Extracting a random sublist from a list in Python

Question

I have a python dictionary as follows:

{'APPLE_PROVIDERS' : ["some", "provider","can","be", "null"],
  ....
}

What i want to do is get a random sublist from the list (which is a value) of a key. Not just one element, but a totally random sublist. Here is what I tried:

a_list = a_dict.get('APPLE_PROVIDERS', "")
for item in a_list[randrange(0,len(a_list)) : randrange(0,len(a_list))]:
  ...do something.. 

This thing has two problems :

1. If the list is empty, or if the dict lookup fails, the program fails since randrange has (0,0) as arguments, which results in an error

2. Many times both randrange() calls generate the same number, especially when the list is small. This returns an empty list. For example a_list[5:5]

So what is the best way to get a random sublist with above cases handled ? Also, I do not care about the ordering. Anything works. I just want a totally random sublist of either 0,1... till len(a_list) elements each time the for loop starts.

If the list can be changed in some other data structure which can hold similar elements, that works for me too.

Answer 1

So, assuming that you want an empty list returned if you get an empty list, here's an example solution:

from random import shuffle

def get_random_sublist(the_dict, key, number):
    l = the_dict.get(key, [])
    shuffle(l)
    return l[:number]

So, I'd use random.shuffle. This allows me to avoid the issue of asking for a sublist that's larger than the actual list that we get.

>>> DICT = {'a' : "1 2 3 4 5".split(), 'b': [], 'c': [1]}
>>> get_random_sublist(DICT, 'a', 3)
['4', '1', '2']
>>> get_random_sublist(DICT, 'b', 10)
[]

Answer 2

>>> from random import randint
>>> left = randint(0, len(L))
>>> right = randint(left, len(L))
>>> L[left:right]
['null']

if you don't want the possiblity of empty lists

>>> left = randint(0, len(L) - 1)
>>> right = randint(left + 1, len(L))

Answer 3

Sample it.

>>> random.sample(["some", "provider", "can", "be", "null"], 3)
['some', 'can', 'provider']
>>> random.sample(["some", "provider", "can", "be", "null"], 3)
['can', 'null', 'provider']
>>> random.sample(["some", "provider", "can", "be", "null"], 3)
['null', 'some', 'provider']

Answer 4

Ignacio's answer is great. If you want to minimally modify your code, you can do this:

a_list = a_dict.get('APPLE_PROVIDERS', "")
if len(a_list) > 1:
    index1 = randrange(0,len(a_list)-1)
    index2 = randrange(index1+1,len(a_list))
    for item in a_list[index1:index2]:
        pass #do stuff

I do two things here: 1) I check to see if a_list has more than one element, and 2) I generate indices using randrange, but in such a way that the second is guaranteed to be greater than the first.