From Django to GitHub Pages

// under Django GitHub Pages Jekyll Markdown

I finally joined the cool club of hackers who host their blogs on GitHub Pages. I won't rave about the sheer awesomeness of using GitHub, Jekyll, and Markdown to write blogs because there are already about a hundred million other blogs out there doing just that. I will, though, mention a thing or two about migrating my blog from its previous framework, Django.

Migrating Entries

Jekyll's required layout is well-defined, so to get a bunch of database stuff to look like Markdown, I found it was easiest to run them all through a templating script that I wrote for just the purpose.

The template is fairly straight-forward post.html:

{% raw %}

Title: "{{ entry.title }}"
Date: {{ entry.pub_date|date:"Y-m-d H:i:s" }}

{{ entry.content|safe }}

{% endraw %}

The migration script was a little tricky because some of my posts were stored in Markdown already, but some older ones were actually still in raw HTML from way-back-when I used a WYSIWYG editor.

I used an open-source convertor called html2text with minimal issues.

The migration script converts anything written prior to May, 2012 and runs it through html2text.

from django.template import Context, Template
from django.template.defaultfilters import slugify
from django.utils import timezone
import html2text

from blog.models import Entry


posts_dir = "~/Sites/dustinfarris.github.com/_posts"


def run():
    print "Starting..."
    template = Template(open('post.html').read())
    count = 0
    # The day I started using Markdown in Django
    md_started = timezone.get_default_timezone()
    for entry in Entry.objects.all():
        print "Migrating '%s'..." % entry.title
        pub_date = timezone.make_naive(entry.pub_date, default_timezone)
        if pub_date < md_started:
            entry.content = html2text.html2text(entry.content)

        # Open a new file using Jekyll's required naming format
        format = "{dir}/{date}-{slug}.md"
        filename = format.format(
            dir=posts_dir,
            date=entry.pub_date.strftime("%Y-%m-%d"),
            slug=slugify(entry.title))
        f = open(filename, 'w')

        # Render the template and encode in UTF-8 (just in case)
        context = Context({'entry': entry})
        content = template.render(context).encode('utf-8')

        # Done
        f.write(content)
        f.close()
        count += 1

    print "Finished migrating %d entries." % count

That took care of the heavy lifting, and I was left to tidy up the remains as I saw fit.

Bash Script

In case this blogging setup isn't hackerish enough, I wrote a quick bash script to jumpstart a new blog post. In ~/.profile:

# Create a new blog post with boilerplate front-matter
function blog() {
  shopt -s xpg_echo
  cd ~/Sites/dustinfarris.github.com/_posts
  SLUGIFIED="$(echo -n "$1" | sed -e 's/[^[:alnum:]]/-/g' | tr -s '-' | tr A-Z a-z)" ;
  FILENAME="$(echo `date +\%Y-\%m-\%d`-$SLUGIFIED)" ;
  TIMESTAMP="$(echo `date +\%Y-\%m-\%d` `date +\%H:\%M:\%S`)" ;
  echo "---\n\nlayout: default\ntitle: \"$1\"\ndate: $TIMESTAMP\n\n---\n\n" > $FILENAME.md ;
  vi +":set syntax=markdown" + $FILENAME.md ;
}

Now if I can just find an editor that doesn't go berserk on embedded code blocks.

Go Top