Move SQLite3 blob encoding to ActiveModel

Author: olefriis

activemodel/lib/active_model/type/immutable_string.rb

@@ -46,13 +46,20 @@ def type
       def serialize(value)
         case value
         when ::Numeric, ::Symbol, ActiveSupport::Duration then value.to_s
+        when ::String then serialize_cast_value(value)
         when true then @true
         when false then @false
         else super
         end
       end
 
       def serialize_cast_value(value) # :nodoc:
+        if value&.encoding == Encoding::BINARY
+          # If we can treat the bytes as UTF-8 without changing them, then use UTF-8 as encoding
+          new_value = value.dup.force_encoding(Encoding::UTF_8)
+          return new_value if new_value.valid_encoding?
+        end
+
         value
       end
 

activemodel/test/cases/type/immutable_string_test.rb

@@ -17,6 +17,26 @@ class ImmutableStringTest < ActiveModel::TestCase
         assert_same s, type.cast(s)
         assert_same s, type.deserialize(s)
       end
+
+      test "leaves validly encoded strings untouched" do
+        s = "string with àccénts".encode(Encoding::ISO_8859_1)
+        type = Type::ImmutableString.new
+        assert_same s, type.serialize(s)
+      end
+
+      test "serializes valid, binary-encoded strings to UTF-8" do
+        s = "string with àccénts".b
+        type = Type::ImmutableString.new
+        serialized = type.serialize(s)
+        assert_equal Encoding::UTF_8, serialized.encoding
+        assert_equal s.bytes, serialized.bytes
+      end
+
+      test "leaves true binary data untouched" do
+        binary_data = "\xEE\x49\xC7".b
+        type = Type::ImmutableString.new
+        assert_same binary_data, type.serialize(binary_data)
+      end
     end
   end
 end

activerecord/lib/active_record/connection_adapters/sqlite3/quoting.rb

@@ -79,12 +79,6 @@ def type_cast(value) # :nodoc:
           case value
           when BigDecimal
             value.to_f
-          when String
-            if value.encoding == Encoding::ASCII_8BIT
-              super(value.encode(Encoding::UTF_8))
-            else
-              super
-            end
           else
             super
           end