From 34b63a8375985486c2b4deeac2fced2c20e482d5 Mon Sep 17 00:00:00 2001
From: Jacques Distler <distler@golem.ph.utexas.edu>
Date: Tue, 1 Dec 2009 12:03:15 -0600
Subject: [PATCH] Fix a Ruby 1.9 Character Encoding Bug

Wow, this stuff is complicated!
Some things really want to be UTF-8;
others really want to be byte strings.
---
 app/controllers/wiki_controller.rb      |  2 +-
 lib/stringsupport.rb                    |  6 +++---
 test/functional/wiki_controller_test.rb | 12 ++++++++++++
 3 files changed, 16 insertions(+), 4 deletions(-)
 mode change 100755 => 100644 test/functional/wiki_controller_test.rb

diff --git a/app/controllers/wiki_controller.rb b/app/controllers/wiki_controller.rb
index 2e41b6ce..23b1ec7e 100644
--- a/app/controllers/wiki_controller.rb
+++ b/app/controllers/wiki_controller.rb
@@ -270,7 +270,7 @@ class WikiController < ApplicationController
       prev_content = ''
       filter_spam(the_content)
       raise Instiki::ValidationError.new('Your name cannot contain a "."') if author_name.include? '.'
-      cookies['author'] = { :value => author_name, :expires => Time.utc(2030) }
+      cookies['author'] = { :value => author_name.dup.as_bytes, :expires => Time.utc(2030) }
       if @page
         new_name = params['new_name'] ? params['new_name'].purify : @page_name
         prev_content = @page.current_revision.content
diff --git a/lib/stringsupport.rb b/lib/stringsupport.rb
index 731a1f95..459ca752 100644
--- a/lib/stringsupport.rb
+++ b/lib/stringsupport.rb
@@ -30,9 +30,9 @@ class String
 # returns a valid utf-8 string, purged of any subsequences of illegal bytes.
 #--
    def purify
-     text = check_ncrs
-     if text.respond_to?(:encoding)
-       text.split(//).collect{|c| c.as_bytes}.grep(UTF8_REGEX).join.as_utf8
+     text = self.dup.check_ncrs.as_utf8
+     if text.respond_to?(:force_encoding)
+       text.chars.collect{|c| c.as_bytes}.grep(UTF8_REGEX).join.as_utf8
      else
        text.split(//u).grep(UTF8_REGEX).join
      end
diff --git a/test/functional/wiki_controller_test.rb b/test/functional/wiki_controller_test.rb
old mode 100755
new mode 100644
index 8676b1ac..de43c267
--- a/test/functional/wiki_controller_test.rb
+++ b/test/functional/wiki_controller_test.rb
@@ -653,6 +653,18 @@ class WikiControllerTest < ActionController::TestCase
     assert_equal 'AuthorOfNewPage', new_page.author
   end
 
+  def test_save_astral_plane_characters
+    r = process 'save', 'web' => 'wiki1', 'id' => 'NewPage', 'content' => "Double-struck A: \xF0\x9D\x94\xB8", 
+      'author' => "\xF0\x9D\x94\xB8\xC3\xBCthorOfNewPage"
+    
+    assert_redirected_to :web => 'wiki1', :controller => 'wiki', :action => 'show', :id => 'NewPage'
+    assert_match @eternity, r.headers["Set-Cookie"][0]
+    new_page = @wiki.read_page('wiki1', 'NewPage')
+    assert_equal "Double-struck A: \360\235\224\270", new_page.content
+    assert_equal "\360\235\224\270\303\274thorOfNewPage", new_page.author
+    assert_equal "\360\235\224\270\303\274thorOfNewPage", r.cookies['author']
+  end
+
   def test_save_not_utf8
     r = process 'save', 'web' => 'wiki1', 'id' => 'NewPage', 'content' => "Cont\000ents of a new page\r\n\000", 
       'author' => 'AuthorOfNewPage'