请问这段代码（为列表中的字典元素去重复）如何能写成列表表达式呢

[pinkman]

>>> t1 = [{'a':1},{'a':1},{'b':2}]
>>> t11 = []
>>> for i in t1:
if i not in t11:
t11.append(i)

[VYSE]

你把{}变成()
t1 = [('a',1),('a',1),('b',2)]
一个dict不久搞定了？
d1 = dict(t1)

[ipconfiger]

[{k:v} for k,v in dict([list(d.iteritems())[0] for d in t1]).iteritems()]

搞定收工

[pinkman]

@ipconfiger 厉害！但看起来好难懂啊，我还是用最简单我原来的方法算了

[saturnisbig]

可以参考stackoverflow上的回答：http://stackoverflow.com/questions/9427163/remove-duplicate-dict-in-list-in-python

[ipconfiger]

@pinkman 随手写的写复杂了点
[{k:v} for k,v in dict([d.items()[0] for d in t1]).items()]
简化后就这样了，其实dict([d.items()[0] for d in t1]).items() 以外的部分纯粹是为了把列表内还原成一个个字典

[jokaye]

[t11.append(t) for t in t1 if t not in t11] ?

[pinkman]

@jokaye 原来如此，还是需要先生成个t11 = []的空字典，随后的list表达式就好写了，之前我就一直在想不自己先定义t11（因为考虑到列表表达式本来就要生成列表），所以才老是想不出来，现在清楚了，感谢jokaye

[mengzhuo]

[t11.append(t) for t in l if t not in t11]
1000000 loops, best of 3: 793 ns per loop
# 不过话说这么个复杂度是2n吧

[dict(t) for t in set([tuple(d.items()) for d in l])]
100000 loops, best of 3: 9.25 us per loop

[{k:v} for k,v in dict([list(d.iteritems())[0] for d in t1]).iteritems()]
100000 loops, best of 3: 10.4 us per loop

[{k:v} for k,v in dict([d.items()[0] for d in t1]).items()]
100000 loops, best of 3: 5.01 us per loop
# 这个很漂亮

---------------------------

升级一下测试：
t2 = [{random.randint(0, 4095):1} for k in xrange(9999)]
len([d for d in t2 if t2.count(d) > 1]) # 这次随机出9144项重复的

[t22.append(t) for t in t2 if t not in t22]
1 loops, best of 3: 1.54 s per loop
# 直接跪了 = =

[dict(t) for t in set([tuple(d.items()) for d in t2])]
10 loops, best of 3: 21.9 ms per loop

[{k:v} for k,v in dict([list(d.iteritems())[0] for d in t2]).iteritems()]
10 loops, best of 3: 26 ms per loop

[{k:v} for k,v in dict([d.items()[0] for d in t2]).items()]
100 loops, best of 3: 11.4 ms per loop
# 用iter，会更好
[{k:v} for k,v in dict([d.items()[0] for d in t2]).iteritems()]
100 loops, best of 3: 10.7 ms per loop

[mengzhuo]

我觉得黑魔法了的一种方法：
[{k:v} for k,v in dict([d.iteritems().next() for d in t2]).iteritems()]
100 loops, best of 3: 10.4 ms per loop

不过性能没啥提升了，还是@ipconfiger 的iter版给力了