How to generate random vector with specific hamming weight in GF(q) ?

edit

If under the hamming weight you understand the number of nonzero components, then something along these lines will do the job:

def random_vector(q,n,w):
    F.<b> = GF(q, modulus='primitive')
    S = Subsets(range(n),w).random_element()
    return vector([b^randint(0,q-2) if k in S else F.zero() for k in range(n)])

Here random_vector(5^2,10,3) returns a random vector over $GF(5^2)$ with 3 nonzero out of total 10 components, such as (0, 0, 2*z2 + 3, 0, 4*z2 + 1, 0, 0, 4*z2 + 1, 0, 0).