Goodbye Wordpress

After using Wordpress as my blogging engine of choice for over a year (after moving away from S9Y), I decided to switch this site to Octopress. Why? Because it’s cool! No, seriously. I was actually quite happy with WP, it just always felt like way more than I actually needed. OP on the other hand is super lightweight, ritten in Ruby and uses some of my favorite technologies, like Sinatra and Markdown. It doesn’t stop there though: having the site rendered to static files means I can get rid of MySQL and PHP on my little VPS, which definitely is a plus. And it’s also very cool to have all your posts in a directory on your local file system, so you can use all your favorite command line tools to search through or batch modify them. Last but not least I’m now writing my posts in Emacs instead of some crappy web input form (ok, granted, I could have done that before too).

Now to the actual migration. I started by exporting my WP blog to XML. This can be done in the administration backend under “Tools - Export”. Once I had this XML, I used the following script to convert the posts to individual files for Octopress:

require 'fileutils' require 'date' require 'yaml' require 'rexml/document' include REXML doc = Document.new(File.new(ARGV[0])) FileUtils.mkdir_p "_posts" doc.elements.each("rss/channel/item[wp:status = 'publish' and wp:post_type = 'post']") do |e| p e.elements['wp:post_name'].text post = e.elements slug = post['wp:post_name'].text date = DateTime.parse(post['wp:post_date'].text) name = "%02d-%02d-%02d-%s.textile" % [date.year, date.month, date.day, slug] date_string = "#{date.year}-#{date.month}-#{date.day}" # convert all tags and categories into categories categories = [] post.each('category') do |cat| categories << cat.text end content = post['content:encoded'].text.encode("UTF-8") # convert <code></code> blocks to {% codeblock %}{% encodebloc %} #content = content.gsub(/<code>(.*?)<\/code>/, '`\1`') content = content.gsub(/<code>/, '{% codeblock %}') content = content.gsub(/<\/code>/, '{% endcodeblock %}') # convert <pre></pre> blocks to {% codeblock %}{% encodebloc %} #content = content.gsub(/<pre lang="([^"]*)">(.*?)<\/pre>/m, '`\1`') content = content.gsub(/<pre>/, '{% codeblock %}') content = content.gsub(/<pre lang="([^"]*)">/, '{% codeblock %}') content = content.gsub(/<\/pre>/m, '{% endcodeblock %}') # convert headers (1..3).each do |i| content = content.gsub(/<h#{i}>([^<]*)<\/h#{i}>/, ('#'*i) + ' \1') end puts "Converting: #{name}" data = { 'layout' => 'post', 'title' => post['title'].text, 'date' => date_string, 'comments' => true, 'categories' => categories, }.delete_if { |k,v| v.nil? || v == ''}.to_yaml File.open("_posts/#{name}", "w") do |f| f.puts data f.puts "---" f.puts content end end

This was a fork of another script, but tailored for my specific needs (e.g. all tags and categories of a post were converted to categories and I exported the files to Textile instead of Markdown because I used the former in my WP install through a plugin).

Since some of my posts are quite popular and get linked to a lot, I didn’t want to loose my old links. I therefore wrote another script that created 301 redirects for all of them:

require 'date' require 'rexml/document' include REXML doc = Document.new(File.new(ARGV[0])) format = ARGV[1] || "apache" formats = { :apache => "RewriteRule ^%s$ /%s [R=301,L]", :nginx => "rewrite ^%s /%s permanent;" } File.open("redirects.txt", "w") do |f| doc.elements.each("rss/channel/item[wp:status = 'publish' and wp:post_type = 'post']") do |e| post = e.elements date = DateTime.parse(post['wp:post_date'].text) slug = post['wp:post_name'].text post['link'].text =~ %r{.*(/archives/\d+)} old_url = $1 new_url = "blog/#{date.strftime("%Y/%m/%d")}/#{slug}/" f.puts formats[format.intern] % [old_url, new_url] end end

The script can generate the necessary directives for Apache or Nginx, so all you have to do is copy them into your web server configuration. I also added another redirect rule for the RSS feed to not annoy my subscribers:

RewriteRule ^/feed /atom.xml [R=301,L]

Of course I also didn’t want to loose all my comments, so I had to come up with a solution for that. I first installed the WP Disqus plugin and exported all my current comments. I then used some good old Emacs magic to convert the list of redirects I had already generated to a CSV for Disqus’ URL map migration tool. So far this has not migrated all comments to the correct location, but it should eventually work out (worst case I’ll have to email Disqus support).

Last but not least I used Octopress’ excellent customization suport to re-add some stuff to my headers, like the OpenID delegate.

Of course there are still some things that need to be done, some of them more urgently then others. First off this site needs a better stylesheet. The default is actually quite ok, but I want something with more personality. I’ll also have to go through my old posts and clean them up a bit, the automatic conversion did some funny things at times. I also want to re-integrate Flattr and PostRank, but I’m not exactly in a hurry when it comes to that.

I hope this post was useful to some of you, I’m definitely happy with my new blogging setup.

citizen428.blog()

Try to learn something about everything

Comments