Automate Jekyll Production

2 minute read

The main benefit I see of using jekyll is being able to edit a plain markdown file and have it appear live on the Internet in just a few seconds. In order for this to happen, I have to perform the following:

  1. Make an edit
  2. Build the website
  3. Commit and push the changes to gh-pages branch
  4. Copy the contents under _site from the gh-pages folder into the production folder
  5. CD into the production-ready folder
  6. Commit and push the changes to the master branch
  7. Wait 5 seconds for github to publish the changes live on github.io.

I wanted a way to automate this.

I have to perform the above steps since I am using jekyll-maps which is not one of the approved plugins that I can use on github pages. This forces me to NOT have github automatically build my jekyll site. I must therefore upload all of the static HTML content on my master branch and have my source files in a gh-pages branch. Github pages will serve the static content from the master branch live for you for free!

I created two folders, one for production and one for source edits. I did the following after already creating the gh-pages branch previously:

  1. cd ~/Documents/github
  2. mkdir production source
  3. cd production; git pull <my_git_repo>.git; git checkout master
  4. cd ../source; git pull <my_git_repo>.git; git checkout gh-pages

I have already forked a jekyll theme that includes all of the scaffolding. I can CD into the source folder, make an edit, and push my changes. I can see the website live on my localhost by running bundle exec jekyll serve. All of the links are relative to my localhost, so I now need to export JEKYLL_ENV=production so that jekyll will produce links with my http://bauerjj.github.io baseurl.

This is where I encounter a jekyll bug that causes a double slash (//) in place of where a single slash should be (/). For example, links will show up as http://bauerjj.github.io//blog instead of http://bauerjj.github.io/blog. I followed jira 6057 and its proposed solution of using URLFilters, however I was not able to resolve my problem. I created a simple script instead that fixes the slashes in the _site directory by iteratiing through all of the files and performing a replace. My config has this: baseurl : "////". Notice how I put multiple slahes. This was so that the script could look for 5 slashes and do the replace since some valid // occur such as those in an http:// url.

#!/bin/bash
# Small script that replaces \\\\\ with a single \ 
FILES=$(find ./ -type f -name '*')
for file in $FILES; do
   sed -i -e 's/\/\/\/\/\//\//g' $file
done

I then needed a method to monitor the source directory for any changes and then commit, push, and deploy the static content up to github for me. I used gitwatch and forked it so that I could customize it do run my replace.sh script, copy the static contents to my production folder, and then commit/push everything to github. You can see my simple changes to the script on my own fork. Now, just execute ./gitwatch and watch it do its magic whenever you edit a source file!

Please let me know if you have any problems.

Categories:

Updated:

Leave a Comment

Your email address will not be published. Required fields are marked *

Note - Sometimes a comment submission will fail. Simply press the "Submit Comment" button again until a green success box is shown

Loading...