As an example. Lets say I have a grid-based game, where the player agent can walk to all squares around the current position, but not stay still.
Thus we could see each move as a tuple (dx,dy), with the total set of moves being
[(1,1), (1,0), (1,-1), (0,1), (0,-1), (-1,1), (-1,0), (-1,-1)]
Staying still would be (dx,dy) = (0,0).
With this problem in mind, one could make use of the cartesian product of [1,0,-1] with itself, which in Python would be itertools.product(iterables*). However, this would yield all those tuples listed above, including (0,0)
One solution is to convert it to a list, and then use remove() on the list:
steps = list(itertools.product([-1,0,1],[-1,0,1])).remove((0,0))
or alternatively [x for x in itertools.product([1,0,-1], [1,0,-1]) if x != (0,0)]. Both of these solutions require conversion to lists.
What's your cleanest way to yield and exclude (0,0)?
Generally speaking you would define another iterator based on the first one and apply a condition in a comprehension.
In this case:
moves = ( p for p in itertools.product((-1,0,1),repeat=2) if p != (0,0) )
Here moves is a new generator that will filter out (0,0) from the output of itertools.product. Notice that it is enclosed in parentheses, not square brackets which would make it a list instead of an iterator