gistfile1.txt

require 'rubygems'
require 'mechanize'

# Generates a CVS list: Postal Code, District ID, District Name
# Use like this:
# $~/postal_codes/A$ ruby /path/to/extract_elections_ca 2>/dev/null >> districts.csv

postal_codes = File.open("postal_codes.txt").read.split("\n")
postal_codes.each do |postcode|
  postcode.sub! ' ', ''
  file_name = "pages/#{postcode}"
  if File.exist?(file_name)
    doc = open(file_name) { |f| Hpricot(f) }
    if !(doc/"#myMessages_lblMessage").empty?
      $stderr.puts "#{postcode},invalid,invalid"
    else
      district_name = (doc/"#lblElectoralDistrict").text
      links = doc.search("//td[@class='Banner']/a").select {|link| link.attributes["href"].match('ED=([0-9]+)') }
      if links.empty?
        $stderr.puts "#{postcode},invalid,invalid"
      else
        district_id = links.first.attributes["href"].match('ED=([0-9]+)')[1]
        $stdout.puts "#{postcode},#{district_id},#{district_name}"
      end
    end
  else
    $stderr.puts "#{postcode},unknown,unknown"
  end
end