A script to import your Wordpress posts on Tumblr

wordpress2tumblr.rb

# Export your Wordpress posts to Tumblr (also allows to delete some.)
# Author: Alexandre Solleiro <[email protected]> 
# How-to:
# 1. Export your Wordpress posts (http://en.blog.wordpress.com/2006/06/12/xml-import-export/)
# 2. Edit the information below to match your settings
# 3. To import, type "ruby wordpress2tumblr.rb --run" in the command line 
# 4. To delete some posts, type "ruby wordpress2tumblr.rb --delete 1,2,3,4" where 1,2,3,4 are post numbers. (Be careful not to use spaces between IDs)

# Edit these
WP_EXPORT_PATH = "./wordpress.2009-09-14.xml"   # path to the Wordpress Export file
TUMBLR_BLOG_URL = "http://yourblog.tumblr.com"  # without a trailing slash!
TUMBLR_EMAIL = "[email protected]"               # the email of your Tumblr account
TUMBLR_PASSWORD = "password"                    # your Tumblr account password
TUMBLR_DRAFT = 1                                # 1 saves your posts as drafts on Tumblr; 0 publishes them

# Dependencies: the following gems
require 'rubygems'
require 'open-uri'
require 'rest_client'
require 'hpricot'
require 'json'
require 'time'

# No need to touch here
TUMBLR_WRITE_API = "http://www.tumblr.com/api/write"
TUMBLR_DELETE_API = "http://www.tumblr.com/api/delete"

# This was borrowed from http://choonkeat.svnrepository.com/svn/rails-plugins/misc/hpricot/rss.rb
# It trims the CDATA[] declaration in WP's RSS output
def trim_cdata(text)
  text.gsub!(/\<\!\[CDATA\[(.*)\]\]\>/m, '\1') if text =~ /<\!\[CDATA\[/ && text =~ /\]\]>/
end

# Post a Wordpress Post to Tumblr
def post_to_tumblr(wp_post)
tumblr_post = RestClient.post(TUMBLR_WRITE_API, 
        :email => TUMBLR_EMAIL,
        :password => TUMBLR_PASSWORD,
        :generator => "imported with wordpress2tumblr.rb",
        :type => "regular",
        :format => "html", # can also be markdown, c.f. Tumblr's API
        :title => wp_post[:title],
        :body => wp_post[:body],
        :date => wp_post[:date],
        :tags => wp_post[:tags],
        :private => TUMBLR_DRAFT # saves as draft; 0 to publish
        )
end

# Delete a Post from Tumblr
def delete_from_tumblr(tumblr_post_id)
  begin
  tumblr_deletion = RestClient.post(TUMBLR_DELETE_API, 
          :email => TUMBLR_EMAIL,
          :password => TUMBLR_PASSWORD,
          :"post-id" => tumblr_post_id
          )
  rescue => e
    puts e
  end
end

# The real action
def all_posts_to_tumblr
  # Parsing the WP eXtended RSS for published posts that we want to export
  doc = Hpricot.parse(File.read(WP_EXPORT_PATH))
  @posts_with_comments = []
  (doc/:item).each do |xml_item|
    # Check if post is supposed to be published - these are the ones we want
    if (xml_item/'wp:status').innerHTML == "publish"
      wp_post = {}
      wp_post[:title] = (xml_item/:title).innerHTML
      wp_post[:date] = Time.parse((xml_item/:pubdate).inner_html).to_s
      wp_post[:body] = trim_cdata((xml_item/'content:encoded').innerHTML)
      # Now we work the tags
      tags = [] 
      (xml_item/:category).each do |category|
        if category.attributes['domain'] == 'tag' and category.attributes['nicename'] != nil
          tags << category.attributes['nicename']
        end
      end
      wp_post[:tags] = tags.join(',')
      # IMPORTANT line, actually publishes to Tumblr. Comment it if you want to dry run.
      tumblr_post = post_to_tumblr(wp_post)
      puts "Tumblr: " + tumblr_post.to_s
    end
  end
end # all_posts_to_tumblr

# Delete the Posts with given IDs from Tumblr
def delete_some_from_tumblr(tumblr_post_ids)
  tumblr_post_ids.each do |tumblr_post|
    puts delete_from_tumblr(tumblr_post.to_s)
  end
end

# Execution
if ARGV[0] == '--run'
  all_posts_to_tumblr
elsif ARGV[0] == '--delete' and ARGV[1] not nil
  delete_some_from_tumblr(ARGV[1])
else
  puts "To import to Tumblr, type 'ruby wordpress2tumblr.rb --run'"
  puts "To delete some posts from Tumblr, type 'ruby wordpress2tumblr.rb --delete 1,2,3,4' where 1,2,3,4 are Tumblr Post IDs - do not use spaces between IDs."
end