Fix unicode characters

Fix #547

Update test

lib/gollum/markup.rb

@@ -76,7 +76,7 @@ module Gollum
       doc,toc = process_headers(doc)
       @toc = @sub_page ? ( @parent_page ? @parent_page.toc_data : "[[_TOC_]]" ) : toc
       yield doc if block_given?
-      data = doc.to_html
+      data = doc.to_xhtml
 
       data = process_toc_tags(data)
       data = process_tex(data)
@@ -95,16 +95,12 @@ module Gollum
     def process_headers(doc)
       toc = nil
       doc.css('h1,h2,h3,h4,h5,h6').each do |h|
-        id = encodeURIComponent(h.content.gsub(' ','-'))
+        h_name = h.content.gsub(' ','-')
+
         level = h.name.gsub(/[hH]/,'').to_i
 
         # Add anchors
-        anchor = Nokogiri::XML::Node.new('a', doc)
+        h.add_child(%Q{<a class="anchor" id="#{h_name}" href="##{h_name}"></a>})
-        anchor['class'] = 'anchor'
-        anchor['id'] = id
-        # % -> %25 so anchors work on Firefox. See issue #475
-        anchor['href'] = '#' + id.gsub('%', '%25')
-        h.add_child(anchor)
 
         # Build TOC
         toc ||= Nokogiri::XML::DocumentFragment.parse('<div class="toc"><div class="toc-title">Table of Contents</div></div>')
@@ -122,7 +118,7 @@ module Gollum
         end
         node = Nokogiri::XML::Node.new('li', doc)
         # % -> %25 so anchors work on Firefox. See issue #475
-        node.add_child(%Q{<a href="##{id.gsub("%", "%25")}">#{h.content}</a>})
+        node.add_child(%Q{<a href="##{h_name}">#{h.content}</a>})
         tail.add_child(node)
       end
       toc = toc.to_xhtml if toc != nil

test/test_unicode.rb

@@ -38,10 +38,33 @@ context "Unicode Support" do
     anchors = h1 / :a
     assert_equal 1, h1s.size
     assert_equal 1, anchors.size
-    assert_equal '#%25ED%2595%259C%25EA%25B8%2580', anchors[0]['href']
+    assert_equal '#한글',  anchors[0]['href']
-    assert_equal '%ED%95%9C%EA%B8%80',              anchors[0]['id']
+    assert_equal  '한글',  anchors[0]['id']
-    assert_equal 'anchor',                          anchors[0]['class']
+    assert_equal 'anchor', anchors[0]['class']
-    assert_equal '',                                anchors[0].text
+    assert_equal '',       anchors[0].text
+  end
+
+  test "create and read non-latin page with anchor 2" do
+    @wiki.write_page("test", :markdown, "# La faune d'Édiacara")
+
+    page = @wiki.page("test")
+    assert_equal Gollum::Page, page.class
+    assert_equal "# La faune d'Édiacara", utf8(page.raw_data)
+
+    # markup.rb
+    # #简介
+    # href.gsub('%', '%25') so the anchor works in Firefox.
+    # <a href="#%25ED%2595%259C%25EA%25B8%2580" id="%ED%95%9C%EA%B8%80" class="anchor"></a>
+    doc     = Nokogiri::HTML page.formatted_data
+    h1s     = doc / :h1
+    h1      = h1s.first
+    anchors = h1 / :a
+    assert_equal 1, h1s.size
+    assert_equal 1, anchors.size
+    assert_equal %q(#La-faune-d'Édiacara), anchors[0]['href']
+    assert_equal %q(La-faune-d'Édiacara),  anchors[0]['id']
+    assert_equal 'anchor',                 anchors[0]['class']
+    assert_equal '',                       anchors[0].text
   end
 
   test "unicode with existing format rules" do