How can I make an object with an interface like a random number generator, but that actually generates a specified sequence?

I'd like to construct an object that works like a random number generator, but generates numbers in a specified sequence.

# a random number generator
rng = lambda : np.random.randint(2,20)//2

# a non-random number generator
def nrng():
    numbers = np.arange(1,10.5,0.5)
    for i in range(len(numbers)):
        yield numbers[i]

for j in range(10):
    print('random number', rng())
    print('non-random number', nrng())

The issue with the code above that I cannot call nrng in the last line because it is a generator. I know that the most straightforward way to rewrite the code above is to simply loop over the non-random numbers instead of defining the generator. I would prefer getting the example above to work because I am working with a large chunk of code that include a function that accepts a random number generator as an argument, and I would like to add the functionality to pass non-random number sequences without rewriting the entire code.

EDIT: I see some confusion in the comments. I am aware that python's random number generators generate pseudo-random numbers. This post is about replacing a pseudo-random-number generator by a number generator that generates numbers from a non-random, user-specified sequence (e.g., a generator that generates the number sequence 1,1,2,2,1,0,1 if I want it to).

Solution

Edit:

The cleanest way to do this would be to use a lambda to wrap your call to next(nrng) as per great comment from @GACy20:

def nrng_gen():
    yield from range(10)

nrng = nrng_gen()

nrng_func = lambda: next(nrng)

for i in range(10):
    print(nrng_func())

Original answer:

If you want your object to keep state and look like a function, create a custom class with __call__ method.

eg.

class NRNG:
    def __init__(self):
        self.numbers = range(10)
        self.state = -1
    def __call__(self):
        self.state += 1
        return self.numbers[self.state]
        
nrng = NRNG()


for i in range(10):
    print(nrng())

However, I wouldn't recommend this unless absolutely necessary, as it obscures the fact that your nrng keeps a state (although technically, most rngs keep their state internally).

It's best to just use a regular generator with yield by calling next on it or to write a custom iterator (also class-based). Those will work with things like for loops and other python tools for iteration (like the excellent itertools package).