Fix Busted Functional Tests

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

vendor/plugins/HTML5lib/lib/html5/inputstream.rb

@@ -281,7 +281,11 @@ module HTML5
             @tell += 1 if @buffer[@tell] == "\n"
             c = "\n"
           when "\x80" .. "\x9F"
-            c = ''.force_encoding('UTF-8') << ENTITIES_WINDOWS1252[c.ord-0x80]
+            c = ENTITIES_WINDOWS1252[c.ord-0x80].chr('utf-8')
+          when "\xA0" .. "\xFF"
+            if c.encoding == Encoding::ASCII_8BIT
+              c = c.encode('utf-8','iso-8859-1')
+            end
           end
 
           if c == "\x0D"
@@ -299,8 +303,7 @@ module HTML5
             @col += 1
           end
 
-          # binary utf-8
-          c.ord > 126 ? [c.ord].pack('U') : c
+          c
 
         when 0x01..0x7F
           if c == 0x0D

vendor/plugins/HTML5lib/lib/html5/sniffer.rb

@@ -7,9 +7,9 @@ module Sniffer
 
     while pos < s.length
       case s[pos]
-      when 0x09, 0x20, 0x0A, 0x0D # tab, space, LF, CR
+      when ?\t, ?\ , ?\n, ?\r # 0x09, 0x20, 0x0A, 0x0D == tab, space, LF, CR
         pos += 1
-      when  0x3C # "<"
+      when ?< # 0x3C
         pos += 1
         if s[pos..pos+2] == "!--" # [0x21, 0x2D, 0x2D]
           pos += 3
@@ -17,13 +17,13 @@ module Sniffer
             pos += 1
           end
           pos += 3
-        elsif s[pos] == 0x21 # "!"
+        elsif s[pos] == ?! # 0x21
           pos += 1
-          until s[pos] == 0x3E or pos >= s.length # ">"
+          until s[pos] == ?> or pos >= s.length # 0x3E
             pos += 1 
           end
           pos += 1
-        elsif s[pos] == 0x3F # "?"
+        elsif s[pos] == ?? # 0x3F
           until s[pos..pos+1] == "?>" or pos >= s.length # [0x3F, 0x3E]
             pos +=  1
           end
@@ -42,4 +42,4 @@ module Sniffer
     "text/html"
   end
 end
-end
+end

vendor/plugins/HTML5lib/lib/html5/tokenizer.rb

@@ -118,7 +118,11 @@ module HTML5
       end
 
       if 0 < charAsInt and charAsInt <= 1114111 and not (55296 <= charAsInt and charAsInt <= 57343)
-        char = [charAsInt].pack('U')
+        if String.method_defined? :force_encoding
+          char = charAsInt.chr('utf-8')
+        else
+          char = [charAsInt].pack('U')
+        end
       else
         char = [0xFFFD].pack('U')
         @token_queue << {:type => :ParseError, :data => "cant-convert-numeric-entity", :datavars => {"charAsInt" => charAsInt}}