如何在整数列表中找到重复项并创建重复项的另一个列表?


当前回答

还有其他测试。当然要做……

set([x for x in l if l.count(x) > 1])

...代价太大了。使用下一个final方法大约快500倍(数组越长结果越好):

def dups_count_dict(l):
    d = {}

    for item in l:
        if item not in d:
            d[item] = 0

        d[item] += 1

    result_d = {key: val for key, val in d.iteritems() if val > 1}

    return result_d.keys()

只有2个循环,没有非常昂贵的l.count()操作。

下面是一个比较方法的代码。代码如下,输出如下:

dups_count: 13.368s # this is a function which uses l.count()
dups_count_dict: 0.014s # this is a final best function (of the 3 functions)
dups_count_counter: 0.024s # collections.Counter

测试代码:

import numpy as np
from time import time
from collections import Counter

class TimerCounter(object):
    def __init__(self):
        self._time_sum = 0

    def start(self):
        self.time = time()

    def stop(self):
        self._time_sum += time() - self.time

    def get_time_sum(self):
        return self._time_sum


def dups_count(l):
    return set([x for x in l if l.count(x) > 1])


def dups_count_dict(l):
    d = {}

    for item in l:
        if item not in d:
            d[item] = 0

        d[item] += 1

    result_d = {key: val for key, val in d.iteritems() if val > 1}

    return result_d.keys()


def dups_counter(l):
    counter = Counter(l)    

    result_d = {key: val for key, val in counter.iteritems() if val > 1}

    return result_d.keys()



def gen_array():
    np.random.seed(17)
    return list(np.random.randint(0, 5000, 10000))


def assert_equal_results(*results):
    primary_result = results[0]
    other_results = results[1:]

    for other_result in other_results:
        assert set(primary_result) == set(other_result) and len(primary_result) == len(other_result)


if __name__ == '__main__':
    dups_count_time = TimerCounter()
    dups_count_dict_time = TimerCounter()
    dups_count_counter = TimerCounter()

    l = gen_array()

    for i in range(3):
        dups_count_time.start()
        result1 = dups_count(l)
        dups_count_time.stop()

        dups_count_dict_time.start()
        result2 = dups_count_dict(l)
        dups_count_dict_time.stop()

        dups_count_counter.start()
        result3 = dups_counter(l)
        dups_count_counter.stop()

        assert_equal_results(result1, result2, result3)

    print 'dups_count: %.3f' % dups_count_time.get_time_sum()
    print 'dups_count_dict: %.3f' % dups_count_dict_time.get_time_sum()
    print 'dups_count_counter: %.3f' % dups_count_counter.get_time_sum()

其他回答

我在寻找相关的东西时遇到了这个问题-想知道为什么没有人提供基于生成器的解决方案?解决这个问题的方法是:

>>> print list(getDupes_9([1,2,3,2,1,5,6,5,5,5]))
[1, 2, 5]

我很关心可伸缩性,所以测试了几种方法,包括在小列表上工作得很好的naive项,但随着列表变大,可伸缩性很差(注意-使用timeit会更好,但这只是说明)。

我加入了@moooeeeep作为比较(它的速度非常快:如果输入列表是完全随机的,速度最快)和itertools方法,对于大多数排序的列表,它甚至更快……现在包括来自@ fireynx的熊猫方法-缓慢,但不是可怕的,而且简单。注意:在我的机器上,sort/tee/zip方法对于大型有序列表始终是最快的,moooeeeep对于洗牌列表是最快的,但您的情况可能会有所不同。

优势

非常快速简单的测试'任何'重复使用相同的代码

假设

重复项只应报告一次 重复的订单不需要保留 Duplicate可能位于列表中的任何位置


最快的解决方案,1m条目:

def getDupes(c):
        '''sort/tee/izip'''
        a, b = itertools.tee(sorted(c))
        next(b, None)
        r = None
        for k, g in itertools.izip(a, b):
            if k != g: continue
            if k != r:
                yield k
                r = k

方法测试

import itertools
import time
import random

def getDupes_1(c):
    '''naive'''
    for i in xrange(0, len(c)):
        if c[i] in c[:i]:
            yield c[i]

def getDupes_2(c):
    '''set len change'''
    s = set()
    for i in c:
        l = len(s)
        s.add(i)
        if len(s) == l:
            yield i

def getDupes_3(c):
    '''in dict'''
    d = {}
    for i in c:
        if i in d:
            if d[i]:
                yield i
                d[i] = False
        else:
            d[i] = True

def getDupes_4(c):
    '''in set'''
    s,r = set(),set()
    for i in c:
        if i not in s:
            s.add(i)
        elif i not in r:
            r.add(i)
            yield i

def getDupes_5(c):
    '''sort/adjacent'''
    c = sorted(c)
    r = None
    for i in xrange(1, len(c)):
        if c[i] == c[i - 1]:
            if c[i] != r:
                yield c[i]
                r = c[i]

def getDupes_6(c):
    '''sort/groupby'''
    def multiple(x):
        try:
            x.next()
            x.next()
            return True
        except:
            return False
    for k, g in itertools.ifilter(lambda x: multiple(x[1]), itertools.groupby(sorted(c))):
        yield k

def getDupes_7(c):
    '''sort/zip'''
    c = sorted(c)
    r = None
    for k, g in zip(c[:-1],c[1:]):
        if k == g:
            if k != r:
                yield k
                r = k

def getDupes_8(c):
    '''sort/izip'''
    c = sorted(c)
    r = None
    for k, g in itertools.izip(c[:-1],c[1:]):
        if k == g:
            if k != r:
                yield k
                r = k

def getDupes_9(c):
    '''sort/tee/izip'''
    a, b = itertools.tee(sorted(c))
    next(b, None)
    r = None
    for k, g in itertools.izip(a, b):
        if k != g: continue
        if k != r:
            yield k
            r = k

def getDupes_a(l):
    '''moooeeeep'''
    seen = set()
    seen_add = seen.add
    # adds all elements it doesn't know yet to seen and all other to seen_twice
    for x in l:
        if x in seen or seen_add(x):
            yield x

def getDupes_b(x):
    '''iter*/sorted'''
    x = sorted(x)
    def _matches():
        for k,g in itertools.izip(x[:-1],x[1:]):
            if k == g:
                yield k
    for k, n in itertools.groupby(_matches()):
        yield k

def getDupes_c(a):
    '''pandas'''
    import pandas as pd
    vc = pd.Series(a).value_counts()
    i = vc[vc > 1].index
    for _ in i:
        yield _

def hasDupes(fn,c):
    try:
        if fn(c).next(): return True    # Found a dupe
    except StopIteration:
        pass
    return False

def getDupes(fn,c):
    return list(fn(c))

STABLE = True
if STABLE:
    print 'Finding FIRST then ALL duplicates, single dupe of "nth" placed element in 1m element array'
else:
    print 'Finding FIRST then ALL duplicates, single dupe of "n" included in randomised 1m element array'
for location in (50,250000,500000,750000,999999):
    for test in (getDupes_2, getDupes_3, getDupes_4, getDupes_5, getDupes_6,
                 getDupes_8, getDupes_9, getDupes_a, getDupes_b, getDupes_c):
        print 'Test %-15s:%10d - '%(test.__doc__ or test.__name__,location),
        deltas = []
        for FIRST in (True,False):
            for i in xrange(0, 5):
                c = range(0,1000000)
                if STABLE:
                    c[0] = location
                else:
                    c.append(location)
                    random.shuffle(c)
                start = time.time()
                if FIRST:
                    print '.' if location == test(c).next() else '!',
                else:
                    print '.' if [location] == list(test(c)) else '!',
                deltas.append(time.time()-start)
            print ' -- %0.3f  '%(sum(deltas)/len(deltas)),
        print
    print

“all dupes”测试的结果是一致的,在这个数组中找到“first”重复,然后是“all”重复:

Finding FIRST then ALL duplicates, single dupe of "nth" placed element in 1m element array
Test set len change :    500000 -  . . . . .  -- 0.264   . . . . .  -- 0.402  
Test in dict        :    500000 -  . . . . .  -- 0.163   . . . . .  -- 0.250  
Test in set         :    500000 -  . . . . .  -- 0.163   . . . . .  -- 0.249  
Test sort/adjacent  :    500000 -  . . . . .  -- 0.159   . . . . .  -- 0.229  
Test sort/groupby   :    500000 -  . . . . .  -- 0.860   . . . . .  -- 1.286  
Test sort/izip      :    500000 -  . . . . .  -- 0.165   . . . . .  -- 0.229  
Test sort/tee/izip  :    500000 -  . . . . .  -- 0.145   . . . . .  -- 0.206  *
Test moooeeeep      :    500000 -  . . . . .  -- 0.149   . . . . .  -- 0.232  
Test iter*/sorted   :    500000 -  . . . . .  -- 0.160   . . . . .  -- 0.221  
Test pandas         :    500000 -  . . . . .  -- 0.493   . . . . .  -- 0.499  

当列表首先被打乱时,排序的代价就变得明显了——效率显著下降,@moooeeeep方法占主导地位,set和dict方法类似,但性能较差:

Finding FIRST then ALL duplicates, single dupe of "n" included in randomised 1m element array
Test set len change :    500000 -  . . . . .  -- 0.321   . . . . .  -- 0.473  
Test in dict        :    500000 -  . . . . .  -- 0.285   . . . . .  -- 0.360  
Test in set         :    500000 -  . . . . .  -- 0.309   . . . . .  -- 0.365  
Test sort/adjacent  :    500000 -  . . . . .  -- 0.756   . . . . .  -- 0.823  
Test sort/groupby   :    500000 -  . . . . .  -- 1.459   . . . . .  -- 1.896  
Test sort/izip      :    500000 -  . . . . .  -- 0.786   . . . . .  -- 0.845  
Test sort/tee/izip  :    500000 -  . . . . .  -- 0.743   . . . . .  -- 0.804  
Test moooeeeep      :    500000 -  . . . . .  -- 0.234   . . . . .  -- 0.311  *
Test iter*/sorted   :    500000 -  . . . . .  -- 0.776   . . . . .  -- 0.840  
Test pandas         :    500000 -  . . . . .  -- 0.539   . . . . .  -- 0.540  

试试这个检查副本

>>> def checkDuplicate(List):
    duplicate={}
    for i in List:
            ## checking whether the item is already present in dictionary or not
            ## increasing count if present
            ## initializing count to 1 if not present

        duplicate[i]=duplicate.get(i,0)+1

    return [k for k,v in duplicate.items() if v>1]

>>> checkDuplicate([1,2,3,"s",1,2,3])
[1, 2, 3]

你可以使用iteration_utilities.duplicate:

>>> from iteration_utilities import duplicates

>>> list(duplicates([1,1,2,1,2,3,4,2]))
[1, 1, 2, 2]

或者如果你只想要一个副本,可以结合iteration_utilities.unique_everseen:

>>> from iteration_utilities import unique_everseen

>>> list(unique_everseen(duplicates([1,1,2,1,2,3,4,2])))
[1, 2]

它也可以处理不可哈希的元素(但是以性能为代价):

>>> list(duplicates([[1], [2], [1], [3], [1]]))
[[1], [1]]

>>> list(unique_everseen(duplicates([[1], [2], [1], [3], [1]])))
[[1]]

这是只有少数其他方法可以处理的问题。

基准

我做了一个快速的基准测试,其中包含了这里提到的大部分(但不是全部)方法。

第一个基准测试只包含了很小范围的列表长度,因为一些方法具有O(n**2)行为。

在图表中,y轴代表时间,所以值越低越好。它还绘制了log-log,以便更好地可视化广泛的值范围:

除去O(n**2)方法,我在一个列表中做了另一个多达50万个元素的基准测试:

正如您所看到的iteration_utilities。duplicate方法比任何其他方法都快,甚至连链接unique_everseen(duplicate(…))也比其他方法更快或同样快。

这里需要注意的另一件有趣的事情是,熊猫方法对于小列表非常慢,但可以轻松地竞争较长的列表。

然而,由于这些基准测试显示大多数方法的性能大致相同,因此使用哪一种并不重要(除了有O(n**2)运行时的3种方法)。

from iteration_utilities import duplicates, unique_everseen
from collections import Counter
import pandas as pd
import itertools

def georg_counter(it):
    return [item for item, count in Counter(it).items() if count > 1]

def georg_set(it):
    seen = set()
    uniq = []
    for x in it:
        if x not in seen:
            uniq.append(x)
            seen.add(x)

def georg_set2(it):
    seen = set()
    return [x for x in it if x not in seen and not seen.add(x)]   

def georg_set3(it):
    seen = {}
    dupes = []

    for x in it:
        if x not in seen:
            seen[x] = 1
        else:
            if seen[x] == 1:
                dupes.append(x)
            seen[x] += 1

def RiteshKumar_count(l):
    return set([x for x in l if l.count(x) > 1])

def moooeeeep(seq):
    seen = set()
    seen_add = seen.add
    # adds all elements it doesn't know yet to seen and all other to seen_twice
    seen_twice = set( x for x in seq if x in seen or seen_add(x) )
    # turn the set into a list (as requested)
    return list( seen_twice )

def F1Rumors_implementation(c):
    a, b = itertools.tee(sorted(c))
    next(b, None)
    r = None
    for k, g in zip(a, b):
        if k != g: continue
        if k != r:
            yield k
            r = k

def F1Rumors(c):
    return list(F1Rumors_implementation(c))

def Edward(a):
    d = {}
    for elem in a:
        if elem in d:
            d[elem] += 1
        else:
            d[elem] = 1
    return [x for x, y in d.items() if y > 1]

def wordsmith(a):
    return pd.Series(a)[pd.Series(a).duplicated()].values

def NikhilPrabhu(li):
    li = li.copy()
    for x in set(li):
        li.remove(x)

    return list(set(li))

def firelynx(a):
    vc = pd.Series(a).value_counts()
    return vc[vc > 1].index.tolist()

def HenryDev(myList):
    newList = set()

    for i in myList:
        if myList.count(i) >= 2:
            newList.add(i)

    return list(newList)

def yota(number_lst):
    seen_set = set()
    duplicate_set = set(x for x in number_lst if x in seen_set or seen_set.add(x))
    return seen_set - duplicate_set

def IgorVishnevskiy(l):
    s=set(l)
    d=[]
    for x in l:
        if x in s:
            s.remove(x)
        else:
            d.append(x)
    return d

def it_duplicates(l):
    return list(duplicates(l))

def it_unique_duplicates(l):
    return list(unique_everseen(duplicates(l)))

基准1

from simple_benchmark import benchmark
import random

funcs = [
    georg_counter, georg_set, georg_set2, georg_set3, RiteshKumar_count, moooeeeep, 
    F1Rumors, Edward, wordsmith, NikhilPrabhu, firelynx,
    HenryDev, yota, IgorVishnevskiy, it_duplicates, it_unique_duplicates
]

args = {2**i: [random.randint(0, 2**(i-1)) for _ in range(2**i)] for i in range(2, 12)}

b = benchmark(funcs, args, 'list size')

b.plot()

基准2

funcs = [
    georg_counter, georg_set, georg_set2, georg_set3, moooeeeep, 
    F1Rumors, Edward, wordsmith, firelynx,
    yota, IgorVishnevskiy, it_duplicates, it_unique_duplicates
]

args = {2**i: [random.randint(0, 2**(i-1)) for _ in range(2**i)] for i in range(2, 20)}

b = benchmark(funcs, args, 'list size')
b.plot()

免责声明

1这是我写的一个第三方库:iteration_utilities。

假设我们有这个元素列表:

a = [1, 2, 3, 2, 1, 5, 6, 5, 5, 5]

我们可以只使用集合来找到唯一的元素:

unique = set()
for num in a:
    if num not in unique:
        unique.add(num)
    else:
        unique = unique - set([num])

最后:

>>> unique
{3, 6}

如果你想要得到副本,你可以简单地做:

>>> duplicates = set(a) - unique
>>> duplicates
{1, 2, 5}

注:

集合中的元素查找是O(1) 从集合中移除的元素是O(1)

不需要转换为列表,可能最简单的方法是如下所示。 在面试中,当他们要求不要使用集合时,这可能会很有用

a=[1,2,3,3,3]
dup=[]
for each in a:
  if each not in dup:
    dup.append(each)
print(dup)

======= else获取唯一值和重复值的2个单独列表

a=[1,2,3,3,3]
uniques=[]
dups=[]

for each in a:
  if each not in uniques:
    uniques.append(each)
  else:
    dups.append(each)
print("Unique values are below:")
print(uniques)
print("Duplicate values are below:")
print(dups)