Automatically deploying Jekyll sites with PHP

By Daniel Klco Daniel Klco on 

I’ve been hosting a couple sites using GitHub as a repository and Jekyll to create the site. Due to legacy URLs and general use of .htaccess rules, these sites need to be hosted on Apache httpd instead of GitHub pages.

One of the things I miss about GitHub pages is the automatic deployments. Initially, I attempted to fix this with periodic rebuilding, which was frustrating as it either wasted CPU cycles by running too often or ran to infrequently to really be useful. Next, I tried adding hooks on a per site basis, but it seemed inefficient and I didn’t want to have server information is public repositories.

Finally, I settled on creating an application in PHP to update all of the sites via a GitHub Post Receive Hook. For those not familiar, these are integrations GitHub will invoke after code is pushed into GitHub’s repositories. In this case, the receiving end is a simple PHP script which maps the data GitHub posts into a local git repo and builds the Jekyll site for you. This script uses a configuration file to avoid the direct input being used to regenerate the sites, thus avoid potential command injection vulnerabilities.

To get started, you can simply clone the repository into a directory or virtual host served by Apache and create your own config.json. There is a sample config file you can use as a basis, but the structure is pretty basic:

{
  "time_limit": 0,
  "jekyll_path": "/usr/local/bin/jekyll",
  "git_path": "git",
  "projects_root": "/var/scratch",
  "sites": {
	"https://github.com/user/repo":{
      "id": "repo",
      "jekyll_args": "build -d /var/www/html"
    }
  }
}

Each site you want to deploy should have a corresponding site object entry, keyed by the repository url in GitHub.

To learn more, check out the PHP Jekyll Post Receive Hook in GitHub.


Tags

PHP   Jekyll   GitHub  

comments powered by Disqus