mp_csv.rb

# Get a list of all MPs as a CSV file

require 'rubygems'
require 'hpricot'
require 'open-uri'

doc = Hpricot(open('http://www.parliament.uk/mpslordsandoffices/mps_and_lords/alms.cfm').read)

mps = ((doc / "table").to_a[4] / "tr").to_a
mps.shift

line_format = "%s,%s,%s,%s,%s,%s"
puts line_format % %W( Name Party Constituency Email Website Biography )
mps.each do |mp|
  name, consituency = (mp / "td text()").to_a
  next if consituency.nil?
  name, party = name.to_s.scan(/(.*) \((.*)\)/)[0]
  options = {}
  (mp / "td a").to_a.each{ |a|
    case a.inner_html.to_s.strip
    when /website/i
      options[:website] = a['href']
    when /bio/i
      options[:bio] = a['href']
    when /email/i
      options[:email] = a['href']
    end
  }
  puts line_format % [ name, party.upcase, consituency, *options.values_at(:email, :website, :bio) ].map{|a| '"' + a.to_s.strip + '"'}
end