01-export-posts-from-mysql

### SQL
# Use the below query in Sequel Pro => Export to XML
#
# This includes tags, not categories
#
# SELECT p.ID, p.post_date, p.post_content, p.post_title, p.post_name, GROUP_CONCAT(t.name SEPARATOR ', ') AS tags
# FROM wp_posts p
# LEFT JOIN wp_term_relationships tr ON tr.object_id = p.id
# LEFT JOIN wp_term_taxonomy tt ON tr.term_taxonomy_id = tt.term_taxonomy_id
# LEFT JOIN wp_terms t ON tt.term_id = t.term_id
# WHERE post_status = 'publish'
# AND post_parent = 0
# AND post_author = 2
# AND post_type = 'post'
# GROUP BY p.ID;

02-migrate_posts

## Run in 'articles' directory:
#  ruby ~/Desktop/migrate.rb

require 'rubygems'
require 'nokogiri'

doc = Nokogiri::XML(open("/Users/ariejan/Desktop/ajn.xml"))

doc.xpath("//custom/row").each do |row|
  
  title = row.xpath('post_title').text
  
  puts "Processing: #{title}"
  
  author = "Ariejan de Vroom"
  slug = row.xpath('post_name').text
  content = row.xpath('post_content').text
  pub_date = Time.parse(row.xpath('post_date').text)
  tags = row.xpath('tags').text
  
  File.open("#{pub_date.strftime("%Y-%m-%d")}-#{slug}.textile", "w") do |f|
    f.write("---\n")
    f.write("title: \"#{title}\"\n")
    f.write("author: #{author}\n")
    f.write("date: #{pub_date.strftime("%Y/%m/%d")}\n")
    f.write("slug: #{slug}\n")
    f.write("tags: #{tags}\n")
    f.write("---\n")

    f.write(content)
  end
end

03-check_for_images

# Just checks for images and reports them.
# I intend to copy/paste my wp-content/uploads dir entirely to 'public'.

require 'rubygems'
require 'nokogiri'

Dir.glob("/Users/ariejan/dev/ariejannet/articles/*.txt").each do |art|
  doc = Nokogiri::HTML(open(art))
  puts "Processing: #{art}"
  
  doc.css("img").each do |img|
    src = img["src"]
    
    puts " -- #{src}"
    
  end
end