utf8uri.rb

# Converts URIs that go beyond the set of ASCII Charset ("Umlautdomains")
# into usable ones for ruby
#
# EXPERIMENTAL - but seems to work pretty good so far
#
module UTF8URI
  # needs punycode4r gem !
  require 'punycode'
  
  def self.encode(url)
    scheme, uri = url.split('//')
    parts = uri.split('/')
    domains = parts.shift.split('.')
    
    parts = parts.collect do |p|
      URI.encode(p)
    end
    
    domains = domains.collect do |d|
      tmp = Punycode.encode(d)
      tmp.gsub(/-$/, '') != d ? "xn--#{tmp}" : d
    end
    
    [scheme, '//', domains.join('.'), '/', parts.join('/')].join
  end
end