How to delete all instances of a repeated number in a list? [duplicate]

Question

I want a code that deletes all instances of any number that has been repeated from a list.

E.g.:

Inputlist = [2, 3, 6, 6, 8, 9, 12, 12, 14]
 
Outputlist = [2,3,8,9,14]

I have tried to remove the duplicated elements in the list already (by using the "unique" function), but it leaves a single instance of the element in the list nevertheless!

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

I went through a lot of StackOverflow articles too, but all of them differ in the idea that they are searching for removing common elements from two different lists, or that they want just one instance of each element to still be kept. I want to simply remove all common elements.

Answer 1

You can use a Counter

>>> from collections import Counter
>>> l = [2, 3, 6, 6, 8, 9, 12, 12, 14]
>>> res = [el for el, cnt in Counter(l).items() if cnt==1]
>>> res
[2, 3, 8, 9, 14]

Answer 2

You can always have two sets. One to check if seen and another one to keep unique only. set.discard(el) will remove if exists.

Inputlist = [2, 3, 6, 6, 8, 9, 12, 12, 14]

seen = set()
ans = set()

for el in Inputlist:
    if el not in seen:
        seen.add(el)
        ans.add(el)
    else:
        ans.discard(el)

print(list(ans))

EDIT: for giggles I measured the performance of these two solutions

from timeit import timeit


first = """
def get_from_two_sets():
    seen = set()
    ans = set()

    for el in (2, 3, 6, 6, 8, 9, 12, 12, 14):
        if el not in seen:
            seen.add(el)
            ans.add(el)
        else:
            ans.discard(el)"""


second = """

def get_from_counter():
    return [el for el, cnt in Counter((2, 3, 6, 6, 8, 9, 12, 12, 14)).items() if cnt == 1]
    """


print(timeit(stmt=first, number=10000000))
print(timeit(stmt=second, number=10000000, setup="from collections import Counter"))

yields

0.3130729760000577
0.46127468299982866

so yay! it seems like my solution is slightly faster. Don't waste those nanoseconds you saved!

@abc solution is clean and pythonic, go for it.

Answer 3

A simple list comprehension will do the trick:

Inputlist = [2, 3, 6, 6, 8, 9, 12, 12, 14]
 
Outputlist = [item for item in Inputlist if Inputlist.count(item) == 1]

Answer 4

Alternate solution for case where only consecutive duplicates should be removed:

from itertools import groupby

inputlist = [2, 3, 6, 6, 8, 9, 12, 12, 14]

outputlist = [x for _, (x, *extra) in groupby(inputlist) if not extra]

All this does is group together runs of identical values, unpack the first copy into x, and the rest enter a list; we check if said list is empty to determine whether there was just one value, or more than one, and only keep the ones where it was a single value.

If you don't like even the temporary extra list, using one of the ilen solutions that doesn't listify the group would allow a similar solution with no unbounded temporary storage:

outputlist = [x for x, grp in groupby(inputlist) if ilen(grp) == 1]

Or with a helper that just checks "at least 2" without iterating beyond that point:

def more_than_one(it):
    next(it)  # Assumes at least once, which is already the case with groupby groups
    try:
        next(it)
    except StopIteration:
        return True
    return False

outputlist = [x for x, grp in groupby(inputlist) if not more_than_one(grp)]

Note: I'd actually prefer abc's Counter-based solution in general, but if you actually want to only delete adjacent duplicates, it's not adequate to the task.

Answer 5

Another solution using sets: Convert the input list to a set and remove all elements of this set from the input list. This leaves only duplicates in the list. Now convert this to a set and you can subtract one set from another. Sounds complicated, but is quite short and efficient for short lists:

l = [2, 3, 6, 6, 8, 9, 12, 12, 14]
inset = set(l)

for i in inset:   # <-- usually the element to remove is in the front,
    l.remove(i)   # <-- but in a worst case, this is slower than O(n)

result = list(inset - set(l))

irrelevant performance for the short example list:

# %timeit this solution
1.18 µs ± 1.97 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)
# %timeit solution with seen-set
1.23 µs ± 1.49 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)
# %timeit solution with Counter class
2.76 µs ± 4.85 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)

For a list with 1000 elements and 10% duplicates the Counter-solution is fastest!

Answer 6

If input is sorted and can be bounded by a min and a max, this can be done in O(n):

min = -1
max = 99999999  # put whatever you need
J = [min] + I + [max]
[y for (x,y,z) in zip(J, J[1:], J[2:]) if x < y and y < z]