Use '?' for replacement char sometimes

natalie-lang · Jun 22, 2024 · 1839164 · 1839164
1 parent 4365e7b
commit 1839164
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 11 deletions.
diff --git a/spec/core/string/encode_spec.rb b/spec/core/string/encode_spec.rb
@@ -104,11 +104,9 @@
     end
 
     it "replaces undefined encoding in destination with default replacement" do
-      NATFIXME 'undef option' do
-        encoded = "B\ufffd".encode(Encoding::US_ASCII, undef: :replace)
-        encoded.should == "B?".encode(Encoding::US_ASCII)
-        encoded.encode("UTF-8").should == "B?"
-      end
+      encoded = "B\ufffd".encode(Encoding::US_ASCII, undef: :replace)
+      encoded.should == "B?".encode(Encoding::US_ASCII)
+      encoded.encode("UTF-8").should == "B?"
     end
 
     it "replaces undefined encoding in destination with a specified replacement" do

diff --git a/spec/core/string/shared/encode.rb b/spec/core/string/shared/encode.rb
@@ -158,10 +158,8 @@
     end
 
     it "replaces invalid characters in the destination encoding" do
-      NATFIXME 'encode options' do
-        xFF = [0xFF].pack('C').force_encoding('utf-8')
-        "ab#{xFF}c".send(@method, Encoding::ISO_8859_1, invalid: :replace).should == "ab?c"
-      end
+      xFF = [0xFF].pack('C').force_encoding('utf-8')
+      "ab#{xFF}c".send(@method, Encoding::ISO_8859_1, invalid: :replace).should == "ab?c"
     end
 
     it "calls #to_hash to convert the options object" do

diff --git a/src/encoding_object.cpp b/src/encoding_object.cpp
@@ -78,7 +78,10 @@ Value EncodingObject::encode(Env *env, EncodingObject *orig_encoding, StringObje
                     temp_string.append(options.replace_option);
                     continue;
                 }
-                unicode_codepoint = 0xFFFD;
+                if (is_single_byte_encoding())
+                    unicode_codepoint = '?';
+                else
+                    unicode_codepoint = 0xFFFD;
             }
         } else {
             unicode_codepoint = orig_encoding->to_unicode_codepoint(source_codepoint);
@@ -134,7 +137,10 @@ Value EncodingObject::encode(Env *env, EncodingObject *orig_encoding, StringObje
                     temp_string.append(options.replace_option);
                     continue;
                 }
-                destination_codepoint = 0xFFFD;
+                if (is_single_byte_encoding())
+                    destination_codepoint = '?';
+                else
+                    destination_codepoint = 0xFFFD;
             }
         }