Have a look at itertools.combinations:

itertools.combinations(iterable, r)

Return r length subsequences of elements from the input iterable.

Combinations are emitted in lexicographic sort order. So, if the input iterable is sorted, the combination tuples will be produced in sorted order.

Since 2.6, batteries are included!

This is an approach that can be easily transfered to all programming languages supporting recursion (no itertools, no yield, no list comprehension):

def combs(a):
    if len(a) == 0:
        return [[]]
    cs = []
    for c in combs(a[1:]):
        cs += [c, c+[a[0]]]
    return cs

>>> combs([1,2,3,4,5])
[[], [1], [2], [2, 1], [3], [3, 1], [3, 2], ..., [5, 4, 3, 2, 1]]

3 functions:

1. all combinations of n elements list
2. all combinations of n elements list where order is not distinct
3. all permutations

import sys

def permutations(a):
    return combinations(a, len(a))

def combinations(a, n):
    if n == 1:
        for x in a:
            yield [x]
    else:
        for i in range(len(a)):
            for x in combinations(a[:i] + a[i+1:], n-1):
                yield [a[i]] + x

def combinationsNoOrder(a, n):
    if n == 1:
        for x in a:
            yield [x]
    else:
        for i in range(len(a)):
            for x in combinationsNoOrder(a[:i], n-1):
                yield [a[i]] + x
    
if __name__ == "__main__":
    for s in combinations(list(map(int, sys.argv[2:])), int(sys.argv[1])):
        print(s)

from itertools import combinations


features = ['A', 'B', 'C']
tmp = []
for i in range(len(features)):
    oc = combinations(features, i + 1)
    for c in oc:
        tmp.append(list(c))

output

[
 ['A'],
 ['B'],
 ['C'],
 ['A', 'B'],
 ['A', 'C'],
 ['B', 'C'],
 ['A', 'B', 'C']
]

You can also use the powerset function from the excellent more_itertools package.

from more_itertools import powerset

l = [1,2,3]
list(powerset(l))

# [(), (1,), (2,), (3,), (1, 2), (1, 3), (2, 3), (1, 2, 3)]

We can also verify, that it meets OP's requirement

from more_itertools import ilen

assert ilen(powerset(range(15))) == 32_768

I know it's far more practical to use itertools to get the all the combinations, but you can achieve this partly with only list comprehension if you so happen to desire, granted you want to code a lot

For combinations of two pairs:

lambda l: [(a, b) for i, a in enumerate(l) for b in l[i+1:]]

And, for combinations of three pairs, it's as easy as this:

lambda l: [(a, b, c) for i, a in enumerate(l) for ii, b in enumerate(l[i+1:]) for c in l[i+ii+2:]]

The result is identical to using itertools.combinations:

import itertools
combs_3 = lambda l: [
    (a, b, c) for i, a in enumerate(l) 
    for ii, b in enumerate(l[i+1:]) 
    for c in l[i+ii+2:]
]
data = ((1, 2), 5, "a", None)
print("A:", list(itertools.combinations(data, 3)))
print("B:", combs_3(data))
# A: [((1, 2), 5, 'a'), ((1, 2), 5, None), ((1, 2), 'a', None), (5, 'a', None)]
# B: [((1, 2), 5, 'a'), ((1, 2), 5, None), ((1, 2), 'a', None), (5, 'a', None)]

Without itertools in Python 3 you could do something like this:

def combinations(arr, carry):
    for i in range(len(arr)):
        yield carry + arr[i]
        yield from combinations(arr[i + 1:], carry + arr[i])

where initially carry = "".

Here are two implementations of itertools.combinations

One that returns a list

def combinations(lst, depth, start=0, items=[]):
    if depth <= 0:
        return [items]
    out = []
    for i in range(start, len(lst)):
        out += combinations(lst, depth - 1, i + 1, items + [lst[i]])
    return out

One returns a generator

def combinations(lst, depth, start=0, prepend=[]):
    if depth <= 0:
        yield prepend
    else:
        for i in range(start, len(lst)):
            for c in combinations(lst, depth - 1, i + 1, prepend + [lst[i]]):
                yield c

Please note that providing a helper function to those is advised because the prepend argument is static and is not changing with every call

print([c for c in combinations([1, 2, 3, 4], 3)])
# [[1, 2, 3], [1, 2, 4], [1, 3, 4], [2, 3, 4]]

# get a hold of prepend
prepend = [c for c in combinations([], -1)][0]
prepend.append(None)

print([c for c in combinations([1, 2, 3, 4], 3)])
# [[None, 1, 2, 3], [None, 1, 2, 4], [None, 1, 3, 4], [None, 2, 3, 4]]

This is a very superficial case but better be safe than sorry

Without using itertools:

def combine(inp):
    return combine_helper(inp, [], [])


def combine_helper(inp, temp, ans):
    for i in range(len(inp)):
        current = inp[i]
        remaining = inp[i + 1:]
        temp.append(current)
        ans.append(tuple(temp))
        combine_helper(remaining, temp, ans)
        temp.pop()
    return ans


print(combine(['a', 'b', 'c', 'd']))

flag = 0
requiredCals =12
from itertools import chain, combinations

def powerset(iterable):
    s = list(iterable)  # allows duplicate elements
    return chain.from_iterable(combinations(s, r) for r in range(len(s)+1))

stuff = [2,9,5,1,6]
for i, combo in enumerate(powerset(stuff), 1):
    if(len(combo)>0):
        #print(combo , sum(combo))
        if(sum(combo)== requiredCals):
            flag = 1
            break
if(flag==1):
    print('True')
else:
    print('else')

I'm late to the party but would like to share the solution I found to the same issue: Specifically, I was looking to do sequential combinations, so for "STAR" I wanted "STAR", "TA", "AR", but not "SR".

lst = [S, T, A, R]
lstCombos = []
for Length in range(0,len(lst)+1):
    for i in lst:
        lstCombos.append(lst[lst.index(i):lst.index(i)+Length])

Duplicates can be filtered with adding in an additional if before the last line:

lst = [S, T, A, R]
lstCombos = []
for Length in range(0,len(lst)+1):
    for i in lst:
         if not lst[lst.index(i):lst.index(i)+Length]) in lstCombos:
             lstCombos.append(lst[lst.index(i):lst.index(i)+Length])

If for some reason this returns blank lists in the output, which happened to me, I added:

for subList in lstCombos:
    if subList = '':
         lstCombos.remove(subList)

I'm a bit late on this topic, but think I can help someone.

You can use product from itertools:

from itertools import product

n = [1, 2, 3]

result = product(n, repeat=3) # You can change the repeat more then n length

print(list(result))

Output:

[(1, 1, 1), (1, 1, 2), (1, 1, 3), (1, 2, 1), (1, 2, 2), (1, 2, 3), (1, 3, 1),
 (1, 3, 2), (1, 3, 3), (2, 1, 1), (2, 1, 2), (2, 1, 3), (2, 2, 1), (2, 2, 2),
 (2, 2, 3), (2, 3, 1), (2, 3, 2), (2, 3, 3), (3, 1, 1), (3, 1, 2), (3, 1, 3), 
(3, 2, 1), (3, 2, 2), (3, 2, 3), (3, 3, 1), (3, 3, 2), (3, 3, 3)]

Another example, but changing repeat arguement:

from itertools import product

n = [1, 2, 3]

result = product(n, repeat=4) # Changing repeat to 4
print(list(result))

Output:

(1, 1, 2, 3), (1, 1, 3, 1), (1, 1, 3, 2), (1, 1, 3, 3), (1, 2, 1, 1), 
(1, 2, 1, 2), (1, 2, 1, 3), (1, 2, 2, 1), (1, 2, 2, 2), (1, 2, 2, 3), 
(1, 2, 3, 1), (1, 2, 3, 2), (1, 2, 3, 3), (1, 3, 1, 1), (1, 3, 1, 2), 
(1, 3, 1, 3), (1, 3, 2, 1), (1, 3, 2, 2), (1, 3, 2, 3), (1, 3, 3, 1), 
(1, 3, 3, 2), (1, 3, 3, 3), (2, 1, 1, 1), (2, 1, 1, 2), (2, 1, 1, 3), 
(2, 1, 2, 1), (2, 1, 2, 2), (2, 1, 2, 3), (2, 1, 3, 1), (2, 1, 3, 2),
 (2, 1, 3, 3), (2, 2, 1, 1), (2, 2, 1, 2), (2, 2, 1, 3), (2, 2, 2, 1), 
(2, 2, 2, 2), (2, 2, 2, 3), (2, 2, 3, 1), (2, 2, 3, 2), (2, 2, 3, 3), 
(2, 3, 1, 1), (2, 3, 1, 2), (2, 3, 1, 3), (2, 3, 2, 1), (2, 3, 2, 2), 
(2, 3, 2, 3), (2, 3, 3, 1), (2, 3, 3, 2), (2, 3, 3, 3), (3, 1, 1, 1), 
(3, 1, 1, 2), (3, 1, 1, 3), (3, 1, 2, 1), (3, 1, 2, 2), (3, 1, 2, 3), 
(3, 1, 3, 1), (3, 1, 3, 2), (3, 1, 3, 3), (3, 2, 1, 1), (3, 2, 1, 2), 
(3, 2, 1, 3), (3, 2, 2, 1), (3, 2, 2, 2), (3, 2, 2, 3), (3, 2, 3, 1), 
(3, 2, 3, 2), (3, 2, 3, 3), (3, 3, 1, 1), (3, 3, 1, 2), (3, 3, 1, 3), 
(3, 3, 2, 1), (3, 3, 2, 2), (3, 3, 2, 3), (3, 3, 3, 1), (3, 3, 3, 2), 
(3, 3, 3, 3)]```

I like this problem because there are so many ways to implement it. I decided to create a reference answer for the future.

What to use in production?

The intertools' documentation has a self-contained example, why not use it in your code? Some people suggested using more_itertools.powerset, but it has exactly the same implementation! If I were you I wouldn't install the whole package for one tiny thing. Probably this is the best way to go:

def powerset(iterable):
    "powerset([1,2,3]) --> () (1,) (2,) (3,) (1,2) (1,3) (2,3) (1,2,3)"
    s = list(iterable)
    return chain.from_iterable(combinations(s, r) for r in range(len(s)+1))

Other possible approaches

Approach 0: Using combinations

import itertools

def subsets(nums):
    result = []
    for i in range(len(nums) + 1):
        result += itertools.combinations(nums, i)
    return result

Approach 1: Straightforward recursion

def subsets(nums):
    result = []

    def powerset(alist, index, curr):
        if index == len(alist):
            result.append(curr)
            return

        powerset(alist, index + 1, curr + [alist[index]])
        powerset(alist, index + 1, curr)

    powerset(nums, 0, [])
    return result

Approach 2: Backtracking

def subsets(nums):
    result = []

    def backtrack(index, curr, k):
        if len(curr) == k:
            result.append(list(curr))
            return
        for i in range(index, len(nums)):
            curr.append(nums[i])
            backtrack(i + 1, curr, k)
            curr.pop()

    for k in range(len(nums) + 1):
        backtrack(0, [], k)
    return result

or

def subsets(nums):
    result = []

    def dfs(nums, index, path, result):
        result.append(path)
        for i in range(index, len(nums)):
            dfs(nums, i + 1, path + [nums[i]], result)

    dfs(nums, 0, [], result)
    return result

Approach 3: Bitmasking

def subsets(nums):
    res = []
    n = len(nums)
    for i in range(1 << n):
        aset = []
        for j in range(n):
            value = (1 << j) & i  # value = (i >> j) & 1
            if value:
                aset.append(nums[j])
        res.append(aset)
    return res

or (not really bitmasking, using intuition that there's exactly 2^n subsets)

def subsets(nums):
    subsets = []
    expected_subsets = 2 ** len(nums)

    def generate_subset(subset, nums):
        if len(subsets) >= expected_subsets:
            return
        if len(subsets) < expected_subsets:
            subsets.append(subset)
        for i in range(len(nums)):
            generate_subset(subset + [nums[i]], nums[i + 1:])

    generate_subset([], nums)
    return subsets

Approach 4: Cascading

def subsets(nums):
    result = [[]]
    for i in range(len(nums)):
        for j in range(len(result)):
            subset = list(result[j])
            subset.append(nums[i])
            result.append(subset)
    return result

If you do not want to use combinations library, here is the solution:

nums = [1,2,3]
p = [[]]
fnl = [[],nums]

for i in range(len(nums)):
    for j in range(i+1,len(nums)):
        p[-1].append([i,j])

for i in range(len(nums)-3):
    p.append([])
    for m in p[-2]:
        p[-1].append(m+[m[-1]+1])

for i in p:
    for j in i:
        n = []
        for m in j:
            if m < len(nums):
                n.append(nums[m])
        if n not in fnl:
            fnl.append(n)

for i in nums:
    if [i] not in fnl:
        fnl.append([i])

print(fnl)

Output:

[[], [1, 2, 3], [1, 2], [1, 3], [2, 3], [1], [2], [3]]

As mentioned by James Brady, you itertools.combinations is a key. But it is not a full solution.

Solution 1

import itertools
def all(lst):
    # ci is a bitmask which denotes particular combination,
    # see explanation below
    for ci in range(1, 2**len(lst)):
        yield tuple(itertools.compress(
            lst,
            [ci & (1<<k) for k in  range(0, len(lst))]
        ))

Solution 2

import itertools
def all_combs(lst):
    for r in range(1, len(lst)+1):
        for comb in itertools.combinations(lst, r):
            yield comb

Example

>>> list(all_combs([1,2,3]))
[(1,), (2,), (3,), (1, 2), (1, 3), (2, 3), (1, 2, 3)]
>>> len(list(all_combs([1,2,3])))
7
>>> len(list(all_combs(range(0, 15))))
32767
>>> list(all([1,2,3]))
[(1,), (2,), (1, 2), (3,), (1, 3), (2, 3), (1, 2, 3)]
>>> len(list(all(range(15))))
32767

Explanation

Assume your array A has length N. Let a bitmask B of length N will denote a particular combination C. If B[i] is 1, then A[i] belongs to combination C.

Solution 1 explanation

So we can just go over all bitmasks and filter source array A with this bitmask, which might be done by itertools.compress.

Solution 2 explanation

...Or, we can represent it as combinations

Now we need to consider cases, when there is single one in B, then when only two ones, and so on. Each case belongs to particular Combination. Thus once we combine all combinations sets we will get all subsequences.

Also, it becomes obvious that amount of all possible combinations in such case is 2^N - 1. We drop case when all B[i] are zeroes, since we assume empty set is not a combination. Otherwise, just don't substract 1.