Gera um YAML com os resultados do Ironman Brasil (2008, 2009 e 2010)

get_resultados_brim.rb

require 'net/http'
require 'nokogiri'

$eventos = { '2008' => '159', '2009' => '280', '2010' => '396' }

unless $eventos.keys.include?(ARGV[0])
  STDERR.puts "Uso: #{$0} <#{$eventos.keys.sort.join('|')}>"
  exit(2)
end

def download_doc(params)
  res = Net::HTTP.post_form(
      URI.parse('http://www.championchipbrasil.com.br/resultados.asp?id_cliente=56'),
      { 'id_evento' => $eventos[ARGV[0]], 'acao_site' => 'consultar' }.merge(params))
  
  unless Net::HTTPSuccess === res
    puts "error (#{params.inspect}): #{res.code}"
    exit(1)
  end
  
  Nokogiri::HTML(res.body)
end

page = 0

while true do
  page += 1
  
  STDERR.puts("Page #{page}")
  
  doc = download_doc('inPage' => page.to_s)
  
  chips = {}
  
  doc.xpath('//a').each do |link|
    if href = link.attributes['href']
      if m = href.value.match(/ver_resultado\((\d*)\)/)
        chips[m[1]] = link.content
      end
    end
  end
  
  break if chips.empty?
  
  chips.each do |chip, nome|
    doc = download_doc('id_chipcode' => chip)
    
    puts "#{chip}:"
    puts "  Nome: '#{nome}'"
    
    table = doc.xpath("//table[tr[1]/td[1]='Evento']").first
    table.xpath("tr").each do |tr|
      tds = tr.xpath("td")
      puts "  #{tds[0].content.strip}: '#{tds[1].content.strip}'"
    end
  end
  
end