>>> L1 = [[1, 2], [3, 4], [5, 6]]
>>> L2 =[["a", "b"], ["c", "d"], ["e", "f"]]
>>> [x + y for x,y in zip(L1,L2)]
[[1, 2, 'a', 'b'], [3, 4, 'c', 'd'], [5, 6, 'e', 'f']]
或者,
>>> [sum(x,[]) for x in zip(L1,L2)]
[[1, 2, 'a', 'b'], [3, 4, 'c', 'd'], [5, 6, 'e', 'f']]
或者,
>>> import itertools
>>> [list(itertools.chain(*x)) for x in zip(L1,L2)]
[[1, 2, 'a', 'b'], [3, 4, 'c', 'd'], [5, 6, 'e', 'f']]
我们也可以不用zip():
>>> [L1[i] + L2[i] for i in xrange(min(len(L1), len(L2)))]
[[1, 2, 'a', 'b'], [3, 4, 'c', 'd'], [5, 6, 'e', 'f']]
>>> [x + L2[i] for i, x in enumerate(L1)] # assuming len(L1) == len(l2)
[[1, 2, 'a', 'b'], [3, 4, 'c', 'd'], [5, 6, 'e', 'f']]
>>> # same as above, but deals with different lengths
>>> Lx, Ly = ((L2,L1), (L1,L2))[len(L1)<=len(L2)] # shortcut for if/else
>>> [x + Ly[i] for i, x in enumerate(Lx)]
一些基准测试
以下是迄今为止提供的答案的一些基准。
看起来最受欢迎的答案 ([x + y for x,y in zip(L1,L2)]) 与 @hammar's map solution 差不多。 另一方面,我给出的替代解决方案已被证明是垃圾!
然而,最快的解决方案(目前)似乎是那些使用列表解析而不使用 zip() 的解决方案。
[me@home]$ SETUP="L1=[[x,x+1] for x in xrange(10000)];L2=[[x+2,x+3] for x in xrange(10000)]"
[me@home]$ # this raises IndexError if len(L1) > len(L2)
[me@home]$ python -m timeit "$SETUP" "[x + L2[i] for i, x in enumerate(L1)]"
100 loops, best of 3: 10.6 msec per loop
[me@home]$ # same as above, but deals with length inconsistencies
[me@home]$ python -m timeit "$SETUP" "Lx,Ly=((L2,L1),(L1,L2))[len(L1)<=len(L2)];[x + Ly[i] for i, x in enumerate(Lx)]"
100 loops, best of 3: 10.6 msec per loop
[me@home]$ # almost as fast as above, but easier to read
[me@home]$ python -m timeit "$SETUP" "[L1[i] + L2[i] for i in xrange(min(len(L1),len(L2)))]"
100 loops, best of 3: 10.8 msec per loop
[me@home]$ python -m timeit "$SETUP" "L3=[x + y for x,y in zip(L1,L2)]"
100 loops, best of 3: 13.4 msec per loop
[me@home]$ python -m timeit "$SETUP" "L3=map(list.__add__, L1, L2)"
100 loops, best of 3: 13.5 msec per loop
[me@home]$ python -m timeit "$SETUP" "L3=[sum(x,[]) for x in zip(L1,L2)]"
100 loops, best of 3: 18.1 msec per loop
[me@home]$ python -m timeit "$SETUP;import itertools" "L3=[list(itertools.chain(*x)) for x in zip(L1,L2)]"
10 loops, best of 3: 32.9 msec per loop
@Zac's suggestion 确实很快,但是我们在这里比较的是苹果和橙子,因为它在L1 上就地 进行了列表扩展,而不是创建第三个列表。因此,如果不再需要 L1,这是一个很好的解决方案。
[me@home]$ python -m timeit "$SETUP" "for index, x in enumerate(L1): x.extend(L2[index])"
100 loops, best of 3: 9.46 msec per loop
但是,如果 L1 必须保持原样,那么一旦包含 deepcopy,性能就会低于标准。
[me@home]$ python -m timeit "$SETUP;from copy import deepcopy" "L3=deepcopy(L1)
> for index, x in enumerate(L1): x.extend(L2[index])"
10 loops, best of 3: 116 msec per loop