ue4同c#通信时的中文乱码问题

本文讨论C#同ue4进行通信，出现的中文乱码情况，其他语言类似。

本文分两种情况讨论，

1.C#向ue4发送string

在C#发送string时，要先区分发送的字符串中是否包含中文，判断方法很简单，如下：

        private bool IsPureAnsi(string str)
        {
            for (int i = 0; i < str.Length; ++i)
            {
                if ((int)str[i] > 127)
                {
                    return false;
                }
            }

            return true;
        }

当string中不包含中文时，每个字符占1个字节，string前要先发送string的长度，占4个字节，string结束要加'\0'结尾（c传统）

当string中包含中文时，每个字节占用2个字节，格式同上（注意string长度无需*2，而且string长度要取反，string结尾需要两个字节的'\0'）

具体代码如下：

        public byte[] StringToBytes(string str)
        {
            Stream stream = new MemoryStream();

            if (this.IsPureAnsi(str))
            {
                byte[] strLenBytes = System.BitConverter.GetBytes(str.Length + 1);
                stream.Write(strLenBytes, 0, strLenBytes.Length);

                byte[] strBytes = Encoding.ASCII.GetBytes(str);
                stream.Write(strBytes, 0, strBytes.Length);
                stream.WriteByte(0);
            }
            else
            {
                int strLen = str.Length + 1;
                byte[] strLenBytes = System.BitConverter.GetBytes(-strLen);
                stream.Write(strLenBytes, 0, strLenBytes.Length);

                byte[] strBytes = Encoding.Unicode.GetBytes(str);
                stream.Write(strBytes, 0, strBytes.Length);
                stream.WriteByte(0);
                stream.WriteByte(0);
            }

            return StreamToBytes(stream);
        }

2.ue4向C#发送string

这种情况，ue4端照常发送，c#解析的时候需要做一定处理

c#在接收到string数据时，先取出string长度信息，如果string长度>=0,则说明c#接收到的string是ascii编码，如果小于0，说明接收到的string是unicode编码

具体代码如下

        public String GetString(byte[] data, int index, out int outIndex)
        {
            int strLen = System.BitConverter.ToInt32(data, index);
            bool Ascii = strLen >= 0;
            if (Ascii)
            {
                outIndex = index + 4 + strLen;

                int begin = index + 4;
                int end = outIndex;
                byte[] byteStr = data.Skip(begin).Take(end - begin - 1).ToArray();

                string str = Encoding.Default.GetString(byteStr);
                return str;
            }
            else
            {
                strLen = -strLen;
                outIndex = index + 4 + strLen*2;

                int begin = index + 4;
                int end = outIndex;
                byte[] byteStr = data.Skip(begin).Take(end - begin - 2).ToArray();

                string str = Encoding.Unicode.GetString(byteStr);
                return str;
            }
        }

以上内容主要参考的是ue4中的源码，具体源码如下：

FArchive& operator<<( FArchive& Ar, FString& A )
{
	// > 0 for ANSICHAR, < 0 for UCS2CHAR serialization

	if (Ar.IsLoading())
	{
		int32 SaveNum;
		Ar << SaveNum;

		bool LoadUCS2Char = SaveNum < 0;
		if (LoadUCS2Char)
		{
			SaveNum = -SaveNum;
		}

		// If SaveNum is still less than 0, they must have passed in MIN_INT. Archive is corrupted.
		if (SaveNum < 0)
		{
			Ar.ArIsError = 1;
			Ar.ArIsCriticalError = 1;
			UE_LOG(LogNetSerialization, Error, TEXT("Archive is corrupted"));
			return Ar;
		}

		auto MaxSerializeSize = Ar.GetMaxSerializeSize();
		// Protect against network packets allocating too much memory
		if ((MaxSerializeSize > 0) && (SaveNum > MaxSerializeSize))
		{
			Ar.ArIsError         = 1;
			Ar.ArIsCriticalError = 1;
			UE_LOG( LogNetSerialization, Error, TEXT( "String is too large" ) );
			return Ar;
		}

		// Resize the array only if it passes the above tests to prevent rogue packets from crashing
		A.Data.Empty           (SaveNum);
		A.Data.AddUninitialized(SaveNum);

		if (SaveNum)
		{
			if (LoadUCS2Char)
			{
				// read in the unicode string and byteswap it, etc
				auto Passthru = StringMemoryPassthru<UCS2CHAR>(A.Data.GetData(), SaveNum, SaveNum);
				Ar.Serialize(Passthru.Get(), SaveNum * sizeof(UCS2CHAR));
				// Ensure the string has a null terminator
				Passthru.Get()[SaveNum-1] = '\0';
				Passthru.Apply();

				INTEL_ORDER_TCHARARRAY(A.Data.GetData())

				// Since Microsoft's vsnwprintf implementation raises an invalid parameter warning
				// with a character of 0xffff, scan for it and terminate the string there.
				// 0xffff isn't an actual Unicode character anyway.
				int Index = 0;
				if(A.FindChar(0xffff, Index))
				{
					A[Index] = '\0';
					A.TrimToNullTerminator();		
				}
			}
			else
			{
				auto Passthru = StringMemoryPassthru<ANSICHAR>(A.Data.GetData(), SaveNum, SaveNum);
				Ar.Serialize(Passthru.Get(), SaveNum * sizeof(ANSICHAR));
				// Ensure the string has a null terminator
				Passthru.Get()[SaveNum-1] = '\0';
				Passthru.Apply();
			}

			// Throw away empty string.
			if (SaveNum == 1)
			{
				A.Data.Empty();
			}
		}
	}
	else
	{
		bool  SaveUCS2Char = Ar.IsForcingUnicode() || !FCString::IsPureAnsi(*A);
		int32 Num        = A.Data.Num();
		int32 SaveNum    = SaveUCS2Char ? -Num : Num;

		Ar << SaveNum;

		A.Data.CountBytes( Ar );

		if (SaveNum)
		{
			if (SaveUCS2Char)
			{
				// TODO - This is creating a temporary in order to byte-swap.  Need to think about how to make this not necessary.
				#if !PLATFORM_LITTLE_ENDIAN
					FString  ATemp  = A;
					FString& A      = ATemp;
					INTEL_ORDER_TCHARARRAY(A.Data.GetData());
				#endif

				Ar.Serialize((void*)StringCast<UCS2CHAR>(A.Data.GetData(), Num).Get(), sizeof(UCS2CHAR)* Num);
			}
			else
			{
				Ar.Serialize((void*)StringCast<ANSICHAR>(A.Data.GetData(), Num).Get(), sizeof(ANSICHAR)* Num);
			}
		}
	}

	return Ar;
}

注意，ue4,发送和接收数据都是用这段代码，Ar.IsLoading()为true代表接收数据，否则为发送数据