Grabs exposition images from www.palazzostrozzi.org

strozziscrape.rb

require "rubygems"
require "mechanize"


url = "http://www.palazzostrozzi.org/AnagraficaOpere.jsp?idForm=13&idMostra=%u"
browser = Mechanize.new { |agent| agent.user_agent_alias = 'Mac Safari' }

(1..20).each do |id_mostra|
  begin
    page = browser.get(url % id_mostra)
    title = page.at('h1.title').text
    raise "No exposition found at #{id_mostra}!" if title.empty?
    folder = "#{id_mostra}_#{title.gsub(/\W+/, '-').upcase}"
    links = page.search('table[summary="Lista Opere"] a[title="Scarica l\'immagine ad alta risoluzione"]')
    puts "#{title}: found #{links.length} images"

    Dir.mkdir(folder) unless File.directory? folder
    links.each do |link|
      href = link.attr(:href)
      filename = File.basename(href)
      `curl #{href} > #{File.join(folder, filename)}`
    end

  rescue Exception => ex
    puts "Error! #{ex}"
  end
end