How to convert java BigDecimal to normal byte array (not 2's complement)

Question

How do I convert from big integer to a byte array which is not in 2's complement format. Bascially I only need to convert positive numbers and do not need the sign bit.

So something like 10 would become a byte 0x0a i.e-> 00001010

[Update] As per comment I tried this

public void testBinary()
{
    BigDecimal test = new BigDecimal(35116031);
    BigInteger theInt = test.unscaledValue();
    byte[] arr = theInt.toByteArray();
    System.out.println(getCounterVal(arr, new BigInteger("256")));
}
public BigInteger getCounterVal(byte[] arr, BigInteger multiplier)
{
    BigInteger counter = BigInteger.ZERO;
    for(int i = (arr.length - 1); i >=0; i--)
    {
        int b = arr[i];
        //int val = (int) b & 0xFF;
        BigInteger augend = BigInteger.valueOf(b);
        counter = counter.add(augend.multiply(multiplier.pow(i)));
    }
    return counter;
}

The out put value I got was -19720446 And with the //int val = (int) b & 0xFF; uncommented and used as augend, I got the value 4292024066

[Update2] Here is a test I ran which works. Not sure if it is bug free but looks fine.

@Test
public void bigIntegerToArray()
{
    BigInteger bigInt = new BigInteger("35116444");
    byte[] array = bigInt.toByteArray();
    if (array[0] == 0)
    {
        byte[] tmp = new byte[array.length - 1];
        System.arraycopy(array, 1, tmp, 0, tmp.length);
        array = tmp;
    }

    BigInteger derived = BigInteger.ZERO;
    BigInteger twofiftysix = new BigInteger("256");
    int j = 0;
    for (int i = array.length - 1; i >= 0; i--)
    {
        int val = (int) array[i] & 0xFF;
        BigInteger addend = BigInteger.valueOf(val);
        BigInteger multiplier = twofiftysix.pow(j);
        addend = addend.multiply(multiplier);
        derived = derived.add(addend);
        j++;
    }

    Assert.assertEquals(bigInt, derived);
}

Answer 1

The difference is largely conceptual. Unsigned numbers are the same in 2's compliment. 2's compliment just describes how to represent negative numbers which you say you don't have.

i.e. 10 is 00001010 in signed and unsigned representation.

To get the bytes from a BigDecimal or BigInteger you can use the methods it provides.

BigDecimal test = new BigDecimal(35116031);
BigInteger theInt = test.unscaledValue();
byte[] arr = theInt.toByteArray();
System.out.println(Arrays.toString(arr));

BigInteger bi2 = new BigInteger(arr);
BigDecimal bd2 = new BigDecimal(bi2, 0);
System.out.println(bd2);

prints

[2, 23, -45, -1]
35116031

The bytes are correct and reproduce the same value.

There is a bug in the way you rebuild your BigInteger. You assume the byte serialization is little endian when Java typically uses big endian http://en.wikipedia.org/wiki/Endianness

Answer 2

Try to split the number in bytes, by dividing by 256 in each iteration and using the remainder, and place all these bytes into an array.

Answer 3

the sign bit in 2-compliment for positive numbers is 0

so signed or unsigned doesn't make a difference for positive numbers

Answer 4

If the value is less than the size of a long then use longValue, then chop the long into bytes. If the value is bigger than a long then probably you need to use an iterative approach, repeatedly dividing the number by 256, taking the remainder as the next byte, then repeating until you get zero. The bytes would be generated right to left. Signed numbers require thought (to generate 2s-complement results) but aren't much more complicated.