Tumblr RB

tumblr.rb

# Put this file inside script/import_scripts
# How to run ? 
# Just get the posts.xml file from the tumblr export that is all.
# bundle exec ruby script/import_scripts/tumblr.rb script/import_scripts/posts.xml 
# 
require "nokogiri"
require File.expand_path(File.dirname(__FILE__) + "/base.rb")

class ImportScripts::Tumblr < ImportScripts::Base

  def initialize
    super

    @tumblr_file =  ARGV[0]

    raise ArgumentError.new('Tumblr file argument missing. Provide full path to tumblr xml file.') if @tumblr_file.blank?

    #Not sure if i want the latest activity.

  end

  def execute

    check_file_exist

    create_tumblr

    parse_file


  end

  private

  def check_file_exist
    raise ArgumentError.new("File does not exist: #{@tumblr_file}") unless File.exist?(@tumblr_file)
  end

  def parse_file

    puts "parsing file...."

    file  = read_file

  end

  def create_tumblr
    puts "Creating Catorgy"

    create_category({
      name: 'Tumblr Blog',
      user_id: -1,
      description: "Articles from the Tumblr blog"
    }, nil) unless Category.find_by_name('Blog')

  end

  def read_file
    puts "reading file..."
    string = Nokogiri::XML(File.read(@tumblr_file))

    posts  = Array.new

    string.xpath("//post").each do |node|

      title = node.xpath('regular-title').text
      url = node.attr('url')
      id = node.attr('id')
      post_date = node.attr('date')
      slug = node.attr('slug')
      post_content = node.xpath('regular-body').text
      topic = node.attr('tumblelog')

      post = {
        id: id,
        user_id: Discourse::SYSTEM_USER_ID,
        raw: post_content,
        created_at: post_date,
        topic_id: 1,
        title: slug
      }

      posts.push(post)
    end

    import_posts(posts)

  end

  def import_posts(posts)

    puts "Importing Posts"

    create_posts(posts) do |post|

      puts "Creating Post"
      {
        id: post[:id],
        user_id: Discourse::SYSTEM_USER_ID,
        category: 'Tumblr Blog',
        raw: post[:raw],
        created_at: post[:post_date],
        title: post[:title]
      }
    end


    end
end

ImportScripts::Tumblr.new.perform