Pega umquetenha.org

pega_umquetenha

#!ruby19
# encoding: utf8
# author: rafael polo
# created_at: 02.mar.2010

require 'open-uri'
require 'timeout'

class PegaUmquetenha

  REGEX_CALENDARIO = /umquetenha.org\/uqt\/\?m=([\d]+)/
  REGEX_POST = /<a href="http:\/\/umquetenha.org\/uqt\/\?p=([\d]+)"/
  REGEX_CAPA = /width: 170px; height: 170px;" src="(.*)" border="0"/
  REGEX_ALBUM = /(.*) - (.*) \(([\d]{4})\)/
  REGEX_DOWNLOAD = /<a href="(.*)" target="blank"><span style=".*"><strong>download<\/strong>/

  def self.pega
    page = load_url("http://umquetenha.org/uqt")
    meses = page.scan(REGEX_CALENDARIO)
    meses.each do |mes|
      page_mes = load_url("http://umquetenha.org/uqt/?m=#{mes}")
      quando = page_mes.scan(/<h2 class="pagetitle">(.*)<\/h2>/)
      puts "\n#{quando}"
      posts = page_mes.scan(REGEX_POST)
      puts "#{posts.size} álbuns"
      posts.each do |post|
        page_album = load_url("http://umquetenha.org/uqt/?p=#{post}")
        album = page_album.scan(/<h2>(.*\([\d]{4}\))<\/h2>/).to_s.gsub("&#8211;", "-").gsub("&#038;", "&")
        capa = page_album.scan(REGEX_CAPA)
        download = page_album.scan(REGEX_DOWNLOAD)
        album_info = album.scan(REGEX_ALBUM)
        if album_info && album_info.first
          album_info = album_info.first
          artista = album_info[0]
          album = album_info[1]
          ano = album_info[2]
          if album!="" && capa!="" && artista!="" && album!="" && ano!=""
            puts "Artista: #{artista}"
            puts "Album: #{album}"
            puts "Ano: #{ano}"
            puts "Download: #{download}"
            puts "Capa: #{capa}"
            puts
          end
        end
      end
    end
  end
  
  def self.load_url(url)
    response=""
    begin
      timeout(8) do
        uri = URI.parse(url)
        response = uri.read if uri
      end
    rescue Exception=>error
      response = false
      puts "Erro: #{error}\n\n"
    end
    response
  end
	
end

PegaUmquetenha.pega