crawlea actas del tse

crawler.rb

require 'open-uri'
require 'nokogiri'


for i in 1..16000
 # u = "http://siede.tse.hn/app_dev.php/divulgacionmonitoreo/reporte-acta/#{i}"
 
  u="http://s3.amazonaws.com/actas2013/icr/40/1/%05d104.jpg" % i
 print u
  openu=""
  begin
    openu = open(u)
  rescue
    openu = ""
    print "invalid"
  end
  if openu != ""
    print u
    print "\n"
    doc = Nokogiri::HTML(openu)
  
    # if image_link = doc.at_css(".image-acta")
    #     uri = image_link['src']
    #     File.open(File.basename(uri),'wb'){ |f| f.write(open(uri).read) }
    #   end
    File.open(File.basename(u),'wb'){ |f| f.write(open(openu).read) }
    
    
  end
 
end