gistfile2.rb

parseGov.rb

#!/usr/bin/env ruby
require 'rubygems'
require 'nokogiri'
require 'open-uri'

(1..948067).each do |n|
  url = "http://www.portaltransparencia.gov.br/PortalTransparenciaPesquisaFavorecidoPF.asp?Exercicio=2010&Pagina=#{n}"
  doc = Nokogiri::HTML(open(url))
  doc.xpath('//div[@id="listagem"]/table//tr').each do |c|
    puts "-"
    puts "cpf: #{c.children[0].content}"
    puts "nome: #{c.children[2].content}"
    puts "valor: R$#{c.children[4].content}"
  end
end