Merge pull request #2509 from cuberite/UTF16Types

Improved typing for utf16 and removed a couple of unused methods.

Tools/ProtoProxy/Connection.cpp

@@ -1835,12 +1835,12 @@ bool cConnection::HandleServerKick(void)
 			Reason.append(Split[4]);
 			Reason.push_back(0);
 			Reason.append(Split[5]);
-			AString ReasonBE16 = UTF8ToRawBEUTF16(Reason.data(), Reason.size());
+			auto ReasonBE16 = UTF8ToRawBEUTF16(Reason);
 			AString PacketStart("\xff");
-			PacketStart.push_back(static_cast<char>((ReasonBE16.size() / 2) / 256));
+			PacketStart.push_back(static_cast<char>(ReasonBE16.size() / 256));
-			PacketStart.push_back(static_cast<char>((ReasonBE16.size() / 2) % 256));
+			PacketStart.push_back(static_cast<char>(ReasonBE16.size() % 256));
 			CLIENTSEND(PacketStart.data(), PacketStart.size());
-			CLIENTSEND(ReasonBE16.data(), ReasonBE16.size());
+			CLIENTSEND(reinterpret_cast<const char *>(ReasonBE16.data()), ReasonBE16.size() * sizeof(char16_t));
 			return true;
 		}
 		else

src/ByteBuffer.cpp

@@ -473,22 +473,6 @@ bool cByteBuffer::ReadBool(bool & a_Value)
 
 
 
-bool cByteBuffer::ReadBEUTF16String16(AString & a_Value)
-{
-	CHECK_THREAD
-	CheckValid();
-	UInt16 Length;
-	if (!ReadBEUInt16(Length))
-	{
-		return false;
-	}
-	return ReadUTF16String(a_Value, Length);
-}
-
-
-
-
-
 bool cByteBuffer::ReadVarInt32(UInt32 & a_Value)
 {
 	CHECK_THREAD
@@ -929,24 +913,6 @@ bool cByteBuffer::ReadString(AString & a_String, size_t a_Count)
 
 
 
-bool cByteBuffer::ReadUTF16String(AString & a_String, size_t a_NumChars)
-{
-	// Reads 2 * a_NumChars bytes and interprets it as a UTF16 string, converting it into UTF8 string a_String
-	CHECK_THREAD
-	CheckValid();
-	AString RawData;
-	if (!ReadString(RawData, a_NumChars * 2))
-	{
-		return false;
-	}
-	RawBEToUTF8(RawData.data(), a_NumChars, a_String);
-	return true;
-}
-
-
-
-
-
 bool cByteBuffer::SkipRead(size_t a_Count)
 {
 	CHECK_THREAD

src/ByteBuffer.h

@@ -63,7 +63,6 @@ public:
 	bool ReadBEFloat        (float & a_Value);
 	bool ReadBEDouble       (double & a_Value);
 	bool ReadBool           (bool & a_Value);
-	bool ReadBEUTF16String16(AString & a_Value);  // string length as BE short, then string as UTF-16BE
 	bool ReadVarInt32       (UInt32 & a_Value);
 	bool ReadVarInt64       (UInt64 & a_Value);
 	bool ReadVarUTF8String  (AString & a_Value);  // string length as VarInt, then string as UTF-8
@@ -109,9 +108,6 @@ public:
 	/** Reads a_Count bytes into a_String; returns true if successful */
 	bool ReadString(AString & a_String, size_t a_Count);
 	
-	/** Reads 2 * a_NumChars bytes and interprets it as a UTF16-BE string, converting it into UTF8 string a_String */
-	bool ReadUTF16String(AString & a_String, size_t a_NumChars);
-	
 	/** Skips reading by a_Count bytes; returns false if not enough bytes in the ringbuffer */
 	bool SkipRead(size_t a_Count);
 	

src/Protocol/ProtocolRecognizer.cpp

@@ -200,10 +200,10 @@ void cProtocolRecognizer::SendDisconnect(const AString & a_Reason)
 		static const int Packet = 0xff;  // PACKET_DISCONNECT
 		SendData(reinterpret_cast<const char *>(&Packet), 1);  // WriteByte()
 
-		AString UTF16 = UTF8ToRawBEUTF16(a_Reason.c_str(), a_Reason.length());
+		auto UTF16 = UTF8ToRawBEUTF16(a_Reason);
-		static const u_short Size = htons(static_cast<u_short>(UTF16.size() / 2));
+		static const u_short Size = htons(static_cast<u_short>(UTF16.size()));
 		SendData(reinterpret_cast<const char *>(&Size), 2);      // WriteShort()
-		SendData(UTF16.data(), UTF16.size());  // WriteString()
+		SendData(reinterpret_cast<const char *>(UTF16.data()), UTF16.size() * sizeof(char16_t));  // WriteString()
 	}
 }
 

src/StringUtils.cpp

@@ -500,13 +500,13 @@ static bool isLegalUTF8(const unsigned char * source, int length)
 
 
 
-AString UTF8ToRawBEUTF16(const char * a_UTF8, size_t a_UTF8Length)
+std::u16string UTF8ToRawBEUTF16(const AString & a_UTF8)
 {
-	AString UTF16;
+	std::u16string UTF16;
-	UTF16.reserve(a_UTF8Length * 3);
+	UTF16.reserve(a_UTF8.size() * 2);
 
-	const unsigned char * source    = reinterpret_cast<const unsigned char *>(a_UTF8);
+	const unsigned char * source    = reinterpret_cast<const unsigned char *>(a_UTF8.data());
-	const unsigned char * sourceEnd = source + a_UTF8Length;
+	const unsigned char * sourceEnd = source + a_UTF8.size();
 	const int halfShift  = 10;  // used for shifting by 10 bits
 	const unsigned int halfBase = 0x0010000UL;
 	const unsigned int halfMask = 0x3ffUL;
@@ -546,13 +546,13 @@ AString UTF8ToRawBEUTF16(const char * a_UTF8, size_t a_UTF8Length)
 				ch = ' ';
 			}
 			unsigned short v = htons(static_cast<unsigned short>(ch));
-			UTF16.append(reinterpret_cast<const char *>(&v), 2);
+			UTF16.push_back(static_cast<char16_t>(v));
 		}
 		else if (ch > UNI_MAX_UTF16)
 		{
 			// Invalid value, replace with a space
 			unsigned short v = htons(' ');
-			UTF16.append(reinterpret_cast<const char *>(&v), 2);
+			UTF16.push_back(static_cast<char16_t>(v));
 		}
 		else
 		{
@@ -560,8 +560,8 @@ AString UTF8ToRawBEUTF16(const char * a_UTF8, size_t a_UTF8Length)
 			ch -= halfBase;
 			unsigned short v1 = htons((ch >> halfShift) + UNI_SUR_HIGH_START);
 			unsigned short v2 = htons((ch & halfMask) + UNI_SUR_LOW_START);
-			UTF16.append(reinterpret_cast<const char *>(&v1), 2);
+			UTF16.push_back(static_cast<char16_t>(v1));
-			UTF16.append(reinterpret_cast<const char *>(&v2), 2);
+			UTF16.push_back(static_cast<char16_t>(v2));
 		}
 	}
 	return UTF16;

src/StringUtils.h

@@ -82,7 +82,7 @@ extern void ReplaceString(AString & iHayStack, const AString & iNeedle, const AS
 extern AString & RawBEToUTF8(const char * a_RawData, size_t a_NumShorts, AString & a_UTF8);
 
 /** Converts a UTF-8 string into a UTF-16 BE string. */
-extern AString UTF8ToRawBEUTF16(const char * a_UTF8, size_t a_UTF8Length);
+extern std::u16string UTF8ToRawBEUTF16(const AString & a_String);
 
 /** Creates a nicely formatted HEX dump of the given memory block.
 Max a_BytesPerLine is 120. */