Quicksort函数不会产生预期的输出，但partition函数会产生（python）

Question

我必须在一个学校项目的“slice”对象上实现一个快速排序算法，该slice对象是一个具有：

• “数据”字段（要排序的整个数组）
• “左”和“右”字段（表示数组中切片子部分的索引）

分区函数代码如下：

def partition (s, cmp, piv=0):
"""
Creates two slices from *s* by selecting in the first slice all
elements being less than the pivot and in the second one all other
elements.

:param s: A slice of is a dictionary with 3 fields :
          - data: the array of objects,
          - left: left bound of the slide (a position in the array),
          - right: right bound of the slice.
:type s: dict
:param cmp: A comparison function, returning 0 if a == b, -1 is a < b, 1 if a > b
:type cmp: function
:return: A couple of slices, the first slice contains all elements that are
         less than the pivot, the second one contains all elements that are
         greater than the pivot, the pivot does not belong to any slice.
:rtype: tuple

>>> import generate
>>> import element
>>> import numpy
>>> def cmp (x,y):
...    if x == y:
...       return 0
...    elif x < y:
...       return -1
...    else:
...       return 1
>>> t = numpy.array([element.Element(i) for i in [4,2,3,1]])
>>> p = {'left':0,'right':len(t)-1,'data':t}
>>> pivot = 0
>>> p1,p2 = partition(p,cmp,pivot)
>>> print(p1['data'][p1['left']:p1['right']+1])
[1 2 3 4]
>>> print(p2['data'][p2['left']:p2['right']+1])
[]
"""
#convenience variables to make the function easily readable.
left = s['left']
right = s['right']
swap(s['data'],right,piv)

j = left
for i in range(left,right+1):
    if (cmp(s['data'][i] , s['data'][right]) != 1 ):
        swap(s['data'],i,j)
        j += 1
s1 = {'data':s['data'],'left':left,'right':j-1}
s2 = {'data':s['data'],'left':j+1,'right':right}

return s1,s2

该函数的doctests不会失败，但是当调用quicksort递归函数时，分区函数的行为不会像预期的那样。quicksort函数的代码如下：

def quicksort_slice (s, cmp):
"""
A sorting function implementing the quicksort algorithm

:param s: A slice of an array, that is a dictionary with 3 fields :
          data, left, right representing resp. an array of objects and left
          and right bounds of the slice.
:type s: dict
:param cmp: A comparison function, returning 0 if a == b, -1 is a < b, 1 if a > b
:type cmp: function
:return: Nothing

>>> import generate
>>> import element
>>> import numpy
>>> def cmp (x,y):
...    if x == y:
...       return 0
...    elif x < y:
...       return -1
...    else:
...       return 1
>>> t = numpy.array([element.Element(i) for i in [4,5,1,2,3,8,9,6,7]])
>>> p = {'left':0,'right':len(t)-1,'data':t}
>>> quicksort_slice(p,cmp)
>>> print(p['data'])
[1 2 3 4 5 6 7 8 9]
"""

if s['left'] < s['right'] :
    s1,s2 = partition(s,cmp)
    quicksort_slice(s1,cmp)
    quicksort_slice(s2,cmp)

File "C:\Users\Marion\Documents\S2\ASD\TP\tp2_asd\src\sorting.py", line 148, in __main__.quicksort_slice
Failed example:
    print(p['data'])
Expected:
    [1 2 3 4 5 6 7 8 9]
Got:
    [8 2 1 3 7 4 9 5 6]

编辑：正如下面的一条评论所指出的，我简直忘了更新piv的值，在对函数的第一次调用之后，piv的值并不总是0。我还替换了：

for i in range(left,right+1):
if (cmp(s['data'][i] , s['data'][right]) != 1 ):
    swap(s['data'],i,j)
    j += 1

与

for i in range(left,right):
    if (cmp(s['data'][i] , s['data'][right]) != 1 ):
        swap(s['data'],i,j)
        j += 1
swap(s['data'],right,j)

Answer 1

尝试关闭这个线程之前，只是意识到我需要一个答案来做这件事。解决方案由JasonHarper提供。正如jasonharper所指出的，我“总是使用索引为0的元素作为枢轴--即使0不在切片内”。