merge charlock_holmes to master

1 files changed, 6 insertions, 8 deletions

diff --git a/lib/utils.rb b/lib/utils.rb
index 2de14761444..8e5d4694cac 100644
--- a/lib/utils.rb
+++ b/lib/utils.rb
@@ -17,15 +17,13 @@ module Utils
   end
 
   module CharEncode
-    def encode(string)
-      return '' unless string
-      cd = CharDet.detect(string)
-      if cd.confidence > 0.6
-        string.force_encoding(cd.encoding)
+    def encode(content)
+      content ||= ''
+      detection = CharlockHolmes::EncodingDetector.detect(content)
+      if hash = detection
+        content = CharlockHolmes::Converter.convert(content, hash[:encoding], 'UTF-8') if hash[:encoding]
       end
-      string.encode("utf-8", :undef => :replace, :replace => "?", :invalid => :replace)
-    rescue
-      "Invalid Encoding"
+      content
     end
   end