[GR-18163] Implement rb_enc_interned_str (#3706)

PullRequest: truffleruby/4393

Author: andrykonchin

CHANGELOG.md

@@ -18,6 +18,7 @@ Compatibility:
 * Support the index/length arguments for the string argument to `String#bytesplice` added in 3.3 (#3656, @rwstauner).
 * Implement `rb_str_strlen()` (#3697, @Th3-M4jor).
 * Support `Time.new` with String argument and error when invalid (#3693, @rwstauner).
+* Implement `rb_enc_interned_str()` (#3703, @Th3-M4jor).
 
 Performance:
 

lib/cext/include/truffleruby/truffleruby-abi-version.h

@@ -20,6 +20,6 @@
 // $RUBY_VERSION must be the same as TruffleRuby.LANGUAGE_VERSION.
 // $ABI_NUMBER starts at 1 and is incremented for every ABI-incompatible change.
 
-#define TRUFFLERUBY_ABI_VERSION "3.2.4.5"
+#define TRUFFLERUBY_ABI_VERSION "3.2.4.6"
 
 #endif

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

@@ -581,6 +581,11 @@ static VALUE string_spec_rb_enc_interned_str_cstr(VALUE self, VALUE str, VALUE e
   return rb_enc_interned_str_cstr(RSTRING_PTR(str), e);
 }
 
+static VALUE string_spec_rb_enc_interned_str(VALUE self, VALUE str, VALUE len, VALUE enc) {
+  rb_encoding *e = NIL_P(enc) ? 0 : rb_to_encoding(enc);
+  return rb_enc_interned_str(RSTRING_PTR(str), FIX2LONG(len), e);
+}
+
 static VALUE string_spec_rb_str_to_interned_str(VALUE self, VALUE str) {
   return rb_str_to_interned_str(str);
 }
@@ -687,6 +692,7 @@ void Init_string_spec(void) {
   rb_define_method(cls, "rb_str_locktmp", string_spec_rb_str_locktmp, 1);
   rb_define_method(cls, "rb_str_unlocktmp", string_spec_rb_str_unlocktmp, 1);
   rb_define_method(cls, "rb_enc_interned_str_cstr", string_spec_rb_enc_interned_str_cstr, 2);
+  rb_define_method(cls, "rb_enc_interned_str", string_spec_rb_enc_interned_str, 3);
   rb_define_method(cls, "rb_str_to_interned_str", string_spec_rb_str_to_interned_str, 1);
 }
 

spec/ruby/optional/capi/string_spec.rb

@@ -1261,6 +1261,51 @@ def inspect
     end
   end
 
+  describe "rb_enc_interned_str" do
+    it "returns a frozen string" do
+      str = "hello"
+      val = @s.rb_enc_interned_str(str, str.bytesize, Encoding::US_ASCII)
+
+      val.should.is_a?(String)
+      val.encoding.should == Encoding::US_ASCII
+      val.should.frozen?
+    end
+
+    it "returns the same frozen string" do
+      str = "hello"
+      result1 = @s.rb_enc_interned_str(str, str.bytesize, Encoding::US_ASCII)
+      result2 = @s.rb_enc_interned_str(str, str.bytesize, Encoding::US_ASCII)
+      result1.should.equal?(result2)
+    end
+
+    it "returns different frozen strings for different encodings" do
+      str = "hello"
+      result1 = @s.rb_enc_interned_str(str, str.bytesize, Encoding::US_ASCII)
+      result2 = @s.rb_enc_interned_str(str, str.bytesize, Encoding::UTF_8)
+      result1.should_not.equal?(result2)
+    end
+
+    it 'returns the same string when using non-ascii characters' do
+      str = 'こんにちは'
+      result1 = @s.rb_enc_interned_str(str, str.bytesize, Encoding::UTF_8)
+      result2 = @s.rb_enc_interned_str(str, str.bytesize, Encoding::UTF_8)
+      result1.should.equal?(result2)
+    end
+
+    it "returns the same string as String#-@" do
+      str = "hello"
+      @s.rb_enc_interned_str(str, str.bytesize, Encoding::UTF_8).should.equal?(-str)
+    end
+
+    ruby_bug "#20322", ""..."3.4" do
+      it "uses the default encoding if encoding is null" do
+        str = "hello"
+        val = @s.rb_enc_interned_str(str, str.bytesize, nil)
+        val.encoding.should == Encoding::ASCII_8BIT
+      end
+    end
+  end
+
   describe "rb_str_to_interned_str" do
     it "returns a frozen string" do
       str = "hello"

src/main/c/cext/string.c

@@ -437,6 +437,11 @@ VALUE rb_enc_interned_str_cstr(const char *ptr, rb_encoding *enc) {
   return rb_str_to_interned_str(str);
 }
 
+VALUE rb_enc_interned_str(const char *ptr, long len, rb_encoding *enc) {
+  VALUE str = rb_enc_str_new(ptr, len, enc ? enc : rb_ascii8bit_encoding());
+  return rb_str_to_interned_str(str);
+}
+
 VALUE rb_str_to_interned_str(VALUE str) {
   return RUBY_INVOKE(str, "-@");
 }