Python中使用Type hinting 和 annotations

Posted 2021-01-30 疯人院主任

Type hints最大的好处就是易于代码维护。当新成员加入，想要贡献代码时，能减少很多时间。
也方便我们在调用汉书时提供了错误的类型传递导致运行时错误的检测。

第一个类型注解示例

我们使用一个简单例子，两个整数相加。

def add(a, b):
    return a + b

上面的例子，可工作于任意可以进行+操作符的对象。如果我们仅让该函数只能对整型作为参数，然后也只是返回整型结果呢？

def add(a: int, b: int) -> int:
    return a + b

我们注意到，返回类型是在函数定义末尾使用 -> 符号来指明。

使用mypi及更多例子

Mypy是为Python构建的静态类型检查器。如果我们使用上面的类型注解，mypy可以在代码中
帮我们找到一些错误。你可以使用在开发流程中任意阶段使用它，比如是在CI中作适当的测试。

安装mypy

我们在虚拟环境中安装mypy

$ pip install mypy

我们的示例

我们会进行如下示例描述，虽然代码作用不大，但是我们可以通过它来学习类型注解以及mypy.

class Student:

    def __init__(self, name, batch, branch, roll):
        self.name = name
        self.batch = batch
        self.branch = branch
        self.roll = roll
        self.semester = None
        self.papers = {}

    def is_passed(self):
        "To find if the student has pass the exam in the current semester"
        for k, v in self.papers.items():
            if v < 34:
                return False

        return True


    def total_score(self):
        "Returns the total score of the student"
        total = 0
        for k, v in self.papers.items():
            total += v

        return total


std1 = Student("Kushal", 2005, "cse", "123")
std2 = Student("Sayan", 2005, "cse", 121)
std3 = Student("Anwesha", 2005, "law", 122)

std1.papers = {"english": 78, "math": 82, "science": 77}
std2.papers = {"english": 80, "math": 92, "science": "78"}
std3.papers = {"english": 82, "math": 87, "science": 77}

for std in [std1, std2, std3]:
    print("Passed: {0}. The toral score of {1} is {2}".format(std.is_passed(), std.name, std.total_score()))

你可能发现了代码中其实有错误，但是在实际开发过程我们经常会发生，并且除了在运行时发现，我们并没有其他更好的机制。

使用 mypy

我们通过mypy来执行我们的代码，我们把文件取名为students2.py

加入一些类型注解

我们将会在__init__方法中加入一些类型注解。为减少代码长度，修改如下：

class Student:

    def __init__(self, name: str, batch: int, branch: str, roll: int) -> None:
        self.name = name
        self.batch = batch
        self.branch = branch
        self.roll = roll
        self.semester = None
        self.papers = {}

$ mypy students2.py 
students2.py:9: error: Need type annotation for 'papers'
students2.py:29: error: Argument 4 to "Student" has incompatible type "str"; expected "int"

可以看到mypy有提示哪些变量没有类型注解，还有在29行，参数我们期望的是整型，但在调用时传递了字符串类型，现在让我们来修正他。

from typing import Dict

class Student:

    def __init__(self, name: str, batch: int, branch: str, roll: int) -> None:
        self.name = name
        self.batch = batch
        self.branch = branch
        self.roll = roll
        self.semester = None
        self.papers: Dict[str, int] = {}

    def is_passed(self):
        "To find if the student has pass the exam in the current semester"
        for k, v in self.papers.items():
            if v < 34:
                return False

        return True


    def total_score(self):
        "Returns the total score of the student"
        total = 0
        for k, v in self.papers.items():
            total += v

        return total


std1: Student = Student("Kushal", 2005, "cse", 123)
std2: Student = Student("Sayan", 2005, "cse", 121)
std3: Student = Student("Anwesha", 2005, "law", 122)

std1.papers = {"english": 78, "math": 82, "science": 77}
std2.papers = {"english": 80, "math": 92, "science": 78}
std3.papers = {"english": 82, "math": 87, "science": 77}

for std in [std1, std2, std3]:
    print("Passed: {0}. The toral score of {1} is {2}".format(std.is_passed(), std.name, std.total_score()))

现在，没有任何错误了。在第一行我们还从typing包引入了Dcit。并作为了self.paper的类型注解，这里的意思就是该变量是字典类型，使用字符串作为键，整型作为值。我们设置std1, std2和std3变量注解为Student类。

现在，我们给papers变量赋一些错误类型的值。

std1.papers = ["English", "Math"]

或者错误的字典键值对

std2.papers = {1: "Engish", 2: "Math"}

我们能看到类似如下错误

$ mypy students2.py
students2.py:35: error: Incompatible types in assignment (expression has type List[str], variable has type Dict[str, int])
students2.py:36: error: Dict entry 0 has incompatible type "int": "str"
students2.py:36: error: Dict entry 1 has incompatible type "int": "str"

更多类型注解示例

from typing import List, Tuple, Sequence, Optional

values: List[int] = []
city: int = 350 # The city code, not a name


# This function returns a Tuple of two values, a str and an int
def get_details() -> Tuple[str, int]:
    return "Python", 5

# The following is an example of Tuple unpacking
name: str
marks: int
name, marks = get_details()


def print_all(values: Sequence) -> None:
    for v in values:
        print(v)


print_all([1,2,3])
print_all({"name": "kushal", "class": 5})
# alltypes.py:23: error: Argument 1 to "print_all" has incompatible type Dict[str, object]; expected Sequence[Any]
# But running the code will give us no error with wrong output

def add_ten(number: Optional[int] = None) -> int:
    if number:
        return number + 10
    else:
        return 42

print(add_ten())
print(add_ten(12))

你可以从PEP 484了解更多类型。typing模块进行更多例子解释如何在代码中使用类型注解。