[GR-45621] Add C API rb_enc_interned_str_cstr function

PullRequest: truffleruby/4149

CHANGELOG.md

@@ -56,6 +56,8 @@ Compatibility:
 * Remove deprecated `FileTest.exists?` method (#3039, @andrykonchin).
 * Fix {Method,Proc}#parameters and return `*`, `**` and `&` names for anonymous parameters (@andrykonchin).
 * Remove deprecated `Fixnum` and `Bignum` constants (#3039, @andrykonchin).
+* Add `rb_enc_interned_str_cstr` function (#3408, @goyox86, @thomasmarshall).
+* Add `rb_str_to_interned_str` function (#3408, @thomasmarshall).
 
 Performance:
 

lib/cext/ABI_check.txt

@@ -1 +1 @@
-10
+13

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

@@ -51,12 +51,6 @@ VALUE string_spec_rb_str_set_len_RSTRING_LEN(VALUE self, VALUE str, VALUE len) {
   return INT2FIX(RSTRING_LEN(str));
 }
 
-VALUE rb_fstring(VALUE str); /* internal.h, used in ripper */
-
-VALUE string_spec_rb_str_fstring(VALUE self, VALUE str) {
-  return rb_fstring(str);
-}
-
 VALUE string_spec_rb_str_buf_new(VALUE self, VALUE len, VALUE str) {
   VALUE buf;
 
@@ -578,11 +572,19 @@ static VALUE string_spec_rb_str_unlocktmp(VALUE self, VALUE str) {
   return rb_str_unlocktmp(str);
 }
 
+static VALUE string_spec_rb_enc_interned_str_cstr(VALUE self, VALUE str, VALUE enc) {
+  rb_encoding *e = rb_to_encoding(enc);
+  return rb_enc_interned_str_cstr(RSTRING_PTR(str), e);
+}
+
+static VALUE string_spec_rb_str_to_interned_str(VALUE self, VALUE str) {
+  return rb_str_to_interned_str(str);
+}
+
 void Init_string_spec(void) {
   VALUE cls = rb_define_class("CApiStringSpecs", rb_cObject);
   rb_define_method(cls, "rb_cstr2inum", string_spec_rb_cstr2inum, 2);
   rb_define_method(cls, "rb_cstr_to_inum", string_spec_rb_cstr_to_inum, 3);
-  rb_define_method(cls, "rb_fstring", string_spec_rb_str_fstring, 1);
   rb_define_method(cls, "rb_str2inum", string_spec_rb_str2inum, 2);
   rb_define_method(cls, "rb_str_append", string_spec_rb_str_append, 2);
   rb_define_method(cls, "rb_str_buf_new", string_spec_rb_str_buf_new, 2);
@@ -679,6 +681,8 @@ void Init_string_spec(void) {
   rb_define_method(cls, "rb_str_catf", string_spec_rb_str_catf, 1);
   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_str_to_interned_str", string_spec_rb_str_to_interned_str, 1);
 }
 
 #ifdef __cplusplus

spec/ruby/optional/capi/string_spec.rb

@@ -498,25 +498,6 @@ def inspect
     end
   end
 
-  describe "rb_fstring" do
-    it 'returns self if the String is frozen' do
-      input  = 'foo'.freeze
-      output = @s.rb_fstring(input)
-
-      output.should equal(input)
-      output.should.frozen?
-    end
-
-    it 'returns a frozen copy if the String is not frozen' do
-      input  = 'foo'
-      output = @s.rb_fstring(input)
-
-      output.should.frozen?
-      output.should_not equal(input)
-      output.should == 'foo'
-    end
-  end
-
   describe "rb_str_subseq" do
     it "returns a byte-indexed substring" do
       str = "\x00\x01\x02\x03\x04".force_encoding("binary")
@@ -1227,4 +1208,59 @@ def inspect
       -> { @s.rb_str_unlocktmp("test") }.should raise_error(RuntimeError, 'temporal unlocking already unlocked string')
     end
   end
+
+  describe "rb_enc_interned_str_cstr" do
+    it "returns a frozen string" do
+      str = "hello"
+      val = @s.rb_enc_interned_str_cstr(str, 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_cstr(str, Encoding::US_ASCII)
+      result2 = @s.rb_enc_interned_str_cstr(str, 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_cstr(str, Encoding::US_ASCII)
+      result2 = @s.rb_enc_interned_str_cstr(str, Encoding::UTF_8)
+      result1.should_not.equal?(result2)
+    end
+
+    it "returns the same string as String#-@" do
+      @s.rb_enc_interned_str_cstr("hello", Encoding::UTF_8).should.equal?(-"hello")
+    end
+  end
+
+  describe "rb_str_to_interned_str" do
+    it "returns a frozen string" do
+      str = "hello"
+      result = @s.rb_str_to_interned_str(str)
+      result.should.is_a?(String)
+      result.should.frozen?
+    end
+
+    it "returns the same frozen string" do
+      str = "hello"
+      result1 = @s.rb_str_to_interned_str(str)
+      result2 = @s.rb_str_to_interned_str(str)
+      result1.should.equal?(result2)
+    end
+
+    it "returns different frozen strings for different encodings" do
+      result1 = @s.rb_str_to_interned_str("hello".force_encoding(Encoding::US_ASCII))
+      result2 = @s.rb_str_to_interned_str("hello".force_encoding(Encoding::UTF_8))
+      result1.should_not.equal?(result2)
+    end
+
+    it "returns the same string as String#-@" do
+      @s.rb_str_to_interned_str("hello").should.equal?(-"hello")
+    end
+  end
 end

src/main/c/cext/string.c

@@ -160,7 +160,7 @@ VALUE rb_str_to_str(VALUE string) {
 }
 
 VALUE rb_fstring(VALUE str) {
-  return RUBY_INVOKE(str, "-@");
+  return rb_str_to_interned_str(str);
 }
 
 VALUE rb_str_buf_new(long capacity) {
@@ -437,3 +437,12 @@ long rb_str_coderange_scan_restartable(const char *s, const char *e, rb_encoding
   *cr = ENC_CODERANGE_VALID;
   return e - s;
 }
+
+VALUE rb_enc_interned_str_cstr(const char *ptr, rb_encoding *enc) {
+  VALUE str = rb_enc_str_new_cstr(ptr, enc);
+  return rb_str_to_interned_str(str);
+}
+
+VALUE rb_str_to_interned_str(VALUE str) {
+  return RUBY_INVOKE(str, "-@");
+}

src/main/java/org/truffleruby/cext/CExtNodes.java

@@ -796,6 +796,8 @@ RubyString rbStrNewNul(int byteLength,
 
     }
 
+    /** Alternative to rb_str_new*() which does not copy the bytes from native memory, to use when the copy is
+     * unnecessary. */
     @CoreMethod(names = "rb_tr_temporary_native_string", onSingleton = true, required = 3, lowerFixnum = 2)
     public abstract static class TemporaryNativeStringNode extends CoreMethodArrayArgumentsNode {