foo.py

    def set_dims(parts):
        assert(len(parts) is 2, "Expected dimensions specification to have exactly two components")
        assert(all([isinstance(p, int) and p > 0 for p in parts]),
               "Expected all components in dimensions to be positive integers")
        options.update({
            "imageWidth": parts[0],
            "imageHeight": parts[1]
        })

    if raster_dimensions is not None:
        if isinstance(raster_dimensions, tuple):
            set_dims(raster_dimensions)
        elif isinstance(raster_dimensions, str):
            p = list(map(lambda s: int(s), raster_dimensions.split(',')))
            set_dims(p)

At line 2 I would probably prefer an equality test rather than is. Not exactly sure why.

at line 3 we could also attempt coercion to int.

try:
    parts = [int(p) for p in parts]
    assert(all([p > 0 for p in parts]), 'nice message')
except ValueError:
    raise
    # or some nicer errror handling?

At line 10 also check for list : if isinstance(raster_dimensions, (list, tuple))

If you take my suggestion above about doing int coercion within set_dims function then you don't need to do it at 14. If you still want to do it, i think more readable to do anotehr list comprehension at 14 instead of list(map(...) . say [int(s) for s in raster_dimensions.split(',')]

At line 2 I would probably prefer an equality test rather than is. Not exactly sure why.

I honestly don't know the difference.

IDK but maybe this seems correct

is will return True if two variables point to the same object, == if the objects referred to by the variables are equal.

And as I recall there is some interesting things like if i say a=2 and b=2 then a is b is true because both objects will point ot the same underlying object. I think.