From 4ee94a6574634496346dca1e5a8c1eb119aeb82a Mon Sep 17 00:00:00 2001
From: bootstraponline <code@bootstraponline.com>
Date: Thu, 29 Nov 2012 23:47:25 -0700
Subject: [PATCH] Fix UTF-8 header and add test

---
 lib/gollum/frontend/views/page.rb |  7 ++++---
 test/test_app.rb                  | 16 +++++++++++++++-
 2 files changed, 19 insertions(+), 4 deletions(-)

diff --git a/lib/gollum/frontend/views/page.rb b/lib/gollum/frontend/views/page.rb
index a359e66a..4329d64a 100644
--- a/lib/gollum/frontend/views/page.rb
+++ b/lib/gollum/frontend/views/page.rb
@@ -109,7 +109,7 @@ module Precious
       # Wraps page formatted data to Nokogiri::HTML document.
       #
       def build_document(content)
-        Nokogiri::HTML(%{<div id="gollum-root">} + content + %{</div>})
+        Nokogiri::HTML::fragment(%{<div id="gollum-root">} + content.to_s + %{</div>}, 'UTF-8')
       end
 
       # Finds header node inside Nokogiri::HTML document.
@@ -134,7 +134,7 @@ module Precious
       def page_header_from_content(content)
         doc = build_document(content)
         title = find_header_node(doc)
-        Sanitize.clean(title.to_html(:encoding => 'UTF-8')).strip unless title.empty?
+        Sanitize.clean(title.to_xhtml(:encoding => 'UTF-8')).strip unless title.empty?
       end
 
       # Returns page content without title if it was extracted.
@@ -143,7 +143,8 @@ module Precious
         doc = build_document(content)
         title = find_header_node(doc)
         title.remove unless title.empty?
-        doc.css("div#gollum-root").inner_html
+        # .inner_html will cause href escaping on UTF-8
+        doc.css("div#gollum-root").children.to_xhtml(:encoding => 'UTF-8')
       end
     end
   end
diff --git a/test/test_app.rb b/test/test_app.rb
index dadbc85c..02a48066 100644
--- a/test/test_app.rb
+++ b/test/test_app.rb
@@ -38,10 +38,24 @@ context "Frontend" do
     assert_equal 'm-plus-f', 'μ†ℱ'.to_url
   end
 
-  test 'utf-8 kcode' do
+  test "utf-8 kcode" do
     assert_equal 'μ†ℱ'.scan(/./), ["μ", "†", "ℱ"]
   end
 
+  test "UTF-8 headers href preserved" do
+    page = 'utfh1'
+    text = '한글'
+
+    # don't use h1 or it will be promoted to replace file name
+    # which doesn't generate a normal header link
+    @wiki.write_page(page, :markdown, '## ' + text,
+                     { :name => 'user1', :email => 'user1' });
+
+    get page
+
+    assert_match /<h2>#{text}<a class="anchor" id="#{text}" href="##{text}"><\/a><\/h2>/, last_response.body
+  end
+
   test "retain edit information" do
     page1 = 'page1'
     user1 = 'user1'