笔记：序列的修改、散列和切片

1. from array import array
2. import reprlib, math, numbers
3. from functools import reduce
4. from operator import xor
5. from itertools import chain

复制代码

1. # Vector_v1
2. class Vector:
3.     typecode = 'd'
5.     def __init__(self, components) -> None:
6.         self._componeents = array(self.typecode, components)
8.     def __iter__(self):
9.         return iter(self._componeents)
11.     def __repr__(self):
12.         components = reprlib.repr(self._componeents)  # array('d', [1, 2, 3, 4, ...])
13.         components = components[components.find('['):-1]  # [1, 2, 3, 4, ...]
14.         return 'Vector({})'.format(components)  # Vector([1, 2, 3, 4])
15.    
16.     def __str__(self):
17.         return str(tuple(self))
18.    
19.     def __bytes__(self):
20.         return (bytes([ord(self.typecode)]) + bytes(self._componeents))
22.     def __eq__(self, other):
23.         return tuple(self) == tuple(other)
25.     def __abs__(self):
26.         return math.sqrt(sum(x * x for x in self))
28.     def __bool__(self):
29.         return bool(abs(self))
30.    
31.     def __len__(self):
32.         return len(self._componeents)
33.    
34.     def __getitem__(self, index):
35.         return self._componeents[index]
37.     @classmethod
38.     def frombytes(cls, octets):
39.         typecode = chr(octets[0])
40.         memv = memoryview(octets[1:]).cast(typecode)
41.         return cls(memv)

复制代码

reprlib.repr() 获取对象的有限长度表现形式, 多的用 ... 表示, eg. array('d', [0, 1, 2, 3, 4, ...])

1. # Test
2. v1 = Vector([3, 4, 5])
4. print(len(v1))
5. print(v1[0], v1[-1])
7. v7 = Vector(range(7))
8. print(v7[1:4])

复制代码

1. 3
2. 3.0 5.0
3. array('d', [1.0, 2.0, 3.0])

复制代码

切片原理

1. class Seq:
2.     def __getitem__(self, index):
3.         return index
5. s = Seq()
6. print(1, s[1])
7. print(2, s[1:4])
8. print(3, s[1:4:2])
9. print(4, s[1:4:2, 9])
10. print(5, s[1:4:2, 7:9])

复制代码

1. 1 1
2. 2 slice(1, 4, None)
3. 3 slice(1, 4, 2)
4. 4 (slice(1, 4, 2), 9)
5. 5 (slice(1, 4, 2), slice(7, 9, None))

复制代码

1. print(slice)
2. print(dir(slice))
3. print(help(slice.indices))

复制代码

1. <class 'slice'>
2. ['__class__', '__delattr__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', 'indices', 'start', 'step', 'stop']
3. Help on method_descriptor:
5. indices(...)
6.     S.indices(len) -> (start, stop, stride)
7.    
8.     Assuming a sequence of length len, calculate the start and stop
9.     indices, and the stride length of the extended slice described by
10.     S. Out of bounds indices are clipped in a manner consistent with the
11.     handling of normal slices.
13. None

复制代码

S.indices(len) -> (start, stop, stride)
自动将 slice 适配到 长度为 len 的对象上

1. print(slice(None, 10, 2).indices(5))
2. print(slice(-3, None, None).indices(5))

复制代码

1. (0, 5, 2)
2. (2, 5, 1)

复制代码

1. print('ABCDE'[:10:2])  # 等价于
2. print('ABCDE'[0:5:2])
4. print('ABCDE'[-3:])  # 等价于
5. print('ABCDE'[2:5:1])

复制代码

1. ACE
2. ACE
3. CDE
4. CDE

复制代码

1. # Vector_v2
2. class Vector:
3.     typecode = 'd'
5.     def __init__(self, components) -> None:
6.         self._componeents = array(self.typecode, components)
8.     def __iter__(self):
9.         return iter(self._componeents)
11.     ################################# NEW ######################################
12.     def __getitem__(self, index):
13.         cls = type(self)
14.         if isinstance(index, slice):  # slice 切片
15.             return cls(self._componeents[index])
16.         elif isinstance(index, numbers.Integral):  # int 索引
17.             return self._componeents[index]
18.         else:  # 抛出异常
19.             msg = '{cls.__name__} indices must be integers'
20.             raise TypeError(msg.format(cls=cls))
22.     def __repr__(self):
23.         components = reprlib.repr(self._componeents)  # array('d', [1, 2, 3, 4, ...])
24.         components = components[components.find('['):-1]  # [1, 2, 3, 4, ...]
25.         return 'Vector({})'.format(components)  # Vector([1, 2, 3, 4])
26.    
27.     def __str__(self):
28.         return str(tuple(self))
29.    
30.     def __bytes__(self):
31.         return (bytes([ord(self.typecode)]) + bytes(self._componeents))
33.     def __eq__(self, other):
34.         return tuple(self) == tuple(other)
36.     def __abs__(self):
37.         return math.sqrt(sum(x * x for x in self))
39.     def __bool__(self):
40.         return bool(abs(self))
41.    
42.     def __len__(self):
43.         return len(self._componeents)
45.     @classmethod
46.     def frombytes(cls, octets):
47.         typecode = chr(octets[0])
48.         memv = memoryview(octets[1:]).cast(typecode)
49.         return cls(memv)

复制代码

1. v7 = Vector(range(7))
2. print(1, v7[-1])
3. print(2, v7[1:4])
4. print(3, v7[-1:])
5. print(4, v7[1, 2])  # 报误，不支持多维切片

复制代码

1. 1 6.0
2. 2 (1.0, 2.0, 3.0)
3. 3 (6.0,)
5. ---------------------------------------------------------------------------
7. TypeError                                 Traceback (most recent call last)
9. Cell In [28], line 5
10.       3 print(2, v7[1:4])
11.       4 print(3, v7[-1:])
12. ----> 5 print(4, v7[1, 2])
14. Cell In [22], line 19, in Vector.__getitem__(self, index)
15.      17 else:  # 抛出异常
16.      18     msg = '{cls.__name__} indices must be integers'
17. ---> 19     raise TypeError(msg.format(cls=cls))
19. TypeError: Vector indices must be integers

复制代码

1. # Vector_v3 动态存取属性
3. class Vector:
4.     typecode = 'd'
5.     shortcut_numes = 'xyzt'
7.     def __init__(self, components) -> None:
8.         self._componeents = array(self.typecode, components)
10.     def __iter__(self):
11.         return iter(self._componeents)
12.    
13.     def __getitem__(self, index):
14.         cls = type(self)
15.         if isinstance(index, slice):  # slice 切片
16.             return cls(self._componeents[index])
17.         elif isinstance(index, numbers.Integral):  # int 索引
18.             return self._componeents[index]
19.         else:  # 抛出异常
20.             msg = '{cls.__name__} indices must be integers'
21.             raise TypeError(msg.format(cls=cls))
23.     ################################# NEW ######################################
24.     def __getattr__(self, name):
25.         cls = type(self)
26.         if len(name) == 1:
27.             pos = cls.shortcut_numes.find(name)
28.             if 0 <= pos < len(self._componeents):
29.                 return self._componeents[pos]
30.         
31.         msg = '{.__name__!r} object has no attribute {!r}'
32.         raise AttributeError(msg.format(cls, name))

复制代码

通过 __setattr__ 方法防止其修改部分属性

1. v5 = Vector(range(5))
2. print(1, v5)
3. print(2, v5.x)
5. v5.x = 10
6. print(3, v5.x)  # v5[0] 变了？
8. print(4, v5)  # v5 实际上并没有变

复制代码

1. 1 (0.0, 1.0, 2.0, 3.0, 4.0)
2. 2 0.0
3. 3 10
4. 4 (0.0, 1.0, 2.0, 3.0, 4.0)

复制代码

散列 & 快速等值测试

__hash__
需要依次计算 v[0] ^ v[1] ^ v[2] ...
reduce()
__eq__

1. # 改进
3. # Vector_v3 动态存取属性
5. class Vector:
6.     typecode = 'd'
7.     shortcut_numes = 'xyzt'
9.     def __init__(self, components) -> None:
10.         self._componeents = array(self.typecode, components)
12.     def __getattr__(self, name):
13.         cls = type(self)
14.         if len(name) == 1:
15.             pos = cls.shortcut_numes.find(name)
16.             if 0 <= pos < len(self._componeents):
17.                 return self._componeents[pos]
18.         
19.         msg = '{.__name__!r} object has no attribute {!r}'
20.         raise AttributeError(msg.format(cls, name))
22.     ################################# NEW ######################################
23.     def __setattr__(self, name, value) -> None:
24.         cls = type(self)
25.         if len(name) == 1:
26.             if name in cls.shortcut_numes:
27.                 error = 'readonly attribute {attr_name!r}'
28.             elif name.islower():
29.                 error = "can't set attributes 'a' to 'z' in {cls_name!r}"
30.             else:
31.                 error = ''
33.             if error:
34.                 msg = error.format(cls_name=cls.__name__, attr_name=name)
35.                 raise AttributeError(msg)
36.         super().__setattr__(name, value)  

复制代码

1. v6 = Vector(range(6))
2. print(1, v6)
3. print(2, v6.x)
5. v6.x = 10

复制代码

1. 1 <__main__.Vector object at 0x000001BD0AD009A0>
2. 2 0.0
4. ---------------------------------------------------------------------------
6. AttributeError                            Traceback (most recent call last)
8. Cell In [35], line 5
9.       2 print(1, v6)
10.       3 print(2, v6.x)
11. ----> 5 v6.x = 10
13. Cell In [34], line 35, in Vector.__setattr__(self, name, value)
14.      33     if error:
15.      34         msg = error.format(cls_name=cls.__name__, attr_name=name)
16. ---> 35         raise AttributeError(msg)
17.      36 super().__setattr__(name, value)
19. AttributeError: readonly attribute 'x'

复制代码

1. # 复习一下 reduce
2. print(reduce(lambda a, b: a * b, range(1, 6)))

复制代码

1. 120

复制代码

注：
使用 reduce() 最好提供三个参数 reduce(function, iterable, initializer)
通常, 对于 +, |, ^ initializer = 0
通常, 对于 *, &      initializer = 1
__eq__ 要和 __hash__ 在一起哦~

1. # 计算多个数异或与
2. print(reduce(lambda a, b: a ^ b, range(233)))
3. print(reduce(xor, range(233)))
5. n = 0
6. for i in range(1, 233):
7.     n ^= i
8. print(n)

复制代码

all() 只要有一个是 False , 整个都是 False
上面两种 __eq__ 效果相等

1. 232
2. 232
3. 232

复制代码

1. # Vector_v3 加入 __hash__
3. class Vector:
4.     typecode = 'd'
6.     def __init__(self, components) -> None:
7.         self._componeents = array(self.typecode, components)
9.     def __iter__(self):
10.         return iter(self._componeents)
12.     ################################# OLD ######################################
13.     def __eq__(self, other):
14.         return tuple(self) == tuple(other)
16.     ################################# NEW ######################################
17.     def __hash__(self):
18.         hashes = (hash(x) for x in self._componeents)
19.         return reduce(xor, hashes, 0)
20.    

复制代码

zip_longest() 按照最长的 iter 迭代, 空的用 fillvalue 补齐
format 格式化
目标: 得到球面坐标

1. # 使用【映射归约】实现 __hash__ (map, reduce)
2. # Vector_v3 加入 __hash__
4. class Vector:
5.     typecode = 'd'
7.     def __init__(self, components) -> None:
8.         self._componeents = array(self.typecode, components)
10.     def __iter__(self):
11.         return iter(self._componeents)
13.     ################################# NEW ######################################
14.     def __hash__(self):
15.         hashes = map(hash, self._componeents)  # 计算各个元素的hash
16.         return reduce(xor, hashes, 0)
18.     ################################# NEW ######################################
19.     def __eq__(self, other):
20.         if len(self) != len(other):  # 长度不相等, 直接 False
21.             return False
23.         for a, b in zip(self, other):  # 判断每个对应元素是否相等
24.             if a != b:
25.                 return False
27.         return True
29.     ################################# NEW ######################################
30.     # 另一种方法
31.     def __eq__(self, other):
32.         return len(self) != len(other) and all(a == b for a, b in zip(self, other))

复制代码

1. # 回忆一下 zip()
3. print(1, zip(range(3), 'ABC'))
4. print(2, list(zip(range(3), 'ABC')))
6. print(3, list(zip(range(3), 'ABC', [0, 1, 2, 3])))  # 什么？？？一个可迭代对象迭代完了, 就不迭代了
8. from itertools import zip_longest
9. print(4, list(zip_longest(range(3), 'ABC', [0, 1, 2, 3], fillvalue=1)))  # 按照最长的iter迭代, 空的用 fillvalue 补齐

复制代码