Python 的奇淫技巧
列了一些我新發現的各種花招
Function Attributes (python 2.1)
def a():
a.count += 1
a.count = 0
for i in range(10):
a()
print(a.count) # 10
跟 C++ 中，在 function 裡面宣告 static 變數差不多
Keyword-Only Arguments (python 3.0)
def func(a, b, *, c = None):
pass
func(1, 2, 3)不行func(1, 2, c = 3)這樣才合法
Additional Unpacking Generalizations (python 3.5)
a = [1, 2, 3]
b = [*a, 4, 5]
print(b) # [1, 2, 3, 4, 5]
Merge Dicts
a = {1: 2}
b = {3: 4}
c = {**a, **b}
print(c) # {1: 2, 3: 4}
Metaclasses
class A:
pass
B = type('B',
(A,),
{
'attr': 10,
'func': lambda obj: obj.attr
})
b = B()
print(b.attr) # 10
print(b.func()) # 10
class A:
pass
class B(A):
attr = 10
def func(self):
return self.attr
b = B()
print(b.attr) # 10
print(b.func()) # 10
看看就好 xD
shallow vs deep copy
Shallow vs Deep Copying of Python Objects
assign
a = [[1, 2, 3], [4, 5, 6]]
b = a
b.append('hello')
b[0].append('world')
print(a)
# [[1, 2, 3, 'world'], [4, 5, 6], 'hello']
print(b)
# [[1, 2, 3, 'world'], [4, 5, 6], 'hello']
copy
import copy
a = [[1, 2, 3], [4, 5, 6]]
b = copy.copy(a)
b.append('hello')
b[0].append('world')
print(a)
# [[1, 2, 3, 'world'], [4, 5, 6]]
print(b)
# [[1, 2, 3, 'world'], [4, 5, 6], 'new object']
deepcopy
import copy
a = [[1, 2, 3], [4, 5, 6]]
b = copy.deepcopy(a)
b.append('hello')
b[0].append('world')
print(a)
# [[1, 2, 3], [4, 5, 6]]
print(b)
# [[1, 2, 3, 'world'], [4, 5, 6], 'new object']
copy.copy (shallow)只複製該物件，不會複製子物件copy.deepcopy (deep)會遞迴複製所有子物件
shallow copy on list
a = [1, 2, 3]
b = list(a)
a = [1, 2, 3]
b = a[:]
a = [1, 2, 3]
b = a.copy()
import copy
a = [1, 2, 3]
b = copy.copy(a)
annotations
function annotations
def func(a: int, b: list) -> int:
pass
print(func.__annotations__)
# {'a': <class 'int'>, 'b': <class 'list'>, 'return': <class 'int'>}
class annotations
class A():
var: int
print(A.__annotations__)
# {'var': <class 'int'>}
variable annotations
a: int
b: int = 2
print(__annotations__)
# {'a': <class 'int'>, 'b': <class 'int'>}
intern string
s = 'hello'
s = sys.intern(s)
把字串存進快取池，相同的字串只會存一次 做字串比對會比較快且節省空間