Python collections中的双向队列deque简单介绍详解

来源：jb51　　时间：2019/11/5 10:11:34　　对本文有异议

前言

在python神书《Python+Cookbook》中有这么一段话：在队列两端插入或删除元素时间复杂度都是 O(1) ，而在列表的开头插入或删除元素的时间复杂度为 O(N)。
于是就想验证下。

简单使用

基本代码

from collections import deque
q = deque(maxlen=4)#有固定长度的双向队列
qq = deque() #无固定长度
print(dir(q))#看看有哪些可用方法或属性

结果：

['__add__', '__bool__', '__class__', '__contains__', '__copy__', '__delattr__', '__delitem__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getitem__', '__gt__', '__hash__', '__iadd__', '__imul__', '__init__', '__init_subclass__', '__iter__', '__le__', '__len__', '__lt__', '__mul__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__reversed__', '__rmul__', '__setattr__', '__setitem__', '__sizeof__', '__str__', '__subclasshook__', 'append', 'appendleft', 'clear', 'copy', 'count', 'extend', 'extendleft', 'index', 'insert', 'maxlen', 'pop', 'popleft', 'remove', 'reverse', 'rotate']

看到可以append，pop，insert，clear等，还可以像List一样用中括号 [] 对某个index获取或设置值。因为是双向队列，所以也有左操作函数：appendleft，popleft。额外的还要反转函数reverse，计数函数count。

使用ipython验证

In [1]: from collections import deque
…: q = deque(maxlen=4)#有固定长度的双向队列
…: qq = deque() #无固定长度
…: print(dir(q))#看看有哪些可用方法或属性
[‘add', ‘bool', ‘class', ‘contains', ‘copy', ‘delattr', ‘delitem', ‘dir', ‘doc', ‘eq', ‘format', ‘ge', ‘getattribute', ‘getitem', ‘gt', ‘hash', ‘iadd', ‘imul', ‘init', ‘init_subclass', ‘iter', ‘le', ‘len', ‘lt', ‘mul', ‘ne', ‘new', ‘reduce', ‘reduce_ex', ‘repr', ‘reversed', ‘rmul', ‘setattr', ‘setitem', ‘sizeof', ‘str', ‘subclasshook', ‘append', ‘appendleft', ‘clear', ‘copy', ‘count', ‘extend', ‘extendleft', ‘index', ‘insert', ‘maxlen', ‘pop', ‘popleft', ‘remove', ‘reverse', ‘rotate']
In [2]: q
Out[2]: deque([])
In [3]: q.append(1)
In [4]: q.insert(0,33)
In [6]: q
Out[6]: deque([33, 1])
In [8]: q.appendleft(44)
In [9]: q
Out[9]: deque([44, 33, 1])
In [10]: q.pop()
Out[10]: 1
In [12]: q[1]
Out[12]: 33
In [13]: q
Out[13]: deque([44, 33])
In [14]: q.reverse()
In [15]: q
Out[15]: deque([33, 44])
In [17]: q.clear()
In [18]: q
Out[18]: deque([])

性能测试

pop和append

#coding:utf8
import datetime,time
from collections import deque
D = deque()
L=[]
def calcTime(func):  
  def doCalcTime():
    sst = int(time.time()*1000)
    func()
    eed = int(time.time()*1000)
    print(func,'cost time:',eed-sst,'ms')
  return doCalcTime
 
@calcTime
def didDeque():
  for i in range(0,10000000):
    D.append(i)
  while D:
    D.pop()
 
@calcTime
def didList():
  for i in range(0,10000000):
    L.append(i)
  while L:
    L.pop()
    
if __name__=='__main__':
  didDeque()
  print("------------")
  didList()

运行结果：

<function didDeque at 0x000002D6912A4D08> cost time: 1924 ms
------------
<function didList at 0x000002D6912D4048> cost time: 2420 ms

是快了一些。

insert

#coding:utf8
import datetime,time
from collections import deque
D = deque()
L=[]
def calcTime(func):  
  def doCalcTime():
    sst = int(time.time()*1000)
    func()
    eed = int(time.time()*1000)
    print(func,'cost time:',eed-sst,'ms')
  return doCalcTime
 
@calcTime
def didDeque():
  for i in range(0,100000):
    D.insert(5,i)
    
@calcTime
def didList():
  for i in range(0,100000):
    L.insert(5,i)
 
if __name__=='__main__':
  didDeque()
  print("------------")
  didList()

运行结果：

<function didDeque at 0x0000021367F06D08> cost time: 32 ms
------------
<function didList at 0x0000021367F34048> cost time: 3499 ms

快了两个数量级。想想也明白，一个是链表，插入的时候只需要改变指针指向，而List是连续空间，需要移动一大堆的元素。

计算移动平均

>>> import numpy as np
>>> from collections import deque
>>> q=deque(maxlen=5)
>>> q.append(1)
>>> q.append(2)
>>> q.append(3)
>>> q.append(4)
>>> q.append(5)
>>> q.append(6)
>>> q
deque([2, 3, 4, 5, 6], maxlen=5)
>>> np.array(q).mean()
4.0