Good way to convert between short and bytes?

Question

I need to take pairs of bytes in, and output shorts, and take shorts in and output pairs of bytes. Here are the functions i've devised for such a purpose:

static short ToShort(short byte1, short byte2)
{
    short number = (short)byte2;
    number <<= 4;
    number += (short)byte1;
    return number;
}
static void FromShort(short number, out byte byte1, out byte byte2)
{
    byte byte2 = (byte)(number >> 4);
    short tempByte = (short)byte2 << 4;
    byte byte1 = (byte)(number - tempByte);
}

I think this is correct but i'm not sure. If this isn't the right way to do it, what is? is there a way to do this already in the framework?

Solution 1

Shorter version (also shifting 8 bits instead of 4):

static short ToShort(short byte1, short byte2)
{
    return (byte2 << 8) + byte1;
}

static void FromShort(short number, out byte byte1, out byte byte2)
{
    byte2 = (byte)(number >> 8);
    byte1 = (byte)(number & 255);
}

Solution 2

Use BitConverter

short number = 42;
byte[] numberBytes = BitConverter.GetBytes(number);
short converted = BitConverter.ToInt16(numberBytes);

Solution 3

Bytes are 8 bits, not 4, so your shifting is off. You also declared local variables in the second function so you wouldn't end up writing the the out parameters like you intend. It's also clearer/better if you limit yourself to bitwise operations (&, |, and ~) where possible.

static short ToShort(byte byte1, byte byte2)
{
    return (short) ((byte2 << 8) | (byte1 << 0));
}

static void FromShort(short number, out byte byte1, out byte byte2)
{
    byte2 = (byte) (number >> 8);
    byte1 = (byte) (number >> 0);
}

Note that the left and right shifts by zero are unnecessary, strictly speaking. I just put those in for symmetry. Also, personally I'd recommend you just learn bitwise arithmetic cold and skip writing helper functions like these. No need to hide the details with something so fundamental, IMHO.

Solution 4

If you want to take bytes... take bytes; and your shifts are off, and | would be more intuitive:

static short ToShort(byte byte1, byte byte2)
{   // using Int32 because that is what all the operations return anyway...
    return (short)((((int)byte1) << 8) | (int)byte2);
}
static void FromShort(short number, out byte byte1, out byte byte2)
{
    byte1 = (byte)(number >> 8); // to treat as same byte 1 from above
    byte2 = (byte)number;
}

View more solutions

Comments

[RCIX]

that taking bytes thing was actually a mistake... Good tips though, thanks!

[RCIX]

had to wrap the code in the ToShort method with a cast to short, just thought i'd let you know...

[RCIX]

Hmm didn't think of this but i like Ates' slution for now. Thanks!

[TJB]

Whatever works best for you! The upside of using a method like this is that if you have to use this code in other projects it will just be there instead of having to create a library and share code. Also, other developers can re-user their knowledge of BitConverter if they have any. Personally, we used to have some wrapper code for byte conversions we would share, but it became more of a hassle to maintain rather than just use the built in stuff. But seriously, use what works best for you ; )

[Ken Smith]

You'll typically want to bit shift byte2, not byte1. So something like: return (short)((byte2<<8) | byte1);

[Lama]

This is the only solution I'd recommend.

[Yiping]

This is the only solution that is correct for signed short

[Assimilater]

this doesn't compile...why so many upvotes? short + short resolves to an integer operator and cannot implicitly cast int as short

[Ates Goral]

@Assimilater In which compiler?

[Assimilater]

Microsoft Visual C# compiler targeting .NETv4.5.2...does it depend on .NET version?

[Assimilater]

As a c/c++ programmer at heart I find it silly that anyone would find it burdensome to do a simple bitshift and or operation...generating a byte[] and then however BitConverter is implemented to parse said byte[] afterwards just seems silly....

[Assimilater]

casting as a short is required in ToShort just before returning. See answer by @john-kugelman

[Assimilater]

though the int casts are unnecessary it shows what will happen anyways (and I think this is what you meant by the comment), so +1

[Rafael Diego Nicoletti]

I think BitConverter would consider big endian/little endian scenarios.

[Farid]

I just tested this method I don't know why it is returning wrong number after double conversion. Providing 36999 then doing FromShort and then converting it to back to short with ToShort the result becomes 65415