open_directory.rb

namespace :open_directory do
  
  # http://www.google.com/Top/
  # http://rdf.dmoz.org/rdf/structure.rdf.u8.gz

  desc "tasks that deal with Google Open Directory importing"
  
  task :import => :environment do
    seeds = []
    Category.delete_all
    
    File.open(RAILS_ROOT + "/assets/structure.rdf.u8").grep(/<Topic r:id="Top\/Business\//) do |ln|
      parts = ln.sub(/<Topic r:id=\"Top\/Business\/([^>]+)">$/, '\1').gsub('_', " ").split('/').collect(&:strip)
      parent_id = nil
      parts.each do |part|
        unless (categ = Category.find(:first, :conditions => {:name => part, :parent_id => parent_id}))
          categ = Category.create!(:name => part, :parent_id => parent_id)
          puts "new category: #{part} - father: #{categ.parent.andand.name}"
          seeds.push "Category.create(:name => \"#{part}\", :parent_id => #{parent_id.nil? ? 'nil' : parent_id})"
        end
        parent_id = categ.id
      end
    end
    
    open(RAILS_ROOT + "/db/seeds.rb","a").write seeds.join("\n")
  
  end
  
end