Concatenate pyton tuples in numba

Question

I'm looking to fill up an arrays of zeros with numbers taken from some tuples, easy as that.

Now normally this is not a problem even when the tuples are not the same length (witch is the point here). but it seems it wont compile and I cannot figure out a solution.

from numba import jit    

def cant_jit(ls):

    # Array total lenth
    tl = 6
    # Type
    typ = np.int64

    # Array to modify and return
    start = np.zeros((len(ls), tl), dtype=typ)

    for i in range(len(ls)):

        a = np.array((ls[i]), dtype=typ)
        z = np.zeros((tl - len(ls[i]),), dtype=typ)
        c = np.concatenate((a, z))
        start[i] = c

    return start

# Uneven tuples would be no problem in vanilla
cant_jit(((2, 4), (6, 8, 4)))


jt = jit(cant_jit)    
# working fine
jt(((2, 4), (6, 8)))
# non working
jt(((2, 4), (6, 8, 4)))

Within the error.

getitem(Tuple(UniTuple(int64 x 3), UniTuple(int64 x 2)), int64) There are 22 candidate implementations: - Of which 22 did not match due to: Overload of function 'getitem': File: : Line N/A. With argument(s): '(Tuple(UniTuple(int64 x 3), UniTuple(int64 x 2)), int64)': No match.

I tried some things here to no success. Does someone know a way around this so the function can be compiled and still do it's thing ?

Answer 1

I wonder if numba would be any happier with this non-numpy version:

def foo1(ls):
    res = []
    for row in ls:
        res.append(row+((0,)*(6-len(ls))))
    return res