Release 2.4.3

to_html uses source document encoding by default. Set UTF-8.

Add note about KCODE warning.

gollum.gemspec

@@ -5,8 +5,8 @@ Gem::Specification.new do |s|
   s.required_ruby_version = ">= 1.8.7"
 
   s.name              = 'gollum'
-  s.version           = '2.4.1'
-  s.date              = '2012-11-19'
+  s.version           = '2.4.3'
+  s.date              = '2012-11-29'
   s.rubyforge_project = 'gollum'
 
   s.summary     = "A simple, Git-powered wiki."
@@ -32,7 +32,7 @@ Gem::Specification.new do |s|
   s.add_dependency('sanitize', '~> 2.0.3')
   s.add_dependency('nokogiri', '~> 1.5.5')
   s.add_dependency('useragent', '~> 0.4.13')
-  s.add_dependency('stringex', '~> 1.4.0')
+  s.add_dependency('stringex', '~> 1.5.0')
 
   s.add_development_dependency('RedCloth', '~> 4.2.9')
   s.add_development_dependency('mocha', '~> 0.13.0')

lib/gollum.rb

@@ -23,8 +23,13 @@ require File.expand_path('../gollum/sanitization', __FILE__)
 require File.expand_path('../gollum/web_sequence_diagram', __FILE__)
 require File.expand_path('../gollum/frontend/uri_encode_component', __FILE__)
 
+# Set ruby to UTF-8 mode
+# Do not remove because of warning.
+# This is required for Ruby 1.8.7 which gollum still supports.
+$KCODE = 'U'
+
 module Gollum
-  VERSION = '2.4.1'
+  VERSION = '2.4.3'
 
   def self.assets_path
     ::File.expand_path('gollum/frontend/public', ::File.dirname(__FILE__))

lib/gollum/frontend/app.rb

@@ -18,8 +18,7 @@ class String
   # _Header => header which causes errors
   def to_url
     return nil if self.nil?
-    return self if ['_Header', '_Footer', '_Sidebar'].include? self
-    upstream_to_url
+    upstream_to_url :exclude => ['_Header', '_Footer', '_Sidebar']
   end
 end
 

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).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

lib/gollum/markup.rb

@@ -126,7 +126,11 @@ module Gollum
         node.add_child(%Q{<a href="##{h_name}">#{h.content}</a>})
         tail.add_child(node)
       end
-      toc = toc.to_xml(@to_xml) if toc != nil
+      if toc != nil
+        # convert to HTML first before XHTML
+        toc = Nokogiri::HTML::fragment toc.to_s
+        toc = toc.to_xhtml(:encoding => 'UTF-8')
+      end
       [doc, toc]
     end
 

test/test_app.rb

@@ -15,6 +15,47 @@ context "Frontend" do
     FileUtils.rm_rf(@path)
   end
 
+  test "urls transform unicode" do
+    header = '_Header'
+    footer = '_Footer'
+    sidebar = '_Sidebar'
+
+    # header, footer, and sidebar must be preserved
+    # or gollum will not recognize them
+    assert_equal header, header.to_url
+    assert_equal footer, footer.to_url
+    assert_equal sidebar, sidebar.to_url
+
+    # spaces are converted to dashes in URLs
+    # and in file names saved to disk
+    # urls are not case sensitive
+    assert_equal 'title-space', 'Title Space'.to_url
+
+    # ascii only file names prevent UTF8 issues
+    # when using git repos across operating systems
+    # as this test demonstrates, translation is not
+    # great
+    assert_equal 'm-plus-f', 'μ†ℱ'.to_url
+  end
+
+  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'

test/test_unicode.rb

@@ -41,6 +41,28 @@ context "Unicode Support" do
     assert_equal '',       anchors[0].text
   end
 
+  def check_h1 text, page
+      @wiki.write_page(page, :markdown, "# " + text)
+
+      page = @wiki.page(page)
+      assert_equal Gollum::Page, page.class
+      assert_equal '# ' + text, utf8(page.raw_data)
+
+      output = page.formatted_data
+
+      # UTF-8 headers should not be encoded.
+      assert_match /<h1>#{text}<a class="anchor" id="#{text}" href="##{text}"><\/a><\/h1>/,   output
+  end
+
+  test "create and read non-latin page with anchor" do
+    # href="#한글"
+    # href="#%ED%95%9C%EA%B8%80"
+    check_h1 '한글', '1'
+    # href="#한글"
+    # href="#Synht%C3%A8se"
+    check_h1 'Synhtèse', '2'
+  end
+
   test "create and read non-latin page with anchor 2" do
     @wiki.write_page("test", :markdown, "# \"La\" faune d'Édiacara")