Question

I am trying to implement quicksort in python:

def partition(ls):
  if len(ls) == 0:
    return 0
  pivot = ls[0]
  i = 0
  j = 1
  while j < len(ls):
    if ls[j] <= pivot:
      i += 1
      temp = ls[i]
      ls[i] = ls[j]
      ls[j] = temp
    j += 1
  ls[0] = ls[i]
  ls[i] = pivot
  return i

assert(partition([1,2]) == 0)
assert(partition([3,2]) == 1)
assert(partition([3,2,1,4,5]) == 2)
assert(partition([]) == 0)
assert(partition([45]) == 0)

def sort(ls):
  if len(ls) == 0:
    return
  pivotIndex = partition(ls)
  sort(ls[0:pivotIndex])
  sort(ls[(pivotIndex + 1):len(ls)])

ls = [54,1,3,2,4,3,5,4]
sort(ls)
print ls

Based on my assert statements, I know that my partition algorithm works fine.

However, my sort function returns erroneous results. This snippet of code prints

[4, 1, 3, 2, 4, 3, 5, 54]

What should be the boundaries of the recursive calls to sort? I am aiming to partition the sublist to the left of the pivot and the sublist to the right of the pivot, both of which do not include the pivot itself.

Answer 1

sort(ls[0:pivotIndex])
sort(ls[(pivotIndex + 1):len(ls)])

The slicing copies part of the list, so in the recursive calls, you don’t modify the original list. Thus only the first partition modifies the list.