我有一个2项元组的列表,我想将它们转换为2个列表,其中第一个包含每个元组中的第一项,第二个列表包含第二项。

例如:

original = [('a', 1), ('b', 2), ('c', 3), ('d', 4)]
# and I want to become...
result = (['a', 'b', 'c', 'd'], [1, 2, 3, 4])

有没有内置函数可以做到这一点?


当前回答

因为它返回元组(可能会使用大量内存),所以对我来说,zip(*zipped)技巧似乎更聪明而不是有用。

这是一个函数,它会给出zip的倒数。

def unzip(zipped):
    """Inverse of built-in zip function.
    Args:
        zipped: a list of tuples

    Returns:
        a tuple of lists

    Example:
        a = [1, 2, 3]
        b = [4, 5, 6]
        zipped = list(zip(a, b))

        assert zipped == [(1, 4), (2, 5), (3, 6)]

        unzipped = unzip(zipped)

        assert unzipped == ([1, 2, 3], [4, 5, 6])

    """

    unzipped = ()
    if len(zipped) == 0:
        return unzipped

    dim = len(zipped[0])

    for i in range(dim):
        unzipped = unzipped + ([tup[i] for tup in zipped], )

    return unzipped

其他回答

总结一下:

# data
a = ('a', 'b', 'c', 'd')
b = (1, 2, 3, 4)

# forward
zipped = zip(a, b)  # [('a', 1), ('b', 2), ('c', 3), ('d', 4)]

# reverse
a_, b_ = zip(*zipped)

# verify
assert a == a_
assert b == b_

因为它返回元组(可能会使用大量内存),所以对我来说,zip(*zipped)技巧似乎更聪明而不是有用。

这是一个函数,它会给出zip的倒数。

def unzip(zipped):
    """Inverse of built-in zip function.
    Args:
        zipped: a list of tuples

    Returns:
        a tuple of lists

    Example:
        a = [1, 2, 3]
        b = [4, 5, 6]
        zipped = list(zip(a, b))

        assert zipped == [(1, 4), (2, 5), (3, 6)]

        unzipped = unzip(zipped)

        assert unzipped == ([1, 2, 3], [4, 5, 6])

    """

    unzipped = ()
    if len(zipped) == 0:
        return unzipped

    dim = len(zipped[0])

    for i in range(dim):
        unzipped = unzipped + ([tup[i] for tup in zipped], )

    return unzipped

前面的答案都没有有效地提供所需的输出,即一个由列表组成的元组,而不是由元组组成的列表。对于前者,您可以使用tuple与map。区别在于:

res1 = list(zip(*original))              # [('a', 'b', 'c', 'd'), (1, 2, 3, 4)]
res2 = tuple(map(list, zip(*original)))  # (['a', 'b', 'c', 'd'], [1, 2, 3, 4])

此外,前面的大多数解决方案都假设Python 2.7,其中zip返回一个列表而不是迭代器。

对于Python 3。X时,您需要将结果传递给list或tuple等函数以耗尽迭代器。对于内存效率高的迭代器,可以忽略各自解的外部列表和元组调用。

要获得一个列表的元组,如问题中所示:

>>> original = [('a', 1), ('b', 2), ('c', 3), ('d', 4)]
>>> tuple([list(tup) for tup in zip(*original)])
(['a', 'b', 'c', 'd'], [1, 2, 3, 4])

要将两个列表解压缩为单独的变量:

list1, list2 = [list(tup) for tup in zip(*original)]

天真的方法

def transpose_finite_iterable(iterable):
    return zip(*iterable)  # `itertools.izip` for Python 2 users

适用于有限可迭代对象(例如list/tuple/str等序列)的(可能是无限)可迭代对象,可以说明如下

| |a_00| |a_10| ... |a_n0| |
| |a_01| |a_11| ... |a_n1| |
| |... | |... | ... |... | |
| |a_0i| |a_1i| ... |a_ni| |
| |... | |... | ... |... | |

在哪里

N在ℕ, A_ij对应第i个迭代对象的第j个元素,

在应用transpose_finite_iterable之后,我们得到

| |a_00| |a_01| ... |a_0i| ... |
| |a_10| |a_11| ... |a_1i| ... |
| |... | |... | ... |... | ... |
| |a_n0| |a_n1| ... |a_ni| ... |

Python例子,a_ij == j, n == 2

>>> from itertools import count
>>> iterable = [count(), count()]
>>> result = transpose_finite_iterable(iterable)
>>> next(result)
(0, 0)
>>> next(result)
(1, 1)

但我们不能再次使用transpose_finite_iterable来返回原始iterable的结构,因为result是有限可迭代对象(在我们的例子中是元组)的无限可迭代对象:

>>> transpose_finite_iterable(result)
... hangs ...
Traceback (most recent call last):
  File "...", line 1, in ...
  File "...", line 2, in transpose_finite_iterable
MemoryError

那么我们如何处理这种情况呢?

... 接下来是deque

在我们看了itertools的文档之后。tee函数,有一个Python recipe,经过一些修改可以帮助我们的情况

def transpose_finite_iterables(iterable):
    iterator = iter(iterable)
    try:
        first_elements = next(iterator)
    except StopIteration:
        return ()
    queues = [deque([element])
              for element in first_elements]

    def coordinate(queue):
        while True:
            if not queue:
                try:
                    elements = next(iterator)
                except StopIteration:
                    return
                for sub_queue, element in zip(queues, elements):
                    sub_queue.append(element)
            yield queue.popleft()

    return tuple(map(coordinate, queues))

让我们检查

>>> from itertools import count
>>> iterable = [count(), count()]
>>> result = transpose_finite_iterables(transpose_finite_iterable(iterable))
>>> result
(<generator object transpose_finite_iterables.<locals>.coordinate at ...>, <generator object transpose_finite_iterables.<locals>.coordinate at ...>)
>>> next(result[0])
0
>>> next(result[0])
1

合成

现在,我们可以使用functools定义用于处理可迭代对象的可迭代对象的通用函数,其中一个是有限的,另一个可能是无限的。单分派装饰器

from collections import (abc,
                         deque)
from functools import singledispatch


@singledispatch
def transpose(object_):
    """
    Transposes given object.
    """
    raise TypeError('Unsupported object type: {type}.'
                    .format(type=type))


@transpose.register(abc.Iterable)
def transpose_finite_iterables(object_):
    """
    Transposes given iterable of finite iterables.
    """
    iterator = iter(object_)
    try:
        first_elements = next(iterator)
    except StopIteration:
        return ()
    queues = [deque([element])
              for element in first_elements]

    def coordinate(queue):
        while True:
            if not queue:
                try:
                    elements = next(iterator)
                except StopIteration:
                    return
                for sub_queue, element in zip(queues, elements):
                    sub_queue.append(element)
            yield queue.popleft()

    return tuple(map(coordinate, queues))


def transpose_finite_iterable(object_):
    """
    Transposes given finite iterable of iterables.
    """
    yield from zip(*object_)

try:
    transpose.register(abc.Collection, transpose_finite_iterable)
except AttributeError:
    # Python3.5-
    transpose.register(abc.Mapping, transpose_finite_iterable)
    transpose.register(abc.Sequence, transpose_finite_iterable)
    transpose.register(abc.Set, transpose_finite_iterable)

在有限非空可迭代对象上的二元运算符类中,它可以被认为是它自己的逆(数学家称这种函数为“对合”)。


作为单分派的额外好处,我们可以像这样处理numpy数组

import numpy as np
...
transpose.register(np.ndarray, np.transpose)

然后像这样使用它

>>> array = np.arange(4).reshape((2,2))
>>> array
array([[0, 1],
       [2, 3]])
>>> transpose(array)
array([[0, 2],
       [1, 3]])

Note

由于转置返回迭代器,如果有人想要一个像OP中那样由列表组成的元组——这可以用map内置函数如

>>> original = [('a', 1), ('b', 2), ('c', 3), ('d', 4)]
>>> tuple(map(list, transpose(original)))
(['a', 'b', 'c', 'd'], [1, 2, 3, 4])

广告

我在0.5.0版本的lz包中添加了通用解决方案,可以像这样使用

>>> from lz.transposition import transpose
>>> list(map(tuple, transpose(zip(range(10), range(10, 20)))))
[(0, 1, 2, 3, 4, 5, 6, 7, 8, 9), (10, 11, 12, 13, 14, 15, 16, 17, 18, 19)]

P.S.

没有解决方案(至少是明显的)来处理潜在无限迭代对象的潜在无限迭代对象,但这种情况不太常见。