Converts Jim Hightower's site from drupal to wordpress without having to actually log into the drupal site. Makes images and wires podcast media types up etc.

migrate.rb

require 'date'
require 'net/http'
require 'uri'
require 'open-uri'
require 'rubypress'
require 'mime'
require 'mime/types'
require 'imgkit'

# goals
#
#    migrate all podcast types from jimhightower.com/node/8517 and greater to dev.hightowerlowdown.org
#    done:
#       in wordpress we want these to show up as podcasts with the correct original date, title and author (jim hightower)
#       we want the podcast actual audio attached 
#    not done:
#       the media itself needs to have summaries and a title
#       the whole thing needs an image
#
# example usage
#
#    ruby migrate.rb 8517

#################################################################################
# Log into Wordpress
#################################################################################

wp = Rubypress::Client.new(:host => "what", 
                           :username => "why", 
                           :password => "wha")

#################################################################################
# Fetch content from old site with parsing
# and write to new site
#################################################################################

def FetchHighTowerPage(wp,urlcode)

  # Get old site data

  uri = URI.parse("http://jimhightower.com/node/"+urlcode);
  response = Net::HTTP.get_response uri

  # parse

  audio = response.body.index('commentary-audio')
  audio = audio + 18 + 9 + 23
  audif = response.body[audio..-1].index('>') 
  audif = audio + audif - 2;
  soundtrackurl = response.body[audio..audif]
  filenameindex = soundtrackurl.index("files") + 6
  filename = soundtrackurl[filenameindex..-1] 
  mimetype = MIME::Types.type_for(filename).first.to_s,

  head = response.body[audio..-1].index('div')
  head = audio + head + 4
  tail = response.body[head..-1].index('script')
  tail = head + tail - 4
  body = response.body[head..tail]

  title1 = response.body.index('title') + 22
  title2 = response.body.index('/title') - 2
  title = response.body[title1..title2]

  time1 = response.body.index('"info">') + 11
  time2 = response.body[time1..-1].index("em") - 3 + time1
  timestr = response.body[time1..time2]
  time = Date.parse(timestr)
  time = time + 1 # drupal is utc

  #summary
  parts = body.scan(/<p>([^<>]*)<\/p/)
  # parts = body.split(/<p(\/)>/)
  summary = "#{parts[0][0]}".strip

  if summary.length < 5
    abort("bad summary generation")
  end

  # print a few details
  puts soundtrackurl
  puts title
  puts time
  puts body 
  puts mimetype
  puts summary

  # sanity tests
  if !soundtrackurl || !title || !time || !body || soundtrackurl.length < 4 || title.length < 4 || body.length < 4
    abort("Things are going poorly for us; post seems to be empty")
  end

  #
  # post the article to the new server
  # doing this first for now and will patch up relationships after
  #

  postid = 0
  if 1
    postid = wp.newPost(
      :blog_id => "0",
      :content => {
        :post_status     => "publish",
        :post_type       => "podcast",
        :post_date       => time,
        :post_content    => body,
        :post_title      => title,
        :post_name       => title,
      }
    ) 
    puts "Published post with id " + postid
  else
    abort("Such sad cannot post")
  end

  #
  # Fetch sound from old server and store locally for a second
  #

  soundsaved = 0
  if 1 
    # get audio and save locally
    # http://jimhightower.com/sites/jimhightower.civicactions.net/files/33-23_f_show.mp3
    puts "Loading sound from " + soundtrackurl
    puts "Saving sound to " + filename
    Net::HTTP.start("jimhightower.com") do |http|
      resp = http.get(soundtrackurl)
      open(filename, "wb") do |file|
          file.write(resp.body)
      end
    end
    soundsaved = 1 # will throw exception
  end

  #
  # post the sound to the new server
  # the post_id is set here to associate the sound with the post - but it is not clear if this is the right thing to do
  #

  soundresults = 0
  if soundsaved 
    # copying soundtrack to wordpress - the spec says you can set the post_id to associate them
    soundresults = wp.uploadFile(:data => {
      :name            => filename,
      :type            => mimetype,
      # :overwrite       => 1, #broken
      :post_id         => postid,
      :bits => XMLRPC::Base64.new(IO.read(filename))
    })
    puts "saved sound to server id:#{soundresults['id']} file:#{soundresults['file']} url:#{soundresults['url']} type:#{soundresults['type']}"
  else
    puts "There appears to be no audio - this is sad making"
    abort("We canna handle this anymore captain")
  end
  soundid = soundresults['id']
  soundfile = soundresults['file']
  soundurl = soundresults['url']
  soundtype = soundresults['type']
  soundduration = "2:00"
  soundfilesizeraw = File.size(filename).to_f
  soundfilesizefix = soundfilesizeraw / 2**20
  soundfilesize = '%.2f' % soundfilesizefix
  soundfilesize = "#{soundfilesize}M" 

  #
  # make a pretty backdrop from scratch by reaching into an html painting engine
  #

  imagefilename = "#{filename}.jpg"
  makeme = "<h1>#{title}</h1>#{timestr}"
  kit = IMGKit.new(makeme, :quality => 80, :width   => 600, :height  => 400, "crop-w" => 600, "crop-h" => 400, "disable-smart-width" => true, "zoom" => 1.0 )
  kit.to_file(imagefilename)
  puts "made a background image file at #{imagefilename}"

  #
  # upload an image
  #

  imageid = 0
  if 1
    imageresults = wp.uploadFile(:data => {
      :name           => imagefilename,
      :type           => 'image/jpeg',
      # :overwrite      => 1, #broken
      :post_id        => postid,
      :bits           => XMLRPC::Base64.new(IO.read(imagefilename)),
    })
    imageid = imageresults['id']
  end

  if imageid == 0
    abort("could not upload image")
  end

  #
  # unused
  # an incantation to associate the thumbnail
  # it is not totally clear if this is the right thing or if it competes with the above or if both are needed
  # http://stackoverflow.com/questions/17722743/attach-image-to-post-in-wordpress-xmlrpc
  #

  if postid == 0 
    attachresults = wp.editPost(
      :blog_id => 0,
      :post_id => imageid,
      :content => {
        :post_type      => "attachment",
        :post_status    => "inherit",
        :post_title     => title,
        :post_name      => title,
        :post_parent    => postid,
        :guid           => soundresults['url'],
        :post_content   => "",
        :post_mime_type => "image/jpeg",
      }
    )
    puts "done attaching the two with result status:"
    puts attachresults
  end

  #
  # revise the post to set podcast fields and the thumbnail
  #

  if postid
    attachresults = wp.editPost(
      :blog_id => 0,
      :post_id => postid,
      :content => {
        :post_thumbnail  => imageid,
        :custom_fields   => [
                              { 'key' => 'enclosure',    'value' => soundresults['url'] },
                              { 'key' => 'audio_file',   'value' => soundresults['url'] },
                              { 'key' => 'duration',     'value' => soundduration },
                              { 'key' => 'filesize',     'value' => soundfilesize },
                              { 'key' => 'filesize_raw', 'value' => soundfilesizeraw },
                              { 'key' => 'date_recorded','value' => time.strftime("%d-%m-%Y") },
                            ],
      }
    )
    puts "updated post"
  end

  #
  # revise the media blob
  #

  if soundid 
    attachresults = wp.editPost(
      :blog_id => 0,
      :post_id => soundid,
      :content => {
        :post_title          => title,
        :post_caption        => title,
        :post_excerpt        => summary,
        :post_description    => summary,
        :post_content        => summary,
      }
    )
    puts "updated the sound description"
  end

  #
  # Revise the image
  #

  if imageresults
    attachresults = wp.editPost(
      :blog_id => 0,
      :post_id => imageid,
      :content => {
        :post_title          => title,
        :post_caption        => title,
        :post_excerpt        => summary,
        :post_description    => summary,
        :post_content        => summary,
      }
    )
    puts "updated the image description"
  end

end

#################################################################################
# GO!
#################################################################################

FetchHighTowerPage(wp,ARGV[0])