Script to import from Wordpress XML to Markdown for Jekyll

.gitignore

*.xml
*.md

Gemfile

source :rubygems
gemspec

import-wp-to-md.gemspec

Gem::Specification.new do |s|
  s.authors = ["John Bellone"]
  s.email = 'jb@thunkbrightly.com'
  s.homepage = 'http://github.com/johnbellone/wp-to-md'

  s.require_paths = %w[lib]
end

import-wp-to-md.rb

require 'nokogiri'
require 'date'

doc = Nokogiri::XML(File.open(ARGV[0])) 
doc.xpath('//item').each do |node|
  pubDate = DateTime.parse(node.xpath('.//pubDate').inner_text)
  title = node.xpath('.//title').inner_text.gsub(/\:/, '')
  description = node.xpath('.//description').inner_text

  description = description.prepend('description: ') if (description.length > 0)

  tags = node.xpath('.//category/text()').to_a.join(',').split(/(\W)/).map(&:capitalize).join
  content = node.xpath('.//content:encoded/text()').inner_text

  unless content.empty?
    filename = pubDate.strftime('%Y-%m-%d-').concat(title.downcase.strip.gsub(/\s/,'-').gsub(/[^\w-]/,'')).concat('.md')
  
    doc = <<-text
---
layout: post
title: #{title}
#{description}
tags: #{tags}
---
#{content}
text
    File.open(filename, 'w+') { |f| f.write(doc) }
  end
end

Obviously very crude right now, there are some issues converting from the Wordpress XML to Markdown in regards to the content blocks containing malformed HTML (or at the very least, HTML). I haven't had time to go through this and figure out all the problems. But luckily its easy to test!