浅谈Python协程asyncio

时间:2022-09-23 08:59:42

一、协程

官方描述;
协程是子例程的更一般形式。 子例程可以在某一点进入并在另一点退出。 协程则可以在许多不同的点上进入、退出和恢复。 它们可通过 async def 语句来实现。 参见 PEP 492。

  • 协程不是计算机内部提供的,不像进程、线程,由电脑本身提供,它是由程序员人为创造的, 实现函数异步执行。
  • 协程(Coroutine),也可以被称为微线程,是一种用户太内的上下文切换技术,其实就是通过一个线程实现代码块相互切换执行。看上去像子程序,但执行过程中,在子程序内部可中断,然后转而执行别的子程序,在适当的时候再返回来接着执行。例如:
?
1
2
3
4
5
6
7
8
9
10
11
12
# 需要python3.7+
import asyncio
 
 
async def main():
    print('hello')
    await asyncio.sleep(1)
    print('world')
 
asyncio.run(main())
 
# 打印 "hello",等待 1 秒,再打印 "world"

注意:简单地调用一个协程并不会使其被调度执行,

直接main() 调用会有问题:

?
1
2
3
RuntimeWarning: coroutine 'main' was never awaited
  main()
RuntimeWarning: Enable tracemalloc to get the object allocation traceback

 

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
def func1():
    print(1)
    ...
    print(2)
    
def func2():
    print(3)
    ...
    print(4)
 
func1()
func2()
 
# 结果:1 2 3 4

实现协程的方法:

  • greenlet,早期模块【不建议使用】
  • yield关键字,它是python的生成器,具有保存状态,切换到其他函数去执行,再切换回原函数的功能。
  • asyncio装饰器(python3.4引入)
  • async、await 关键字(python3.5)【推荐使用】

1.1 greenlet实现协程

?
1
2
3
# 第三方模块,因此需要安装
 
pip install greenlet
?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
from greenlet import greenlet
 
 
def func1():
    print(1)
    gr2.switch()
    print(2)
    gr2.switch()
 
 
def func2():
    print(3)
    gr1.switch()
    print(4)
 
 
gr1 = greenlet(func1)
gr2 = greenlet(func2)
 
gr1.switch()
 
# 结果:1 3 2 4

1.2 yield关键字

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
def func1():
    yield 1
    yield from func2()
    yield 2
 
 
def func2():
    yield 3
    yield 4
 
f1 = func1()
for item in f1:
    print(item)
    
# 结果:1 3 2 4

1.3 asynico装饰器

python3.4 及之后版本支持

DeprecationWarning: “@coroutine” decorator is deprecated since Python 3.8, use “async def”
翻译:@coroutine"装饰器自Python 3.8起已弃用,请使用"async def"代替

所以这个也不支持。

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
import asyncio
 
@asyncio.coroutine
def func1():
    print(1)
    yield from asyncio.sleep(2# 遇到IO耗时操作,自动切换到tasks中其他任务,比如:网络IO,下载图片
    print(2)
 
@asyncio.coroutine
def func2():
    print(3)
    yield from asyncio.sleep(2# 遇到IO耗时操作,自动切换到tasks中其他任务,比如:网络IO,下载图片
    print(4)
 
tasks = [
    asyncio.ensure_future(func1()),
    asyncio.ensure_future(func2())
]
loop = asyncio.get_event_loop()
loop.run_until_complete(asyncio.wait(tasks))
 
# 结果: 1 3 2 4

1.4 async & await 关键字

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
import asyncio
 
 
async def func1():
    print(1)
    await asyncio.sleep(2# 遇到IO耗时操作,自动切换到tasks中其他任务,比如:网络IO,下载图片
    print(2)
 
 
async def func2():
    print(3)
    await asyncio.sleep(2# 遇到IO耗时操作,自动切换到tasks中其他任务,比如:网络IO,下载图片
    print(4)
 
tasks = [
    asyncio.ensure_future(func1()),
    asyncio.ensure_future(func2())
]
loop = asyncio.get_event_loop()
loop.run_until_complete(asyncio.wait(tasks))

二、协程的意义

充分利用线程。在一个线程中如果遇到IO等待时间线程不会一直等待,利用空闲时间再去干点其他事情。

以下载三张图片为例:

普通方式(同步)下载:

?
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
import time
import requests
 
def download_image(url, img_name):
    print("开始下载:", url)
    # 发送网络请求,下载图片
    response = requests.get(url)
    print("下载完成")
    # 图片保存到本地文件
    file_name = str(img_name) + '.png'
    with open(file_name, mode='wb') as file:
        file.write(response.content)
 
if __name__ == '__main__':
    start = time.time()
    url_list = [
        'https://tse4-mm.cn.bing.net/th/id/OIP.866vRxQ8QvyDsrUuXiu7qwHaNK?w=182&h=324&c=7&o=5&pid=1.7',
        'https://tse2-mm.cn.bing.net/th/id/OIP.HUcWtoYPG-z2pu4ityajbAHaKQ?w=182&h=252&c=7&o=5&pid=1.7',
        'https://tse2-mm.cn.bing.net/th/id/OIP.MvncR0-Pt9hVxKTdrvD9dAHaNK?w=182&h=324&c=7&o=5&pid=1.7',
        'https://tse1-mm.cn.bing.net/th/id/OIP._nGloaeMWbL7NB7Lp6SnXQHaLH?w=182&h=273&c=7&o=5&pid=1.7',
        ]
    img_name = 1
    for item in url_list:
        download_image(item, img_name)
        img_name += 1
    end = time.time()
    print(end - start)
    
 # 最终时间:7.25s

协程方式(异步)下载:

?
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
34
35
import aiohttp
import asyncio
import time
 
 
async def fetch(session, url):
    print("发送请求:", url)
 
    async with session.get(url, verify_ssl=False) as response:
        content = await response.content.read()
        file_name = url.rsplit('_')[-1]
        # print(file_name)
        with open(file_name, mode='wb') as file_object:
            file_object.write(content)
        print("下载完成")
 
 
async def main():
    async with aiohttp.ClientSession() as session:
        url_list = [
            'https://www3.autoimg.cn/newsdfs/g26/M02/35/A9/120x90_0_autohomecar__ChsEe12AXQ6AOOH_AAFocMs8nzU621.jpg',
            'https://www3.autoimg.cn/newsdfs/g26/M02/35/A9/120x90_0_autohomecar__ChsEe12AXQ6AOOH_AAFocMs8nzU621.jpg',
            'https://www3.autoimg.cn/newsdfs/g26/M02/35/A9/120x90_0_autohomecar__ChsEe12AXQ6AOOH_AAFocMs8nzU621.jpg',
            'https://www3.autoimg.cn/newsdfs/g26/M02/35/A9/120x90_0_autohomecar__ChsEe12AXQ6AOOH_AAFocMs8nzU621.jpg',
        ]
        tasks = [asyncio.ensure_future(fetch(session, url)) for url in url_list]
        await asyncio.wait(tasks)
 
if __name__ == '__main__':
    start = time.time()
    asyncio.get_event_loop().run_until_complete(main())
    end = time.time()
    print(end - start)
    
# 结果:0.05s

到此这篇关于浅谈Python协程的文章就介绍到这了,更多相关Python协程内容请搜索服务器之家以前的文章或继续浏览下面的相关文章希望大家以后多多支持服务器之家!

原文链接:https://blog.****.net/zhb_feng/article/details/117930310