Fetch the Queen's new-years speeches into a SQLite3 database... WHY WOULDN'T YOU?

da_queen.rb

# Fetch the Queen's new-years speeches into a SQLite3 database... WHY WOULD YOU?

require "rubygems"
require "open-uri"
require "hpricot"
require "dm-core"

class Speech
  include DataMapper::Resource
  
  property :id, Serial
  property :year, String
  property :body, Text
end

DataMapper.setup(:default, 
  "sqlite3:///#{Dir.pwd}/data.db")
DataMapper.auto_migrate!

class String
  def utfy(from='ISO-8859-1')
    require "iconv"
    Iconv.conv('utf-8', from, self)
  end
end

index = Hpricot(open("http://kongehuset.dk/publish.php?dogtag=k_dk_aktuelt_taler"))
links = (index / "div#content a").select { |a| a.html =~ /tale/i }

links.each do |tale|
  h = Hpricot(open(tale.get_attribute("href")))
  year, body = [tale.html.utfy,
    (h / "div#content").to_html.chop.utfy.gsub!(/(<[^>]*>)/s,"")]
  Speech.new(:year => year, :body => body).save
end