[GR-44710] Fix rb_enc_vsprintf and force message encoding instead o…

…f converting

PullRequest: truffleruby/4012

CHANGELOG.md

@@ -10,6 +10,7 @@ Bug fixes:
 Compatibility:
 
 * Add `Exception#detailed_message` method (#3257, @andrykonchin).
+* Fix `rb_enc_vsprintf` and force String encoding instead of converting it (@andrykonchin).
 
 Performance:
 

lib/cext/ABI_check.txt

@@ -1 +1 @@
-5
+6

spec/ruby/optional/capi/encoding_spec.rb

@@ -657,6 +657,20 @@
     end
   end
 
+  describe "rb_enc_raise" do
+    it "forces exception message encoding to the specified one" do
+      utf_8_incompatible_string = "\x81".b
+
+      -> {
+        @s.rb_enc_raise(Encoding::UTF_8, RuntimeError, utf_8_incompatible_string)
+      }.should raise_error { |e|
+        e.message.encoding.should == Encoding::UTF_8
+        e.message.valid_encoding?.should == false
+        e.message.bytes.should == utf_8_incompatible_string.bytes
+      }
+    end
+  end
+
   describe "rb_uv_to_utf8" do
     it 'converts a Unicode codepoint to a UTF-8 C string' do
       str = ' ' * 6

spec/ruby/optional/capi/ext/encoding_spec.c

@@ -271,6 +271,13 @@ static VALUE encoding_spec_rb_enc_str_asciionly_p(VALUE self, VALUE str) {
   }
 }
 
+static void encoding_spec_rb_enc_raise(VALUE self, VALUE encoding, VALUE exception_class, VALUE format) {
+  rb_encoding *e = rb_to_encoding(encoding);
+  const char *f = RSTRING_PTR(format);
+
+  rb_enc_raise(e, exception_class, f);
+}
+
 static VALUE encoding_spec_rb_uv_to_utf8(VALUE self, VALUE buf, VALUE num) {
   int len = rb_uv_to_utf8(RSTRING_PTR(buf), NUM2INT(num));
   RB_ENC_CODERANGE_CLEAR(buf);
@@ -368,6 +375,7 @@ void Init_encoding_spec(void) {
   rb_define_method(cls, "rb_enc_nth", encoding_spec_rb_enc_nth, 2);
   rb_define_method(cls, "rb_enc_codepoint_len", encoding_spec_rb_enc_codepoint_len, 1);
   rb_define_method(cls, "rb_enc_str_asciionly_p", encoding_spec_rb_enc_str_asciionly_p, 1);
+  rb_define_method(cls, "rb_enc_raise", encoding_spec_rb_enc_raise, 3);
   rb_define_method(cls, "rb_uv_to_utf8", encoding_spec_rb_uv_to_utf8, 2);
   rb_define_method(cls, "ONIGENC_MBC_CASE_FOLD", encoding_spec_ONIGENC_MBC_CASE_FOLD, 1);
   rb_define_method(cls, "rb_enc_left_char_head", encoding_spec_rb_enc_left_char_head, 2);

src/main/c/cext/printf.c

@@ -56,13 +56,17 @@ VALUE rb_enc_vsprintf(rb_encoding *enc, const char *format, va_list args) {
   VALUE types = RUBY_CEXT_INVOKE("rb_tr_sprintf_types", rubyFormat);
   VALUE rubyArgs = rb_tr_get_sprintf_args(args, types);
 
-  return rb_str_conv_enc(rb_tr_wrap(
-                    polyglot_invoke(
-                                    RUBY_CEXT,
-                                    "rb_tr_sprintf",
-                                    rb_tr_unwrap(rubyFormat),
-                                    rb_tr_vsprintf_new_cstr,
-                                    rb_tr_unwrap(rubyArgs))), NULL, enc);
+  VALUE string = rb_tr_wrap(
+    polyglot_invoke(
+      RUBY_CEXT,
+      "rb_tr_sprintf",
+      rb_tr_unwrap(rubyFormat),
+      rb_tr_vsprintf_new_cstr,
+      rb_tr_unwrap(rubyArgs)
+    )
+  );
+
+  return rb_enc_associate(string, enc);
 }
 
 VALUE rb_enc_sprintf(rb_encoding *enc, const char *format, ...) {