libidn + addressable

0001-use-libidn-instead-of-pure-ruby-implementation.patch

From 35ee80699222754ae6dc3a2b933716d215fa7824 Mon Sep 17 00:00:00 2001
From: Ilya Grigorik <ilya@igvita.com>
Date: Thu, 17 Feb 2011 23:42:08 -0500
Subject: [PATCH 1/2] use libidn instead of pure ruby implementation

---
 lib/addressable/uri.rb |   25 ++++++++++++++++++-------
 1 files changed, 18 insertions(+), 7 deletions(-)

diff --git a/lib/addressable/uri.rb b/lib/addressable/uri.rb
index 46ed296..fdb0530 100644
--- a/lib/addressable/uri.rb
+++ b/lib/addressable/uri.rb
@@ -23,7 +23,14 @@
 #++
 
 require "addressable/version"
-require "addressable/idna"
+#require "addressable/idna"
+
+begin
+  require 'rubygems'
+  require 'idn'
+rescue
+  puts "install idn for faster performance"
+end
 
 module Addressable
   ##
@@ -432,7 +439,8 @@ module Addressable
       unencoded = self.unencode_component(component)
       begin
         encoded = self.encode_component(
-          Addressable::IDNA.unicode_normalize_kc(unencoded),
+          IDN::Stringprep.nfkc_normalize(unencoded),
+          #Addressable::IDNA.unicode_normalize_kc(unencoded),
           character_class
         )
       rescue ArgumentError
@@ -530,7 +538,8 @@ module Addressable
         if value != nil
           begin
             components[key] =
-              Addressable::IDNA.unicode_normalize_kc(value.to_str)
+              IDN::Stringprep.nfkc_normalize(value.to_str)
+              #Addressable::IDNA.unicode_normalize_kc(value.to_str)
           rescue ArgumentError
             # Likely a malformed UTF-8 character, skip unicode normalization
             components[key] = value.to_str
@@ -951,9 +960,10 @@ module Addressable
       @normalized_host ||= (begin
         if self.host != nil
           if self.host.strip != ""
-            result = ::Addressable::IDNA.to_ascii(
-              self.class.unencode_component(self.host.strip.downcase)
-            )
+            result = IDN::Idna.toASCII(self.class.unencode_component(self.host.strip.downcase))
+            #result = ::Addressable::IDNA.to_ascii(
+            #  self.class.unencode_component(self.host.strip.downcase)
+            #)
             if result[-1..-1] == "."
               # Trailing dots are unnecessary
               result = result[0...-1]
@@ -1993,7 +2003,8 @@ module Addressable
     # @return [Addressable::URI] A URI suitable for display purposes.
     def display_uri
       display_uri = self.normalize
-      display_uri.host = ::Addressable::IDNA.to_unicode(display_uri.host)
+      #display_uri.host = ::Addressable::IDNA.to_unicode(display_uri.host)
+      display_uri.host = IDN::Idna.toUnicode(display_uri.host) 
       return display_uri
     end
 
-- 
1.7.1

0002-this-is-bizarre.patch

From 389230b72b2a294f6e2e025534c69f358f3e8f67 Mon Sep 17 00:00:00 2001
From: root <root@ip-10-114-167-102.ec2.internal>
Date: Thu, 17 Feb 2011 23:51:24 -0500
Subject: [PATCH 2/2] this is bizarre

---
 lib/addressable/uri.rb |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/lib/addressable/uri.rb b/lib/addressable/uri.rb
index fdb0530..6822626 100644
--- a/lib/addressable/uri.rb
+++ b/lib/addressable/uri.rb
@@ -2105,7 +2105,8 @@ module Addressable
     # @return [TrueClass, FalseClass]
     #   <code>true</code> if the URI is frozen, <code>false</code> otherwise.
     def frozen?
-      self.to_s.frozen?
+      false
+      #self.to_s.frozen?
     end
 
     ##
-- 
1.7.1

notes.txt

Libidn: http://www.gnu.org/software/libidn/manual/libidn.html
Ruby wrapper for libidn: http://idn.rubyforge.org/docs/

Pure-ruby:
 
      user     system      total        real
  9.500000   0.050000   9.550000 ( 10.029430)

LibIDN:

      user     system      total        real
  3.390000   0.020000   3.410000 (  3.527302)

Trace post libidn integration:

  32  10.6%  10.6%       32  10.6% garbage_collector
  29   9.6%  20.1%       29   9.6% Addressable::URI.encode_component
  29   9.6%  29.7%       62  20.5% Addressable::URI.normalize_component
  28   9.2%  38.9%       63  20.8% Addressable::URI#frozen?
  25   8.3%  47.2%       35  11.6% Addressable::URI#to_s
  23   7.6%  54.8%       66  21.8% Addressable::URI#normalized_path
  20   6.6%  61.4%      114  37.6% Addressable::URI#initialize
  11   3.6%  65.0%      109  36.0% Addressable::URI#defer_validation
  11   3.6%  68.6%       59  19.5% Addressable::URI.parse
   9   3.0%  71.6%        9   3.0% Addressable::URI.normalize_path
   8   2.6%  74.3%       10   3.3% Addressable::URI#authority
   7   2.3%  76.6%       26   8.6% Addressable::URI#authority=
   7   2.3%  78.9%        8   2.6% Kernel#gem_original_require
   6   2.0%  80.9%      252  83.2% Integer#times
   5   1.7%  82.5%       12   4.0% Addressable::URI#normalized_authority
   5   1.7%  84.2%        5   1.7% Addressable::URI#validate

Q: URI#frozen? This seems bizarre. Removing this shaves another 10% in performance.