refactor(string): Change Get_Utf8_Size and Get_Unicode_Size to exclude null terminator

Author: bobtista

Core/GameEngine/Source/Common/System/AsciiString.cpp

@@ -276,11 +276,11 @@ void AsciiString::translate(const UnicodeString& stringSrc)
 	// TheSuperHackers @fix bobtista 02/04/2026 Implement UTF-8 conversion replacing 7-bit ASCII only implementation
 	clear();
 	const WideChar* src = stringSrc.str();
-	if (!src[0])
-		return;
 	size_t size = Get_Utf8_Size(src);
-	char* buf = getBufferForRead((Int)(size - 1));
-	Unicode_To_Utf8(buf, src, size);
+	if (size == 0)
+		return;
+	char* buf = getBufferForRead((Int)size);
+	Unicode_To_Utf8(buf, src, size + 1);
 	validate();
 }
 

Core/GameEngine/Source/Common/System/UnicodeString.cpp

@@ -225,11 +225,11 @@ void UnicodeString::translate(const AsciiString& stringSrc)
 	// TheSuperHackers @fix bobtista 02/04/2026 Implement UTF-8 conversion replacing 7-bit ASCII only implementation
 	clear();
 	const char* src = stringSrc.str();
-	if (!src[0])
-		return;
 	size_t size = Get_Unicode_Size(src);
-	WideChar* buf = getBufferForRead((Int)(size - 1));
-	Utf8_To_Unicode(buf, src, size);
+	if (size == 0)
+		return;
+	WideChar* buf = getBufferForRead((Int)size);
+	Utf8_To_Unicode(buf, src, size + 1);
 	validate();
 }
 

Core/GameEngine/Source/GameNetwork/GameSpy/Thread/ThreadUtils.cpp

@@ -36,8 +36,10 @@
 std::wstring MultiByteToWideCharSingleLine( const char *orig )
 {
 	size_t size = Get_Unicode_Size(orig);
-	std::wstring ret(size - 1, L'\0');
-	Utf8_To_Unicode(&ret[0], orig, size);
+	if (size == 0)
+		return std::wstring();
+	std::wstring ret(size, L'\0');
+	Utf8_To_Unicode(&ret[0], orig, size + 1);
 	WideChar *c = nullptr;
 	do
 	{
@@ -64,8 +66,10 @@ std::wstring MultiByteToWideCharSingleLine( const char *orig )
 std::string WideCharStringToMultiByte( const WideChar *orig )
 {
 	size_t size = Get_Utf8_Size(orig);
-	std::string ret(size - 1, '\0');
-	Unicode_To_Utf8(&ret[0], orig, size);
+	if (size == 0)
+		return std::string();
+	std::string ret(size, '\0');
+	Unicode_To_Utf8(&ret[0], orig, size + 1);
 	return ret;
 }
 

Core/Libraries/Source/WWVegas/WWLib/utf8.cpp

@@ -87,13 +87,13 @@ bool Utf8_Validate(const char* str, size_t length)
 size_t Get_Utf8_Size(const wchar_t* src)
 {
 	int bytes = WideCharToMultiByte(CP_UTF8, 0, src, -1, nullptr, 0, nullptr, nullptr);
-	return (bytes > 0) ? (size_t)bytes : 1;
+	return (bytes > 1) ? (size_t)(bytes - 1) : 0;
 }
 
 size_t Get_Unicode_Size(const char* src)
 {
 	int wchars = MultiByteToWideChar(CP_UTF8, 0, src, -1, nullptr, 0);
-	return (wchars > 0) ? (size_t)wchars : 1;
+	return (wchars > 1) ? (size_t)(wchars - 1) : 0;
 }
 
 void Unicode_To_Utf8(char* dest, const wchar_t* src, size_t dest_size)

Core/Libraries/Source/WWVegas/WWLib/utf8.h

@@ -33,12 +33,12 @@ size_t Utf8_Trailing_Invalid_Bytes(const char* str, size_t length);
 bool Utf8_Validate(const char* str);
 bool Utf8_Validate(const char* str, size_t length);
 
-// Returns the number of bytes required to store the UTF-8 representation of src,
-// including the null terminator.
+// Returns the number of bytes in the UTF-8 representation of src, excluding the
+// null terminator. Returns 0 on failure or if src is empty.
 size_t Get_Utf8_Size(const wchar_t* src);
 
-// Returns the number of wchar_t elements required to store the wide character
-// representation of the UTF-8 string src, including the null terminator.
+// Returns the number of wchar_t elements in the wide character representation of
+// the UTF-8 string src, excluding the null terminator. Returns 0 on failure or if src is empty.
 size_t Get_Unicode_Size(const char* src);
 
 // Converts a wide character string to UTF-8. dest_size is in bytes.