Question

I’m trying to implement quicksort in Python using 2 main functions – partition and quicksort.
The partition function is designed so that it returns 2 arrays – bigger and smaller than p.
after that quicksort is called on both of them separately.
so the quicksort works like this:

def quicksort(array)
    pivot = 0 # pivot choice is irrelevant
    left,right = partition(array,pivot)
    quicksort(left)
    quicksoft(right)
    return left+right

But from my understanding it should be possible to design partition to return just one single index – delimiting bigger and smaller arrays and redesign quicksort as follows:

def quicksort(array)
    pivot = 0 # pivot choice is irrelevant
    i = partition(array,pivot)
    quicksort(array[:i-1])
    quicksoft(array[i:])
    return array

but this implementation returns partially sorted array

original array [5, 4, 2, 1, 6, 7, 3, 8, 9]
sorted array [3, 4, 2, 1, 5, 7, 6, 8, 9]

what am i missing here?

Answer 1

without seeing your code it’s hard to be sure, but one possible error is the i-1:

>>> [1,2,3,4][:2]
[1, 2]
>>> [1,2,3,4][2:]
[3, 4]

(although you may be simply skipping the pivot?)

also, slices are new lists, not views:

>>> l = [1,2,3,4]
>>> l[2:][0] = 'three'
>>> l
[1, 2, 3, 4]

which is unfortunate (the typical functional program doing quicksort which is not a quicksort at all, dammit, because it’s creating a pile of new arrays annoys me too…)

you can work round the second problem by passing the entire list plus lo/hi indices:

def quicksort(data, lo=0, hi=None):
    if hi is None: hi = len(data)
    ....