1 Counter

计数，以下的例子，找出列表中元素的重复次数：

from collections import Counter

device_temperatures = [13.5, 14.0, 14.0, 14.5, 14.5, 14.5, 15.0, 16.0]

temperature_counter = Counter(device_temperatures)

# Counter({14.5: 3, 14.0: 2, 13.5: 1, 15.0: 1, 16.0: 1})
print(temperature_counter) 
print(type(temperature_counter)) # <class 'collections.Counter'>
print(temperature_counter[13.5]) # 1
print(temperature_counter[14.0]) # 2
print(temperature_counter[14.5]) # 3
print(temperature_counter[15.0]) # 1
print(temperature_counter[16.0]) # 1
1
2
3
4
5
6
7
8
9
10
11
12
13
14

2 defaultdict

例 1

如下的代码将产生错误 KeyError，因为 my_dict 里没有为 'hi' 的 key:

my_dict={'hello':5}
print(my_dict['hi'])

# Traceback (most recent call last):
#   File "<string>", line 4, in <module>
# KeyError: 'hi'
# > 
1
2
3
4
5
6
7

与之相反， defaultdict 从不会引发 KeyError。
以下是一个由 tuple 组成的列表 coworkers，列出了每个人就读过的学校：

coworkers = [('Rolf', 'MIT'), ('Jen', 'Oxford'), ('Rolf', 'Cambridge'), ('Charlie', 'Manchester')] 
1

现在要得到一个dictionary：

{
	'Rolf': ['MIT', 'Cmbridge'],
	'Jen': ['Oxford'],
	'Charlie': ['Manchester']
}
1
2
3
4
5

一般的写法：

coworkers = [('Rolf', 'MIT'), ('Jen', 'Oxford'), ('Rolf', 'Cambridge'), ('Charlie', 'Manchester')] 
alma_maters = {}  # (alma mater：母校)
for coworker in coworkers:
    if coworker[0] not in alma_maters:
        alma_maters[coworker[0]] = []
    alma_maters[coworker[0]].append(coworker[1])
1
2
3
4
5
6

或者：

coworkers = [('Rolf', 'MIT'), ('Jen', 'Oxford'), ('Rolf', 'Cambridge'), ('Charlie', 'Manchester')] 
alma_maters = {}  # (alma mater：母校)
for coworker, place in coworkers:
    if coworker not in alma_maters:
        alma_maters[coworker] = [] # default value
    alma_maters[coworker].append(place)
    
print(alma_maters)
1
2
3
4
5
6
7
8

改为使用 defaultdict ：

from collections import defaultdict

coworkers = [('Rolf', 'MIT'), ('Jen', 'Oxford'), ('Rolf', 'Cambridge'), ('Charlie', 'Manchester')] 
# 如果词典中的某一个key不存在，则调用参数里的function
# 这里 function 是 list，即调用list,得到一个空的列表 []
alma_maters = defaultdict(list)  
for coworker, place in coworkers:
    alma_maters[coworker].append(place)

# 如果希望在访问不存在的 key 时，能引发异常, 则添加下面的一行
# alma_maters.default_factory = None
# (None 改成 int 时生成 0 值)
        
print(alma_maters['Rolf'])  # ['MIT', 'Cambridge']
print(alma_maters['Jen'])  # ['Oxford']
print(alma_maters['Charlie'])  # ['Manchester']
print(alma_maters['Anne'])  # []
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

例 2

from collections import defaultdict

my_company = 'Teclado'

coworkers = ['Jen', 'Li', 'Charlie', 'Rhys']
other_coworkers = [('Rolf', 'Apple Inc.'), ('Anna', 'Google')]

# 不能直接写 my_company, 因为 defaultdict 接受函数为参数
# lambda: my_company 返回 my_company
coworker_companies = defaultdict(lambda: my_company)

for person, company in other_coworkers:
    coworker_companies[person] = company
    
# coworkers[1] 是 'Li', 输出默认值 Teclado
print(coworker_companies[coworkers[1]])  

# other_coworkers[0][0] 是 'Rolf', 输出 Apple Inc.
print(coworker_companies[other_coworkers[0][0]])  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

3 OrderedDict

这里是 Pascal Case

顾名思义，OrderedDict 是有序词典，是指键值对的顺序按插入顺序排序。

from collections import OrderedDict

o = OrderedDict()
o['Rolf'] = 6
o['Jose'] = 10
o['Jen'] = 3

# keys are always in the order in which they were inserted
# OrderedDict([('Rolf', 6), ('Jose', 10), ('Jen', 3)])
print(o)  

o.move_to_end('Rolf') # 移到末尾
# OrderedDict([('Jose', 10), ('Jen', 3), ('Rolf', 6)])
print(o)  

o.move_to_end('Rolf', last = False) # 移到反向的末尾，即开头
# OrderedDict([('Rolf', 6), ('Jose', 10), ('Jen', 3)])
print(o)

o.popitem()

# OrderedDict([('Rolf', 6), ('Jose', 10)])
print(o)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

o.popitem(False) : 删除开头的元素

但从 Python 3.7 开始，dictionary 已经按插入排序，所以 OrderedDict 用处不是特别大。
Are dictionaries ordered in Python 3.6+?

4 namedtuple

namedtuple： 给 tuple 以及 tuple 中的每一个元素都取名字

如下的代码，account[0]，account[1] 分别指的什么不是显而易见：

account = ('checking', 1850.90)

print(account[0])  # name
print(account[1])  # balance
1
2
3
4

使用 namedtuple:

from collections import namedtuple

account = ('checking', 1850.90)

# 第1个参数是 tuple 名称，和定义名称相同
# 第2个参数是 fields 名称
Account = namedtuple("Account", ['name', 'balance'])

accountNamedTuple_1 = Account('checking', 1850.90)
print(accountNamedTuple_1.name, accountNamedTuple_1.balance)  # checking 1850.9

accountNamedTuple_2 = Account._make(account)
account_name_2, account_balance_2 = accountNamedTuple_2
print(account_name_2, account_balance_2) # checking 1850.9

accountNamedTuple_3 = Account(*account)
account_name_3, account_balance_3 = accountNamedTuple_3
print(account_name_3, account_balance_3) # checking 1850.9

print(accountNamedTuple_1._asdict()['balance']) # 1850.9
print(accountNamedTuple_2._asdict()['balance']) # 1850.9
print(accountNamedTuple_3._asdict()['balance']) # 1850.9
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

从 csv 文件或者 database 读取数据时，使用 namedtuple 可使代码更容易理解。

5 deque

deque：double ended queue 双端队列，使用deque而非 list 的原因首先是deque效率高，其次它保证线程安全 (thread safe)，deque 所有的操作都是线程安全的 ，因此在使用 thread 时可使用 deque。

from collections import deque

friends = deque(('Rolf', 'Charlie', 'Jen', 'Anna'))

friends.append('Jose')
friends.appendleft('Anthony')
print(friends)  # deque(['Anthony', 'Rolf', 'Charlie', 'Jen', 'Anna', 'Jose'])

friends.pop()
print(friends)  # deque(['Anthony', 'Rolf', 'Charlie', 'Jen', 'Anna'])

friends.popleft()  
print(friends)  # deque(['Rolf', 'Charlie', 'Jen', 'Anna'])
1
2
3
4
5
6
7
8
9
10
11
12
13