Oculus VR, Inc.

Bitstream Overview
Description

The bitstream class is a helper class under the namespace RakNet that is used to wrap a dynamic array for the purpose of packing and unpacking bits. Its main four benefits are:
  1. Creating packets dynamically
  2. Compression
  3. Writing Bits
  4. Endian swapping

With structs you have to predefine your structures and cast them to a (char*). With a bitstream, you can choose to write blocks at runtime, depending on the context. Bitstreams can compress the native types using the SerializeBitsFromIntegerRange and SerializeFloat16().

You can also write bits. Most of the time you will not care about this. However, when writing booleans it will automatically only write one bit. This can also be very useful for encryption since your data will no longer be byte aligned.

Writing Data

Bitstream is templated to take any type of data. If this is a built-in type, such as NetworkIDObject, it uses partial template specialization to write the type more efficiently. If it's a native type, or a structure, it writes the individual bits, similar to memcpy. You can pass structs containing multiple data members to bitstreams. However, you may wish to serialize each individual element to do correct endian swapping (needed for communication between PCs and Macs, for example).

struct MyVector
{
float x,y,z;
} myVector;

// No endian swapping
bitStream.Write(myVector);

// With endian swapping
#undef __BITSTREAM_NATIVE_END
bitStream.Write(myVector.x);
bitStream.Write(myVector.y);
bitStream.Write(myVector.z);

// You can also override operator left shift and right shift

// Shift operators have to be in the namespace RakNet or they might use the default one in BitStream.h instead. Error occurs with std::string
namespace RakNet
{

RakNet::BitStream& operator << (RakNet::BitStream& out, MyVector& in)
{
out.WriteNormVector(in.x,in.y,in.z);
return out;
}
RakNet::BitStream& operator >> (RakNet::BitStream& in, MyVector& out)
{
bool success = in.ReadNormVector(out.x,out.y,out.z);
assert(success);
return in;
}

} // namespace RakNet

// Read from bitstream
myVector << bitStream;
// Write to bitstream
myVector >> bitStream;

Optional - One of the constructor versions takes a length in bytes as a parameter. If you have an idea of the size of your data you can pass this number when creating the bitstream to avoid internal reallocations.

See Creating Packets for more details.

Reading Data

Reading data is equally simple. Create a bitstream, and in the constructor assign it your data.

// Assuming we have a Packet *
BitStream myBitStream(packet->data, packet->length, false);
struct MyVector
{
float x,y,z;
} myVector;

// No endian swapping
bitStream.Read(myVector);

// With endian swapping (__BITSTREAM_NATIVE_END should just be commented in RakNetDefines.h)
#undef __BITSTREAM_NATIVE_END
#include "BitStream.h" bitStream.Read(myVector.x);
bitStream.Read(myVector.y);
bitStream.Read(myVector.z);

See Receiving Packets for a more complete example.

Serializing Data

You can have the same function read and write, by using BitStream::Serialize() instead of Read() or Write().

struct MyVector
{
float x,y,z;
// writeToBitstream==true means write, writeToBitstream==false means read
void Serialize(bool writeToBitstream, BitStream *bs)
{
bs->Serialize(writeToBitstream, x);
bs->Serialize(writeToBitstream, y);
bs->Serialize(writeToBitstream, z);
}
} myVector;

See Receiving Packets for a more complete example.

Useful functions

See BitStream.h for a full list of functions.

Reset
Restart the bitstream, clearing all data.

Write functions
The write functions write data to the bitstream at the end of the bitstream. You should use the analogous Read to get the data back out.

Read functions
The read functions read data already in the bitstream, in order from beginning to end. The read function returns false if there is no more data in the bitstream.

WriteBitsFromIntegerRange, ReadBitsFromIntegerRange, SerializeBitsFromIntegerRange
If a number only uses a specific range of values (such as 10-20) this function will determine automatically the number of bits needed to write that range of values.

WriteCasted, ReadCasted
Write a variable of one type as if it were casted to another type. For example, WriteCasted<char>(5); is equivalent to char c=5; Write(c);

WriteNormVector, ReadNormVector
Writes a normalized vector where each component ranges from -1 to +1. Each component is written with 16 bites.

WriteFloat16, ReadFloat16
Given a min and max value for a floating point number, divide the range by 65535 and write the result in 16 bytes (lossy)..

WriteNormQuat, ReadNormQuat
Write a quaternion in 16*3+4 bits (lossy).

WriteOrthMatrix, ReadOrthMatrix
Convert an orthonormal matrix to a quaternion, then call WriteNormQuat/ReadNormQuat.

GetNumberOfBitsUsed
GetNumberOfBytesUsed
Gives you the number of bytes or bits written.

GetData
Gives you a pointer to the internal data of the bitstream. This is a (char*) allocated with malloc and is presented in case you need direct assess to the data.


See Also
Index
Creating Packets
Receiving Packets