BLURI - URI canonicalisation for Business Link, as was. Has some BL-specific bits, but otherwise good for querystring-as-hash sorting/deleting. Has own spec. One careful owner. Requires `addressable` gem, because, well, `URI` is a bit broken.

bluri.rb

require 'addressable/uri'

module URI
  ##
  # Extends a hash with query string reordering/deleting capabilities
  module QueryHash
    def ordered_query_string(*args)
      unmentioned_keys = keys.reject { |key| args.include?(key.to_s) || args.include?(key.to_sym) }
      (
        args.uniq.collect { |key| render_value(key, self[key]) }.reject { |i| i.nil? } +
            unmentioned_keys.collect { |key| render_value(key, self[key]) }
      ).join('&')
    end

    def delete_keys(*args)
      args.uniq.collect { |key| delete(key.to_s) }
      to_s
    end

    def [](key)
      item = super key
      item = super(key.to_s) if item.nil? || item.length == 0
      item.class == Array && item.length == 0 ? nil : item
    end

    def to_s
      keys.collect { |key| render_value(key, self[key]) }.join('&')
    end

    private

    def render_value(key, value)
      return nil if value.nil?
      return value.collect { |el| render_value(key, el) }.join('&') if value.kind_of? Array
      "#{key}=#{CGI::escape(value)}"
    end
  end

  ##
  # A URI class with a bit extra for messing about with query strings
  #
  class BLURI < URI::HTTP
    extend Forwardable

    def_delegators :@uri, :scheme, :path, :host, :host=, :query, :to_s

    def initialize(uri_str)
      @uri = ::Addressable::URI.parse(uri_str)
      raise URI::InvalidURIError, "'#{uri_str}' not a valid URI" unless @uri
    end

    def query_hash
      @query_hash ||
          (
          @query_hash = CGI::parse(self.query)
          @query_hash.each_pair { |k, v| @query_hash[k] = v[0] if v.length == 1 }
          @query_hash.extend QueryHash
          )
    end

    def query=(query_str)
      @query_hash = nil
      @uri.query = query_str
    end

    def self.parse(uri_str)
      # Deal with known URI spec breaks - leading/trailing spaces and unencoded entities
      if uri_str.is_a? String
        uri_str = uri_str.strip.gsub(' ', '%20')
        uri_str.gsub!('&', '%26') if uri_str =~ /^mailto:.*&.*/
      end
      BLURI.new(uri_str)
    end

    def has_query?
      %w{http https}.include?(@uri.scheme) && query
    end

    #
    # Reorder the query string according to symbols or string key values
    # passed in in order
    #
    def reorder_query_string!(*args)
      return self unless has_query?
      self.query = query_hash.ordered_query_string(*args)
      self
    end

    def delete_query_keys!(*args)
      return self unless has_query?
      self.query = query_hash.delete_keys(*args)
      self
    end

    def []=(key, value)
      return self unless has_query?
      query_hash[key] = value
      self.query = query_hash.to_s
      self
    end

    def canonicalize!
      # 'page' for contact is useless, as is 'contactUs'
      self.delete_query_keys!(:contactUs, :page) if path.include?('action/findcontactdetail')
      self.host = host.sub! 'online.', 'www.' if path.include?('action/findcontact')
      # r codes are removed
      self.delete_query_keys_matching! { |k| k =~ /^r\./ }
      # itemId and type are served the wrong way round to the canonical url
      self.reorder_query_string!(:itemId, :type)
    end

    def delete_query_keys_matching!(&block)
      return self unless has_query?
      self.query = query_hash.delete_if(&block).to_s
      self
    end
  end
end

module Kernel
  def BLURI(uri_str)
    ::URI::BLURI.parse(uri_str)
  end

  module_function :BLURI
end

bluri_spec.rb

require "spec_helper"
require 'bluri'

describe URI::BLURI do
  ITEM_URI = 'http://www.businesslink.gov.uk/bdotg/action/detail?type=RESOURCES&itemId=1081912559'
  RCODE_ITEM_URI = 'http://www.businesslink.gov.uk/bdotg/action/detail?type=RESOURCES&r.l1=2&r.l2=3&itemId=1081912559'
  EXPECTED_QUERY = 'itemId=1081912559&type=RESOURCES'

  it "should be an HTTP URI" do
    bluri = BLURI("http://some.where.com")
    bluri.should be_a URI::HTTP
  end

  it "should not allow other schemes" do
    lambda { BLURI('ftp://foo').should raise_error(ArgumentError) }
  end

  it "should not allow nil" do
    lambda { BLURI(nil) }.should raise_error(URI::InvalidURIError)
  end

  it "should support scheme" do
    BLURI('http://foo').scheme.should == 'http'
  end
  it "should support host" do
    BLURI('http://foo').host.should == 'foo'
  end
  it "should support path" do
    BLURI('http://foo/a/path').path.should == '/a/path'
  end
  it "should support query" do
    BLURI('http://foo?to=you&you=foo').query.should == 'to=you&you=foo'
  end
  it "should support mailto:someone@somewhere" do
    BLURI('mailto:[email protected]').to_s.should == 'mailto:[email protected]'
  end
  it "should correct unencoded ampersands in mailto" do # http://www.faqs.org/rfcs/rfc2368.html
    BLURI('mailto:fruit&[email protected]').to_s.should == 'mailto:fruit%[email protected]'
  end
  it "should correct trailing spaces" do
    BLURI('http://www.newspapersoc.org.uk ').to_s.should == 'http://www.newspapersoc.org.uk'
  end
  it "should correct leading spaces" do
    BLURI(' http://www.newspapersoc.org.uk').to_s.should == 'http://www.newspapersoc.org.uk'
  end

  describe "Query string parsing" do
    before do
      @bluri = BLURI('http://some.com/a/path?itemid=1&type=RESOURCE')
    end

    it "indexes the query string" do
      @bluri.query_hash['itemid'].should == '1'
    end

    it "allows indexing by symbol" do
      @bluri.query_hash[:itemid].should == '1'
    end

    it "should show nil for absent items" do
      @bluri.query_hash[:eerie_flash].should == nil
    end

    it "indexes the query string" do
      @bluri.query_hash['type'].should == 'RESOURCE'
    end

    it "allows setting of the query" do
      @bluri.query = "furry=really"
      @bluri.to_s.should == 'http://some.com/a/path?furry=really'
    end

    describe "reordering the query string" do
      it "allows sorting of the query string by in-place replacement" do
        @bluri.reorder_query_string!(:type, :itemid, :type)
        @bluri.query.should == 'type=RESOURCE&itemid=1'
      end

      it "should bunch repeated items up" do
        bluri = BLURI('http://foo?itemid=1&type=2&itemid=3').reorder_query_string!(:itemid, :type)
        bluri.query.should == 'itemid=1&itemid=3&type=2'
      end

      it "should leave out items that weren't there" do
        bluri = BLURI('http://foo?itemid=1&type=RESOURCE')
        bluri.reorder_query_string!(:granny_smith, :itemid)
        bluri.to_s.should_not include('granny_smith')
      end

      it "should not add a query string to a URL without one" do
        bluri = BLURI('http://foo')
        bluri.reorder_query_string!(:things)
        bluri.to_s.should == 'http://foo'
      end

      it "should preserve items which weren't mentioned, but at the end" do
        bluri = BLURI('http://foo?q1=1&q2=2&q3=3')
        bluri.reorder_query_string!(:q3, :q1)
        bluri.query.should == 'q3=3&q1=1&q2=2'
      end

      it "should leave unmentioned singles alone" do
        TOPIC_URI = 'http://www.businesslink.gov.uk/bdotg/action/layer?topicId=1074450344'
        bluri = BLURI(TOPIC_URI)
        bluri.reorder_query_string!(:itemid, :type)
        bluri.to_s.should == TOPIC_URI
      end

      it "should handle cased params" do
        bluri = BLURI(ITEM_URI).reorder_query_string!(:itemId, :type)
        bluri.query.should == EXPECTED_QUERY
      end
    end
  end

  describe "Deleting parts" do
    it "should remove specified parts" do
      BLURI('http://foo?q=1&q2=2&q3=3&q4=4').delete_query_keys!(:q2, :q3).
          query.should == 'q=1&q4=4'
    end

    it "should leave parts not present" do
      BLURI('http://foo?q=1').delete_query_keys!(:not_present).query.should == 'q=1'
    end

    it "should allow removal of keys matching a regex" do
      BLURI('http://foo?q=1&r1.l=2&r2.r=3&p=2').delete_query_keys_matching! { |k, v| k =~ /r[0-9]\..+/ }.
          query.should == 'q=1&p=2'
    end
  end

  describe "changing query keys" do
    it "should let us alter parameters" do
      uri = BLURI('http://foo?q=1&r=2')
      uri['q'] = '3'
      uri.to_s.should == 'http://foo?q=3&r=2'
    end
  end


  describe "Canonicalization" do
    it "should reorder item and type and remove r-codes" do
      BLURI(RCODE_ITEM_URI).canonicalize!.query.should include(EXPECTED_QUERY)
    end

    it "should order page query elements" do
      BLURI('http://online.businesslink.gov.uk/bdotg/action/findcontactbrowse?letter=K&page=1&topicId=1074537159').
          canonicalize!.query.should == 'letter=K&page=1&topicId=1074537159'
    end

    describe "Contact URLs" do
      subject do
        BLURI('https://online.businesslink.gov.uk/bdotg/action/findcontactdetail?itemId=1074044196&type=CONTACT&contactUs=&page=1').
            canonicalize!
      end

      it "should remove contactUs and page for contacts" do
        subject.query.should == 'itemId=1074044196&type=CONTACT'
      end

      it "should replace online. with www. to line up with canonicals" do
        subject.host.should == 'www.businesslink.gov.uk'
      end

      it "should work for contactbrowse" do
        BLURI('http://online.businesslink.gov.uk/bdotg/action/findcontactbrowse?page=1&r.l1=1073909872&topicId=1074537116').
            canonicalize!.to_s.should == 'http://www.businesslink.gov.uk/bdotg/action/findcontactbrowse?page=1&topicId=1074537116'
      end
    end

  end
end