How to solve this problem?Give some Hints please!

> How to solve this problem?Give some Hints please!
The method got from Lalescu Liviu:
1057: I have to calculate how many numbers are from 0 to x, which,
written in
base b, have only digits 0 and 1, and they have exactly k digits of
1. First,
I convert x in base b, I denoted it ax(nx-1), ax(nx-2)...ax(1), ax
(0). I
denote by f(ax,nx,nzeroes) the number of valid numbers from 0 to x,
where
nzeroes is the number of zeroes I have to obtain, and which is nx-k.
Then:
f(ax,nx,nzeroes)=
1) if ax[nx-1]=0, I have to put 0 and return f(ax,nx-1,nzeroes-1)
(remember my
number has to stay below x)
2) if ax[nx-1]=1, I can put 0 and then any combination is below x,
that is we
obtain comb(nx-1,nzeroes-1) variants (comb means binomial coefficient
or
combinations), or put 1 and calculate f(ax,nx-1,nzeroes).
3) if ax[nx-1]>=2, I can put 0 or 1, and obtain
comb(n1-1,nzeroes-1)+comb(n1-1,nzeroes) possibilities, because any
number
generated is below x.