How to find the length of the longest consecutive bit string(either 1 or 0)?
00000000 11110000 00000000 00000000 -> If it is 0 then length will be 20
11111
The following is based on the concept that if you AND
a bit sequence with a shifted version of itself, you're effectively removing the trailing 1 from a row of consecutive 1's.
11101111 (x)
& 11011110 (x << 1)
----------
11001110 (x & (x << 1))
^ ^
| |
trailing 1 removed
Repeating this N
times will reduce any sequence with N
consecutive 1's to 0x00
.
So, to count the number of consecutive 1's:
int count_consecutive_ones(int in) {
int count = 0;
while (in) {
in = (in & (in << 1));
count++;
}
return count;
}
To count the number of consecutive 0's, simply invert and the same routine.
int count_consecutive_zeros(int in) {
return count_consecutive_ones(~in);
}
Proof of concept: http://ideone.com/Z1l0D
int main(void) {
printf("%d has %d consecutive 1's\n", 0, count_consecutive_ones(0));
printf("%d has %d consecutive 0's\n", 0, count_consecutive_zeros(0));
/* 00000000 11110000 00000000 00000000 -> If it is 0 then length will be 20 */
printf("%x has %d consecutive 0's\n", 0x00F00000, count_consecutive_zeros(0x00F00000));
/* 11111111 11110000 11110111 11111111 -> If it is 1 then length will be 12 */
printf("%x has %d consecutive 1's\n", 0xFFF0F7FF, count_consecutive_ones(0xFFF0F7FF));
}
Output:
0 has 0 consecutive 1's
0 has 32 consecutive 0's
f00000 has 20 consecutive 0's
fff0f7ff has 12 consecutive 1's
public static int maxConsecutiveOneInBinaryNumber(int number) {
int count = 0;
int max = 0;
while (number != 0) {
if ((number & 1) == 1) {
count++;
} else {
max = Math.max(count, max);
count = 0;
}
number = number >> 1;
}
return Math.max(count, max);
}
You can this code here: https://github.com/VishalSKumar/DSFiddle/blob/master/src/main/java/com/vishalskumar/hackerrank/MaxConsecutiveOneInBinary.java
You can form a look up table to do it quickly for you. The bigger the table, the faster the lookup. 2x256 entry tables can do 8 bits at a time with a little bit twiddling. Add a 1s version of the table and start adding entries. That's probably how I'd go about it.
If you're just looking for a byte string of four bytes, you can pack these into an unsigned long
and use an algorithm like this:
int CountConsecutiveOnes(unsigned long n)
{
unsigned long m = n;
int k = 0;
while (m)
{
++k;
n >>= 1;
m &= n;
}
return k;
}
For counting zeros, just taking the bitwise complement first.
If you need to count byte strings longer than four, you can just implement the operations x >>= 1
and x & y
either directly on the byte strings or it may be more efficient to use strings of unsigned long
so the carry checks on the implementation of x >>= 1
aren't too expensive.
One simple way would be to simply loop over the bits, and keep track of the number of bits in a row which have had the same value, and the maximum that this value has reached.
Here's a simple C function which does just this:
int num_conseq_matching_bits(int n) {
int i, max, cur, b, prevb;
prevb = n & 1; /* 0th bit */
cur = 1;
max = 1;
for(i=1; i<32; i++) {
b = (n >> i) & 1; /* get the i'th bit's value */
if(b == prevb) {
cur += 1;
if(cur > max)
max = cur;
}
else {
cur = 1; /* count self */
prevb = b;
}
}
return max;
}
To use the table idea, you need something like
static struct {
int lead; /* leading 0 bits */
int max; /* maximum 0 bits */
int trail; /* trailing 0 bits */
} table[256] = { ....data.... };
int mostConsecutiveBits(unsigned char *str, int length, bool count_ones) {
int max = 0; /* max seen so far */
int trail = 0; /* trailing 0s from previous bytes */
while (length-- > 0) {
int byte = *str++;
if (count_ones)
byte ^= 0xff;
if (table[byte].max > max)
max = table[byte].max;
if (trail + table[byte].lead > max)
max = trail + table[byte].lead;
if (byte)
trail = table[byte].trail;
else
trail += 8;
}
return max;
}
initializing the table is straight-forward, but depends on your bit- and byte-ordering (little endian or big endian).