WordpressXML to Jekyll.rb

#!/usr/bin/env ruby
require 'nokogiri'
require 'yaml'

class File
  def writeline(value)
    self.write(value + "\n")
  end
end

inputFile = ARGV[0]

if inputFile then
File.new('_config.yml','w+')
unless(File.exists?('_posts')) then Dir.mkdir('_posts') end
unless(File.exists?('_layouts')) then Dir.mkdir('_layouts') end
unless(File.exists?('css')) then Dir.mkdir('css') end
unless(File.exists?('images')) then Dir.mkdir('images') end

doc = Nokogiri::XML(File.open(inputFile))
doc.xpath('//channel/item').each do |post|
  post_date = Date.parse(post.xpath('wp:post_date/text()').to_s)
  title = post.xpath("title").inner_text
  curr = File.new("_posts/#{post_date.year}-#{post_date.month}-#{post_date.day}-#{title.gsub(/[^\s0-9a-zA-Z]/,'').gsub(/[\s]+/,'-')}.html", 'w+')
  categories = (post.xpath('category[not(@domain)]').map do |cat| cat.inner_text end).to_a
  hash = {"layout"=>"post","title"=> title, "tags"=>categories, "postdate"=>post.xpath('wp:post_date').inner_text}
  status = post.xpath('wp:status').inner_text
  if status != 'publish' then
    hash["status"] = status
  end
  curr.write hash.to_yaml + "---\n"
  curr.write(post.xpath("content:encoded").inner_html)
  curr.close
end
else
  p 'This script requires one argument, the path to your WordPress XML export file.'
end