Question

I have a sequence of generators: (gen_0, gen_1, … gen_n)

These generators will create their values lazily but are finite and will have potentially different lengths.

I need to be able to construct another generator that yields the first element of each generator in order, followed by the second and so forth, skipping values from generators that have been exhausted.

I think this problem is analogous to taking the tuple

((1, 4, 7, 10, 13, 16), (2, 5, 8, 11, 14), (3, 6, 9, 12, 15, 17, 18))

and traversing it so that it would yield the numbers from 1 through 18 in order.

I’m working on solving this simple example using (genA, genB, genC) with genA yielding values from (1, 4, 7, 10, 13, 16), genB yielding (2, 5, 8, 11, 14) and genC yielding (3, 6, 9, 12, 15, 17, 18).

To solve the simpler problem with the tuple of tuples the answer is fairly simple if the
elements of the tuple were the same length. If the variable ‘a’ referred to the tuple, you could use

[i for t in zip(*a) for i in t]

Unfortunately the items are not necessarily the same length and the zip trick doesn’t seem to work for generators anyway.

So far my code is horribly ugly and I’m failing to find anything approaching a clean solution. Help?

Answer 1

You might consider itertools.izip_longest, but in case None is a valid value, that solution will fail. Here is a sample “another generator”, which does exactly what you asked for, and is pretty clean:

def my_gen(generators):
    while True:
        rez = () 
        for gen in generators:
            try:
                rez = rez + (gen.next(),)
            except StopIteration:
                pass
        if rez:
            yield rez
        else:
            break

print [x for x in my_gen((iter(xrange(2)), iter(xrange(3)), iter(xrange(1))))]

[(0, 0, 0), (1, 1), (2,)] #output