site_url_helper.rb

# -*- coding: utf-8 -*-
require 'uri'
require 'addressable/uri'

module SiteUrlHelper
  
  def self.site_url_valid? url
    uri = URI.parse url
    uri.kind_of? URI::HTTP
  rescue URI::InvalidURIError
    false
  end

  def self.get_printed_and_normalised url
    uri = Addressable::URI.parse url
    
    uri = Addressable::URI.parse "http://#{url}" unless uri.scheme
    
    return nil unless self.site_url_valid? uri.normalized_site
    
    [uri.hostname, uri.normalized_site]
  end

end

2.1.2 :169 > SiteUrlHelper.get_printed_and_normalised "президент.рф"
 => ["президент.рф", "http://xn--d1abbgf6aiiy.xn--p1ai"]
2.1.2 :170 > SiteUrlHelper.get_printed_and_normalised "http://президент.рф"
 => ["президент.рф", "http://xn--d1abbgf6aiiy.xn--p1ai"]
2.1.2 :171 > SiteUrlHelper.get_printed_and_normalised "http://test.com"
 => ["test.com", "http://test.com"]
2.1.2 :172 > SiteUrlHelper.get_printed_and_normalised "http://www.test.com"
 => ["www.test.com", "http://www.test.com"]
2.1.2 :173 > SiteUrlHelper.get_printed_and_normalised "test.com"
 => ["test.com", "http://test.com"]
2.1.2 :174 > SiteUrlHelper.get_printed_and_normalised "goo.gl"
 => ["goo.gl", "http://goo.gl"]
2.1.2 :175 > SiteUrlHelper.get_printed_and_normalised "sibext.ru"
 => ["sibext.ru", "http://sibext.ru"]
2.1.2 :176 > SiteUrlHelper.get_printed_and_normalised "bad!"
 => nil