在键上加入python迭代器

Question

I'm looking for a way to join python iterators like itertools.izip_longest() does, but I would like to join elements that have the same "key" (as defined by a parameter) and output None when the key does not exist on all iterators. I'm assuming iterators sorted ascending by "key".

Example:

iter1 = iter((1, 3, 4, 9))
iter2 = iter((3, 5, 6))
iter3 = iter((1, 3, 10))

zipjoiner(iter1, iter2, iter3)

should give:

iter(((1, None, 1), (3, 3, 3), (4, None, None), (None, 5, None), (None, 6, None), (9, None, None), (None, None, 10)))

(in this case key is the default identity lambda x: x )

I've tried to modify the izip_longest() implementation as found in python documentation and it works (at least on my example), but I'm looking for a more elegant solution. Any idea?

This is my code:

def zipjoiner(*args, **kwds):
    # izip_longest('ABCD', 'xy', fillvalue='-') --> Ax By C- D-
    fillvalue = kwds.get('fillvalue')
    key = kwds.get('key', lambda x: x)
    counter = [len(args) - 1]
    def sentinel():
        if not counter[0]:
            raise ZipExhausted
        counter[0] -= 1
        yield fillvalue
    fillers = itertools.repeat(fillvalue)
    iterators = [itertools.chain(it, sentinel(), fillers) for it in args]

    def getkey(x):
        return None if x is None else key(x)

    try:
        while iterators:
            elements = tuple(map(next, iterators))
            keys = tuple(map(getkey, elements))
            minkey = min(_ for _ in keys if not _ is None)
            while not all(k == minkey for k in keys):
                yield tuple(map(lambda (k, v): v if k == minkey else None, zip(keys, elements)))
                elements = tuple(map(lambda (k, it, v): it.next() if k == minkey else v, zip(keys, iterators, elements)))
                keys = tuple(map(getkey, elements))
                minkey = min(_ for _ in keys if not _ is None)
            yield elements

    except ZipExhausted:
        pass

Answer 1

In case you don't want to preserve order you could turn your lists into sets, then loop through the sorted list of all values from all input iterators and yield a tuple with values or None depending on whether the value is in the corresponding set:

def join_iterators(*iterators):
    sets = []
    for iterator in iterators:
        sets.append(set(iterator))

    values = set(itertools.chain(*iterators))
    get_value_or_none = lambda value, s: value if value in s else None
    for value in sorted(values):
        yield tuple(get_value_or_none(value, s) for s in sets)

This does not address your key function but I think you figure out how to apply that ;)

Answer 2

While you could make something that was more readable, I don't think you could make something that was more efficient.

From the readability standpoint I would change the header and first few lines:

def zipjoiner(*iters, fillvalue=None, key=lambda x: x):
    # drop first two lines dealing with fillvalue and key

as there is no reason to deal with **kwds .