A simple script to convert the export file from wordpress to middleman blog style

wordpress2middleman.rb

require "nokogiri"
require "upmark"
require "time"

BLACKLIST = ["pothix", "ilmo francisco m. melo"]

def tags_for element
  element.xpath("category").map do |cat|
    cat.text.downcase if cat["domain"] == "post_tag"
  end.compact
end

def category_for element
  element.xpath("category").find do |cat|
    cat["domain"] == "category"
  end.text.downcase
end

def comments_authors_for element
  comments = element.xpath("wp:comment")
  comments.map do |comment|
    type = comment.at_xpath("wp:comment_type").text
    name = comment.at_xpath("wp:comment_author").text
    name if type != "pingback" && BLACKLIST.all?{|b| b != name.downcase}
  end.compact
end

File.open("pothix.wordpress.2015-05-25.xml") do |file|
  items = Nokogiri::XML(file).xpath("//channel//item")

  items.each do |article|
    com_authors = comments_authors_for article
    tags        = tags_for article
    category    = category_for article

    if com_authors.size > 0
      comments_str = "\n\n\n\n_Old comments by: #{com_authors.join(", ")} | Not available anymore. :(_"
    end

    name     = article.at_xpath("wp:post_name").text.strip
    date_str = article.at_xpath("wp:post_date_gmt").text + " +0000"

    date_utc = Time.parse(date_str).utc
    date_str = date_utc.strftime("%Y-%m-%d")
    filename = "#{date_str}-#{name}.markdown"

    begin
      content = Upmark.convert(article.at_xpath("content:encoded").text)
    rescue => e
      name = article.at_xpath("wp:post_name").text.strip
      puts "error given when parsing '#{name}'. Error: #{e.inspect}"
    end

    title = article.at_xpath("title").text
    post = <<-eos
---
title: "#{title.gsub(/\[(.*)\]/, '\1:')}"
date: "#{date_str}"
tags: "#{tags.join(",")}"
category: "#{category}"
---

#{content}#{comments_str}
eos

    File.open('posts/'+filename, 'w') {|f| f.puts post }
  end
end