Python Decorator装饰器

1. 基本装饰器

1.1. 无参方法无参装饰器

# 无参log装饰器
def log(func):
    def wrapper():
        print('Log: excute method {}()'.format(func.__name__))
        return func()
    return wrapper
# 无参方法
@log
def do_save():
    print('save something!')
# Log: excute method do_save()
# save something!
do_save()

1.2. 有参方法无参装饰器

利用(*args, **kwargs)来表示任意参数。
此时若@log则会出现异常，因为@log装饰器的实现中并没有接收参数。

# 无参装饰器
def log_new(func):
    def wrapper(*args, **kwargs):
        print('Log: excute method {}()'.format(func.__name__))
        return func(*args, **kwargs)
    return wrapper

# 带参数的方法
# @log  # TypeError need 1 argument
@log_new
def do_save_user(user):
    print('save {}'.format(user))

# Log: excute method do_save_user()
# save JiangW
do_save_user('JiangW')

1.3. 有参方法有参装饰器

# 带参数的装饰器
def log_level(level):
    def decorator(func):
        def wrapper(*args, **kwargs):
            print('Log[{}]: excute method {}()'.format(level, func.__name__))
            return func(*args, **kwargs)
        return wrapper
    return decorator


@log_level('INFO')
def do_delete_user(user):
    print('delete {}'.format(user))

# Log[INFO]: excute method do_delete_user()
# delete JiangW
do_delete_user('JiangW')

2. 基于类实现的装饰器

2.1. 简单基于类实现的装饰器

# 带参数
class LoggingLevel(object):
    def __init__(self, level):
        self.level = level

    def __call__(self, func):
        def wrapper(*args, **kwargs):
            print('Log[{}]: excute method {}()'.format(self.level, func.__name__))
            return func(*args, **kwargs)
        return wrapper

@LoggingLevel('WARN')
def do_remove_user(user):
    print('remove {}'.format(user))

# Log[WARN]: excute method do_remove_user()
# remove JiangW
do_remove_user('JiangW')

3. 注意事项

3.1. 拆分装饰器

利用@log_level语法糖来实现的装饰器。

def log_level(level):
    def decorator(func):
        def wrapper(*args, **kwargs):
            print('Log[{}]: excute method {}()'.format(level, func.__name__))
            return func(*args, **kwargs)
        return wrapper
    return decorator

@log_level('INFO')
def do_delete_user(user):
    print('delete {}'.format(user))

# Log[INFO]: excute method do_delete_user()
# delete JiangW
do_delete_user('JiangW')

拆分成每一步实际上最终为:

# 等同于
def do_delete_user(user):
    print('delete {}'.format(user))

# # Log[INFO]: excute method do_delete_user()
# delete JiangW
decorator = log_level('INFO')
wrapper = decorator(do_delete_user)
wrapper('Jian111gW')

3.2. 函数名称改变

如3.1所述拆分装饰器后实际上最终是调用了

wrapper = decorator(do_delete_user)
wrapper('Jian111gW')

此时被@log_level装饰器装饰的函数名已经变成了wrapper

print(do_delete_user.__name__)  # wrapper

利用functools.wraps来修正

def log_level(level):
    def decorator(func):
        @wraps(func)
        def wrapper(*args, **kwargs):
            print('Log[{}]: excute method {}()'.format(level, func.__name__))
            return func(*args, **kwargs)
        return wrapper
    return decorator

@log_level('INFO')
def do_delete_user(user):
    print('delete {}'.format(user))

print(do_delete_user.__name__)  # do_delete_user

此时do_delete_user.__name__为do_delete_user.

4. 使用wrapt和decoractor

4.1. wrapt

4.2. decoractor

5. 延伸

5.1. 内置装饰器

5.2. 另类用法