wordpress_parser.rb

# require "~/Desktop/wordpress_parser.rb"
# parser = WordpressParser.new("/Users/nicolasiensen/Desktop/quintal.wordpress.2015-03-03.xml")
# posts = parser.posts
# posts.each_with_index {|post, i| puts i; parser.create_post post }

require 'nokogiri'

class WordpressParser
  def initialize file_path = "quintal.wordpress.2015-03-03.xml"
    @xml = Nokogiri::XML(File.open(file_path))
  end

  def create_post post
    category = Category.find_by(slug: post['categories'].first) || Category.first
    tags = post['tags'].map do |tag|
      Tag.find_or_create_by(name: tag)
    end

    new_post = Post.create(
      title: post['title'],
      published_at: Time.parse(post['published_at']),
      body: post['content'],
      legacy_id: post['post_id'],
      created_at: Time.parse(post['post_date']),
      slug: post['post_name'],
      category_id: category.id,
      tags: tags
    )

    post['comments'].each do |comment|
      if comment[:author_email].present? && comment[:body].present?
        user = User.find_by(email: comment[:author_email]) || user = User.create!(
          email: comment[:author_email],
          name: comment[:author_name],
          ip: comment[:author_ip]
        )

        Comment.create!(
          post_id: new_post.id,
          user_id: user.id,
          created_at: Time.parse(comment[:created_at]),
          body: comment[:body],
          approved: comment[:approved] == '1' ? true : false
        )
      end
    end
  end

  def posts
    @posts ||= posts_node_set.map do |post|
      hash = {}

      hash.store('title', post.search(".//title").children.text)
      hash.store('published_at', post.search(".//pubDate").children.text)
      hash.store('content', post.search(".//content:encoded").children.text)
      hash.store('post_id', post.search(".//wp:post_id").children.text)
      hash.store('post_date', post.search(".//wp:post_date").children.text)
      hash.store('post_name', post.search(".//wp:post_name").children.text)

      categories = post.search(".//category[@domain = 'category']").map do |category|
        category.attribute("nicename").value
      end
      hash.store('categories', categories)

      tags = post.search(".//category[@domain = 'post_tag']").map do |tag|
        tag.attribute("nicename").value
      end
      hash.store('tags', tags)

      comments = post.search(".//wp:comment").map do |comment|
        {
          author_name: comment.search(".//wp:comment_author").text,
          author_email: comment.search(".//wp:comment_author_email").text,
          author_ip: comment.search(".//wp:comment_author_IP").text,
          created_at: comment.search(".//wp:comment_date").text,
          body: comment.search(".//wp:comment_content").text,
          approved: comment.search(".//wp:comment_approved").text
        }
      end
      hash.store('comments', comments)

      hash
    end
  end

  def posts_node_set
    @posts_node_set ||= @xml.xpath("//item[wp:post_type[text() = 'post']]")
  end

  def fix_all_posts_bodies
    Post.all.each do |p|
      body = p.body.remove(/height\=\"\d+\"/)
      body = body.remove(/width\=\"\d+\"/)
      p.update_attribute :body, body
    end
  end
end