Bisect, is it possible to work with descending sorted lists?

Question

How can I use bisect module on lists that are sorted descending? e.g.

import bisect

x = [1.0,2.0,3.0,4.0] # normal, ascending
bisect.insort(x,2.5)  # -->         


        

Answer 1

Probably the easiest thing is to borrow the code from the library and make your own version

def reverse_insort(a, x, lo=0, hi=None):
    """Insert item x in list a, and keep it reverse-sorted assuming a
    is reverse-sorted.

    If x is already in a, insert it to the right of the rightmost x.

    Optional args lo (default 0) and hi (default len(a)) bound the
    slice of a to be searched.
    """
    if lo < 0:
        raise ValueError('lo must be non-negative')
    if hi is None:
        hi = len(a)
    while lo < hi:
        mid = (lo+hi)//2
        if x > a[mid]: hi = mid
        else: lo = mid+1
    a.insert(lo, x)

Answer 2

I've never used to the bisect package. But if it only works in ascending order and you're always keeping your list sorted (whether ascending or descending) then you could simply sort beforehand and then invert (if you want to keep it descending).

x.sort()
bisect.insort(x,2.5)
x.reverse()

Obviously more a hack then a real solution but at least it would work.

Answer 3

One approach is to negate the keys. For example,

a = [1,2,3,4]
a.reverse()
def comparator(a):
    return -a
b = [comparator(i) for i in a]
bisect.insort_right(b,comparator(2.5))
a = [comparator(c) for c in b]

Result: [4, 3, 2.5, 2, 1]

Answer 4

Hopefully the "key" argument will be added to bisect module functions one day (see: http://bugs.python.org/issue4356). Unfortunately it's not possible without some kind of workaround at the moment (Python version < 3.4).

One possible workaround is to use a wrapper like:

class ReversedSequenceView(collections.MutableSequence):
    def __init__(self, seq):
        self._seq = seq

    def __getitem__(self, i):
        return self._seq[-1 - i]

    def __setitem__(self, i, v):
        self._seq[-1 - i] = v

    def __delitem__(self, i):
        del self._seq[-1 - i]

    def __len__(self):
        return len(self._seq)

    def insert(self, i, v):
        return self._seq.insert(len(self._seq) - i, v)

Usage:

>>> x = [4., 3., 2., 1.]
>>> bisect.insort(ReversedSequenceView(x), 2.5)
>>> x
[4.0, 3.0, 2.5, 2.0, 1.0]

Answer 5

I was looking to get the position to insert a number in a list, to keep it ordered. I ended up using the following solution:

def rev_bisect(l, v):
    '''l -> list (descending order)
       v -> value
    Returns:
       int -> position to insert value in list keeping it sorted
    '''
    h = list(range(len(l))) + [len(l)+1]
    h.reverse()
    l.reverse()
    return h[bisect.bisect_left(l, v)]

Answer 6

From the documentation:

Unlike the sorted() function, it does not make sense for the bisect() functions to have key or reversed arguments because that would lead to an inefficent design (successive calls to bisect functions would not “remember” all of the previous key lookups).

Therefore, if you have a list with inverse order, then you are out of luck.

The main usecase for bisect is the efficient update of an already ordered list.
You may want either to change the data format of your list (e.g. maintaining it in direct order as much as possible, and then reversing it at the very end), either to implement your own version of bisect.
Or, if you are not in the main usecase, you can opt not to use it at all, e.g. by inserting all elements and then sorting them at the very end.