[emscripten::val] Support literal utf-8/16 strings (#16349)

It already has supports for cstrings with Latin1 encoding. And std::string
could be used for utf-8 encoded strings, but with cost on extra copy of
wire type.
I don't want to extend cstring for utf encoding so that it keeps fast in
most cases. Instead, add new static functions to specially support u8string
and u16string.

Author: xbcnn

src/embind/emval.js

@@ -134,6 +134,18 @@ var LibraryEmVal = {
     return Emval.toHandle(getStringOrSymbol(v));
   },
 
+  _emval_new_u8string__sig: 'ii',
+  _emval_new_u8string__deps: ['$Emval'],
+  _emval_new_u8string: function(v) {
+    return Emval.toHandle(UTF8ToString(v));
+  },
+
+  _emval_new_u16string__sig: 'ii',
+  _emval_new_u16string__deps: ['$Emval'],
+  _emval_new_u16string: function(v) {
+    return Emval.toHandle(UTF16ToString(v));
+  },
+
   _emval_take_value__sig: 'iii',
   _emval_take_value__deps: ['$Emval', '$requireRegisteredType'],
   _emval_take_value: function(type, argv) {

system/include/emscripten/val.h

@@ -54,6 +54,8 @@ void _emval_run_destructors(EM_DESTRUCTORS handle);
 EM_VAL _emval_new_array();
 EM_VAL _emval_new_object();
 EM_VAL _emval_new_cstring(const char*);
+EM_VAL _emval_new_u8string(const char*);
+EM_VAL _emval_new_u16string(const char16_t*);
 
 EM_VAL _emval_take_value(TYPEID type, EM_VAR_ARGS argv);
 
@@ -351,6 +353,14 @@ class val {
         return val(internal::_emval_new_object());
     }
 
+    static val u8string(const char* s) {
+        return val(internal::_emval_new_u8string(s));
+    }
+
+    static val u16string(const char16_t* s) {
+        return val(internal::_emval_new_u16string(s));
+    }
+
     static val undefined() {
         return val(EM_VAL(internal::_EMVAL_UNDEFINED));
     }

tests/embind/test_val.cpp

@@ -639,6 +639,26 @@ int main()
   ensure(aAsNumberVectorUint32_t.at(2) == 0);      // 0 is returned if can not be converted for integers
   ensure(aAsNumberVectorUint32_t.at(3) == 100000); // Date returns milliseconds since epoch
 
+  test("val u8string(const char* s)");
+  val::global().set("a", val::u8string(u8"abc"));
+  ensure_js("a == 'abc'");
+  val::global().set("a", val::u8string(u8"你好"));
+  ensure_js_not("a == 'abc'");
+  ensure_js("a == '你好'");
+  auto u8_str = val::global()["a"].as<std::string>();
+  ensure(u8_str == u8"你好");
+
+  test("val u16string(const char16_t* s)");
+  val::global().set("a", val::u16string(u"hello"));
+  ensure_js("a == 'hello'");
+  val::global().set("a", val::u16string(u"世界"));
+  ensure_js_not("a == 'hello'");
+  ensure_js("a == '世界'");
+  // UTF-16 encoded SMILING FACE WITH OPEN MOUTH (U+1F603)
+  const char16_t* s = u"😃 = \U0001F603 is :-D";
+  val::global().set("a", val::u16string(s));
+  ensure_js("a == '😃 = \U0001F603 is :-D'");
+  
   printf("end\n");
   return 0;
 }

tests/embind/test_val.out

@@ -268,4 +268,16 @@ pass
 pass
 pass
 pass
+test:
+val u8string(const char* s)
+pass
+pass
+pass
+pass
+test:
+val u16string(const char16_t* s)
+pass
+pass
+pass
+pass
 end