Implement rb_utf8_str_new etc

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

@@ -411,6 +411,18 @@ static VALUE string_spec_rb_str_modify(VALUE self, VALUE str) {
   return str;
 }
 
+static VALUE string_spec_rb_utf8_str_new_static(VALUE self) {
+  return rb_utf8_str_new_static("nokogiri", 8);
+}
+
+static VALUE string_spec_rb_utf8_str_new(VALUE self) {
+  return rb_utf8_str_new("nokogiri", 8);
+}
+
+static VALUE string_spec_rb_utf8_str_new_cstr(VALUE self) {
+  return rb_utf8_str_new_cstr("nokogiri");
+}
+
 void Init_string_spec(void) {
   VALUE cls = rb_define_class("CApiStringSpecs", rb_cObject);
   rb_define_method(cls, "rb_cstr2inum", string_spec_rb_cstr2inum, 2);
@@ -485,6 +497,9 @@ void Init_string_spec(void) {
   rb_define_method(cls, "rb_String", string_spec_rb_String, 1);
   rb_define_method(cls, "rb_string_value_cstr", string_spec_rb_string_value_cstr, 1);
   rb_define_method(cls, "rb_str_modify", string_spec_rb_str_modify, 1);
+  rb_define_method(cls, "rb_utf8_str_new_static", string_spec_rb_utf8_str_new_static, 0);
+  rb_define_method(cls, "rb_utf8_str_new", string_spec_rb_utf8_str_new, 0);
+  rb_define_method(cls, "rb_utf8_str_new_cstr", string_spec_rb_utf8_str_new_cstr, 0);
 }
 
 #ifdef __cplusplus

spec/ruby/optional/capi/string_spec.rb

@@ -997,4 +997,28 @@ def inspect
       str.should == "2345678".encode("UTF-32LE")
     end
   end
+
+  describe "rb_utf8_str_new_static" do
+    it "returns a UTF-8 string of the correct characters and length" do
+      str = @s.rb_utf8_str_new_static
+      str.should == "nokogiri"
+      str.encoding.should == Encoding::UTF_8
+    end
+  end
+
+  describe "rb_utf8_str_new" do
+    it "returns a UTF-8 string of the correct characters and length" do
+      str = @s.rb_utf8_str_new
+      str.should == "nokogiri"
+      str.encoding.should == Encoding::UTF_8
+    end
+  end
+
+  describe "rb_utf8_str_new_cstr" do
+    it "returns a UTF-8 string of the correct characters and length" do
+      str = @s.rb_utf8_str_new_cstr
+      str.should == "nokogiri"
+      str.encoding.should == Encoding::UTF_8
+    end
+  end
 end

src/main/c/cext/ruby.c

@@ -4276,12 +4276,13 @@ VALUE rb_str_tmp_new(long len) {
 
 #undef rb_utf8_str_new
 VALUE rb_utf8_str_new(const char *ptr, long len) {
-  rb_tr_error("rb_utf8_str_new not implemented");
+  return rb_enc_str_new(ptr, len, rb_utf8_encoding());
 }
 
 #undef rb_utf8_str_new_cstr
 VALUE rb_utf8_str_new_cstr(const char *ptr) {
-  rb_tr_error("rb_utf8_str_new_cstr not implemented");
+  // TODO CS 11-Oct-19 would be nice to read in one go rather than strlen followed by read
+  return rb_utf8_str_new(ptr, strlen(ptr));
 }
 
 VALUE rb_str_new_static(const char *ptr, long len) {
@@ -4293,7 +4294,7 @@ VALUE rb_usascii_str_new_static(const char *ptr, long len) {
 }
 
 VALUE rb_utf8_str_new_static(const char *ptr, long len) {
-  rb_tr_error("rb_utf8_str_new_static not implemented");
+  return rb_utf8_str_new(ptr, len);
 }
 
 void rb_str_shared_replace(VALUE str, VALUE str2) {