[SOLVED] Bit packing

Aquarius_Girl · 03-18-2010, 04:08 AM

Suppose there is a number 1010 1101 0011

I want to set to 0, all the bits after the first 3 ones.

If I AND the following two numbers, I'll solve my problem

1010 1101 0011
&
1010 0000 0000
=
1010 0000 0000

Question:
How to generate the number 1010 0000 0000 ?

It is not clicking me somehow, how to use AND, OR and shift operations to get that number !

Guttorm · 03-18-2010, 04:36 AM

Hi

When you "and" to clear bits, use 1 for all the bits you want to keep and 0 for all the others.

1010 1101 0011
&
1110 0000 0000

Aquarius_Girl · 03-18-2010, 04:55 AM

Thanks for answering !

Now I am thinking how to generalize it.
i.e.

If user specifies

3, we will AND with 111, whereas 111 actually means 7
4, we will AND with 1111, whereas 1111 actually means 15
5, we will AND with 11111, whereas 11111 actually means 31

Now the task is how to make the program understand that when user has specified 3 we have to AND with 7 ... ?

I don't think making a mapping table will be a good idea, there must be some better way to do it ?

Aquarius_Girl · 03-18-2010, 05:04 AM

And I think I have got it now:

3 : 1 + 2 will result in 111
4 : 1 + 2 + 4 + 8 will result in 1111
5 : 1 + 2 + 4 + 8 + 16 will result in 11111

See all the numbers I am adding are the multiples of 2.

Guttorm · 03-18-2010, 05:05 AM

The formula will be (2^number)-1

2^3-1=7
2^4-1=15
...

Aquarius_Girl · 03-18-2010, 05:08 AM

Thanks again!

I think your and my formula are same ! Aren't they ?

Aquarius_Girl · 03-18-2010, 05:19 AM

Your formula is easier to implement than mine, thanks once again

Aquarius_Girl · 03-19-2010, 01:35 AM

crts · 03-19-2010, 02:13 AM

Hi,

I am not sure if I understand your question. If your data is in an array (of element length 1 byte) then you could just XOR the last element with 1110 0000 or the first with 0000 0111. Not exactly sure what you mean with first three bits - MSB or LSB? And am I understanding right that in this context you want to interpret the 4096 byte as 'one huge number'?

Aquarius_Girl · 03-19-2010, 03:41 AM

Aquarius_Girl · 03-19-2010, 04:57 AM

Quote:

Originally Posted by crts

I am not sure if I understand your question. If your data is in an array (of element length 1 byte) then you could just XOR the last element with 1110 0000 or the first with 0000 0111. Not exactly sure what you mean with first three bits - MSB or LSB? And am I understanding right that in this context you want to interpret the 4096 byte as 'one huge number'?

Thanks for bothering to reply !

I apologize for wasting your time, there was a communication gap between me and my senior, hence resulting in a wrong question, and now the question what I have understood is easily solvable !

Thanks again !

Aquarius_Girl · 04-01-2010, 01:33 AM

I have successfully written the bit packing code and now comes the turn of unpacking it:

In the following code data is the name of the array that contains the packed bits.

Code:

void packedMessage :: getUnpackedMsg (unsigned char *emptyArray, unsigned int numberOfBits)
{
    unsigned int   bitsPending   = numberOfBits;
    unsigned int   andWith       = 0;
    unsigned int   andResult     = 0;

    // Calculating the number of bytes
    int            remainder     = numberOfBits % 8;
    int            quotient      = numberOfBits / 8;	
    unsigned short numberOfBytes = quotient;	
    
    if (remainder > 0)
    {
	numberOfBytes++;
    }
    
    unsigned int   i = 0;
    
    while (numberOfBytes > 0)
    {
	if (numberOfBytes == 1)
	{
	    andWith       = (pow (2, bitsPending)) - 1;
	    andResult     = data[index] & andWith;
	    emptyArray[i] = emptyArray[i] | andResult;
	    --index;
	}
	else if ((numberOfBytes > 1) && (bitsPending >= 8))
	{
	    emptyArray[i] = emptyArray[i] | data[index];
	    bitsPending   = bitsPending - 8;
	    
	    ++i;
	    --index;
	}
	
	--numberOfBytes;
    }

    printf ("\nAfter unpacking: %d ", emptyArray[0]);
    printf ("\nAfter unpacking: %d ", emptyArray[1]);
    cout << "\n";
}

Main function from which I am calling the above

Code:

int main ()
{
	
// Test case : 2 bytes 11 bits

	int byte = 2047;
	    
	packedMessage p;	
	p.getPackedMsg   ((unsigned char*)&byte, 11);

	p.setDataArray   ((unsigned char*)&byte, 11);

	unsigned int *empty = new unsigned int [5];
	p.getUnpackedMsg ((unsigned char*)&empty, 11);

Question:
How I am supposed to print what ever I have unpacked, Those print statements in the unpack function do not seem to work properly !

If my question is not clear please let me know !
Help kindly!!