Fix Busted Functional Tests

Fix the functional tests busted by Revision 212.
Sync with latest HTML5lib.

vendor/plugins/HTML5lib/test/test_lxp.rb

@@ -9,7 +9,11 @@ def assert_xml_equal(input, expected=nil, parser=HTML5::XMLParser)
   document = parser.parse(input.chomp, :lowercase_attr_name => false, :lowercase_element_name => false).root
   if not expected
     expected = input.chomp.gsub(XMLELEM,&sortattrs)
-    expected = expected.gsub(/&#(\d+);/) {[$1.to_i].pack('U')}
+    if expected.respond_to? :force_encoding
+      expected = expected.gsub(/&#(\d+);/) {$1.to_i.chr('utf-8')}
+    else
+      expected = expected.gsub(/&#(\d+);/) {[$1.to_i].pack('U')}
+    end
     output = document.to_s.gsub(/'/,'"').gsub(XMLELEM,&sortattrs)
     assert_equal(expected, output)
   else

vendor/plugins/HTML5lib/test/test_sanitizer.rb

@@ -12,11 +12,11 @@ class SanitizeTest < Test::Unit::TestCase
   include HTML5
 
   def sanitize_xhtml stream
-    XHTMLParser.parse_fragment(stream, {:tokenizer => HTMLSanitizer, :encoding => 'utf-8', :lowercase_element_name => false, :lowercase_attr_name => false}).to_s
+    XHTMLParser.parse_fragment(stream, {:tokenizer => HTMLSanitizer, :encoding => 'utf-8', :lowercase_element_name => false, :lowercase_attr_name => false}).join
   end
 
   def sanitize_html stream
-    HTMLParser.parse_fragment(stream, {:tokenizer => HTMLSanitizer, :encoding => 'utf-8', :lowercase_element_name => false, :lowercase_attr_name => false}).to_s
+    HTMLParser.parse_fragment(stream, {:tokenizer => HTMLSanitizer, :encoding => 'utf-8', :lowercase_element_name => false, :lowercase_attr_name => false}).join
   end
 
   def sanitize_rexml stream

vendor/plugins/HTML5lib/test/test_serializer.rb

@@ -50,7 +50,7 @@ class Html5SerializeTestcase < Test::Unit::TestCase
           flunk("Expected: #{expected.inspect}, Received: #{result.inspect}")
         end
 
-        return if test_name == 'optionaltags'
+        next if test_name == 'optionaltags'
 
         result = HTML5::XHTMLSerializer.
           serialize(JsonWalker.new(test["input"]), (test["options"] || {}))
@@ -60,7 +60,6 @@ class Html5SerializeTestcase < Test::Unit::TestCase
         elsif !expected.include?(result)
           flunk("Expected: #{expected.inspect}, Received: #{result.inspect}")
         end
-
       end
 
     end

vendor/plugins/HTML5lib/test/test_stream.rb

@@ -5,6 +5,14 @@ require 'html5/inputstream'
 class HTMLInputStreamTest < Test::Unit::TestCase
   include HTML5
 
+  def getc stream
+    if String.method_defined? :force_encoding
+      stream.char.force_encoding('binary')
+    else
+      stream.char
+    end
+  end
+
   def test_char_ascii
     stream = HTMLInputStream.new("'", :encoding=>'ascii')
     assert_equal('ascii', stream.char_encoding)
@@ -13,23 +21,23 @@ class HTMLInputStreamTest < Test::Unit::TestCase
 
   def test_char_null
     stream = HTMLInputStream.new("\x00")
-    assert_equal("\xef\xbf\xbd", stream.char)
+    assert_equal("\xef\xbf\xbd", getc(stream))
   end
 
   def test_char_utf8
     stream = HTMLInputStream.new("\xe2\x80\x98", :encoding=>'utf-8')
     assert_equal('utf-8', stream.char_encoding)
-    assert_equal("\xe2\x80\x98", stream.char)
+    assert_equal("\xe2\x80\x98", getc(stream))
   end
 
   def test_char_win1252
     stream = HTMLInputStream.new("\xa2\xc5\xf1\x92\x86")
     assert_equal('windows-1252', stream.char_encoding)
-    assert_equal("\xc2\xa2", stream.char)
-    assert_equal("\xc3\x85", stream.char)
-    assert_equal("\xc3\xb1", stream.char)
-    assert_equal("\xe2\x80\x99", stream.char)
-    assert_equal("\xe2\x80\xa0", stream.char)
+    assert_equal("\xc2\xa2", getc(stream))
+    assert_equal("\xc3\x85", getc(stream))
+    assert_equal("\xc3\xb1", getc(stream))
+    assert_equal("\xe2\x80\x99", getc(stream))
+    assert_equal("\xe2\x80\xa0", getc(stream))
   end
 
   def test_bom