有效的方式在运行中按升序和降序排序

Question

From the following code, I am using the function generator which creates random numbers and helps to simulate incoming data.

from random import randint


def generator(n=1000):
    i = 0
    while i < n:
        yield randint(0, n)
        i += 1

For sorting in ascending order and get only the lowest 10 records I use this:

out_size = 10
out = []
for num in generator():
    if not len(out):
        out.append(num)
    else:
        for i in range(out_size):
            if num < out[i]:
                out.insert(i, num)
                break
            elif len(out) < out_size:
                out.append(num)
                break

print out[:out_size]

Is there an efficient and better way to sort the generated data?, how to doing in descending order. ?

The input is a constant stream, but I just need to keep the lowest and highest 10 records.

Answer 1

Use the heapq.nlargest() and heapq.nsmallest() functions; these use a heap to efficiently track largest or smallest K items for you:

import heapq

out = heapq.nsmallest(10, generator())

A heap invariant is maintained in O(logK) complexity, with K being the size (10 here); the nsmallest function creates a heap of size K, fills it with the first K elements, then pushes each next value onto the heap popping off the now-largest each time, in one operation.

Looping over N elements then makes the total operation O(N logK), while sorting would take O(N logN). Provided K remains smaller than N the heapq approach wins.

Once the generator is exhausted, the heap is returned in sorted order.