深圳幻海软件技术有限公司 欢迎您!

计算函数执行时长的方法

2023-02-28

python开发,有时需要做性能分析及性能优化,这时就需要记录一些耗时函数执行时间问题,然后针对函数逻辑进行优化。在python3中一般都有哪些方法呢。1、使用time.time()这种方法较简单,但如果想更精确的计算函数的执行时间,会产生精度缺失,没办法统计时间极短的函数耗时。复制importti

python开发,有时需要做性能分析及性能优化,这时就需要记录一些耗时函数执行时间问题,然后针对函数逻辑进行优化。

在python3中一般都有哪些方法呢。

1、使用time.time()

这种方法较简单,但如果想更精确的计算函数的执行时间,会产生精度缺失,没办法统计时间极短的函数耗时。

import time
 
 def func():
     time.sleep(1)
     
 t = time.time()
 func()
 print(f'耗时:{time.time() - t:.4f}s')
 
 耗时:1.0050s
  • 1.
  • 2.
  • 3.
  • 4.
  • 5.
  • 6.
  • 7.
  • 8.
  • 9.
  • 10.

2、使用time.perf_counter()

perf_counter是在python3.3新添加的,返回性能计数器的值,返回值是浮点型,统计结果包括睡眠的时间,单个函数的返回值无意义,只有多次运行取差值的结果才是有效的函数执行时间。

import time
 def func():
     print('hello world')
 t = time.perf_counter()
 func()
 print(f'耗时:{time.perf_counter() - t:.8f}s')
 hello world
 耗时:0.00051790s
  • 1.
  • 2.
  • 3.
  • 4.
  • 5.
  • 6.
  • 7.
  • 8.

3、使用timeit.timeit ()

timeit()函数有5个参数:
   stmt 参数是需要执行的语句,默认为 pass
   setup 参数是用来执行初始化代码或构建环境的语句,默认为 pass
   timer 是计时器,默认是 perf_counter()
   number 是执行次数,默认为一百万
   globals 用来指定要运行代码的命名空间,默认为 None 
 import timeit
 def func():
     print('hello world')
 print(f'耗时: {timeit.timeit(stmt=func, number=1)}')
 hello world
 耗时: 0.0007705999999999824
  • 1.
  • 2.
  • 3.
  • 4.
  • 5.
  • 6.
  • 7.
  • 8.
  • 9.
  • 10.
  • 11.
  • 12.

4、使用装饰器统计

在实际项目代码中,可以通过装饰器方便的统计函数运行耗时。使用装饰器来统计函数执行耗时的好处是对函数的入侵性小,易于编写和修改。

装饰器装饰函数的方案只适用于统计函数的运行耗时,如果有代码块耗时统计的需求就不能用了,这种情况下可以使用 with 语句自动管理上下文。

(1)同步函数的统计

import time 
 def coast_time(func):
     def fun(*args, **kwargs):
         t = time.perf_counter()
         result = func(*args, **kwargs)
         print(f'函数:{func.__name__} 耗时:{time.perf_counter() - t:.8f} s')
         return result
     return fun
 @coast_time
 def test():
     print('hello world')
 if __name__ == '__main__':
     test()
  • 1.
  • 2.
  • 3.
  • 4.
  • 5.
  • 6.
  • 7.
  • 8.
  • 9.
  • 10.
  • 11.
  • 12.
  • 13.

(2)异步函数的统计

import asyncio
 import time
 from asyncio.coroutines import iscoroutinefunction
 def coast_time(func):
     def fun(*args, **kwargs):
         t = time.perf_counter()
         result = func(*args, **kwargs)
         print(f'函数:{func.__name__} 耗时:{time.perf_counter() - t:.8f} s')
         return result
     async def func_async(*args, **kwargs):
         t = time.perf_counter()
         result = await func(*args, **kwargs)
         print(f'函数:{func.__name__} 耗时:{time.perf_counter() - t:.8f} s')
         return result
     if iscoroutinefunction(func):
         return func_async
     else:
         return fun
 @coast_time
 def test():
     print('hello test')
     time.sleep(1)
 @coast_time
 async def test_async():
     print('hello test_async')
     await asyncio.sleep(1)
 if __name__ == '__main__':
     test()
     asyncio.get_event_loop().run_until_complete(test_async())       
 hello test
 函数:test 耗时:1.00230700 s
 hello test_async
 函数:test_async 耗时:1.00572550 s
  • 1.
  • 2.
  • 3.
  • 4.
  • 5.
  • 6.
  • 7.
  • 8.
  • 9.
  • 10.
  • 11.
  • 12.
  • 13.
  • 14.
  • 15.
  • 16.
  • 17.
  • 18.
  • 19.
  • 20.
  • 21.
  • 22.
  • 23.
  • 24.
  • 25.
  • 26.
  • 27.
  • 28.
  • 29.
  • 30.
  • 31.
  • 32.
  • 33.

5、with语句统计

通过实现 enter 和 exit 函数可以在进入和退出上下文时进行一些自定义动作,例如连接或断开数据库、打开或 关闭文件、记录开始或结束时间等,例如:我们用来统计函数块的执行时间。

with语句不仅可以统计代码块的执行时间,也可以统计函数的执行时间,还可以统计多个函数的执行时间之和,相比装饰器来说对代码的入侵性比较大,不易于修改,好处是使用起来比较灵活,不用写过多的重复代码。

import asyncio
 import time 
 class CoastTime(object):
     def __init__(self):
         self.t = 0
     def __enter__(self):
         self.t = time.perf_counter()
         return self
     def __exit__(self, exc_type, exc_val, exc_tb):
         print(f'耗时:{time.perf_counter() - self.t:.8f} s')
 def test():
     print('hello test')
     with CoastTime():
         time.sleep(1)
 async def test_async():
     print('hello test_async')
     with CoastTime():
         await asyncio.sleep(1)
 if __name__ == '__main__':
     test()
     asyncio.get_event_loop().run_until_complete(test_async())
hello test
耗时:1.00723310 s
hello test_async
耗时:1.00366820 s
  • 1.
  • 2.
  • 3.
  • 4.
  • 5.
  • 6.
  • 7.
  • 8.
  • 9.
  • 10.
  • 11.
  • 12.
  • 13.
  • 14.
  • 15.
  • 16.
  • 17.
  • 18.
  • 19.
  • 20.
  • 21.
  • 22.
  • 23.
  • 24.
  • 25.