel free to issue a pull request if you want to though :)

Before every push I run my test suite.

Then on a push CodeShip runs the tests. If all is good jekyll builds the site and then rsyncs the files over to the server.

All in all it does what I want. I would like to add more tests - such as internal images should always exist, the number of featured work articles is always at two etc. Time will grow it I guess.

The source of the test file can be viewed here.