await
、asyncio.Queue
等)来实现数据交换,简化了并发编程的复杂性。await
关键字等待 I/O 操作完成时,可以在等待期间释放 CPU,执行其他协程任务,提高整体的并发性能。创建协程函数:async def func()
运行协程函数:asyncio.run(func())
注意:当调用协程函数func()
后,内部代码是不会执行的,只是得到了一个协程对象,如果想要运行协程对象,则必须将其交给事件循环来处理
import asyncio
async def func():
print("Hello, coroutine!")
# 调用协程
asyncio.run(func())
也可以这么写
import asyncio
async def func():
print("Hello, coroutine!")
# 创建事件循环对象
loop = asyncio.get_event_loop()
# 将事件封装为任务
task = loop.create_task(func())
# 运行事件直到任务完成
loop.run_until_complete(task)
.wait()
和.gather
await asyncio.wait(tasks)
:接受一个任务集合作为参数,并等待所有任务完成。返回两个集合(Sets):已完成的任务集合和仍在进行中的任务集合。
await asyncio.gather(*tasks)
:接受一个任务集合作为参数,并等待所有任务完成。返回每个任务的实际返回值
await task
:执行单个任务,返回每个任务的实际返回值
await
关键字后面可以定义可等待对象,例如协程对象,Future,Task对象
此处的可等待对象其实就是I/O阻塞,当await
包裹的协程任务遇到阻塞时会自动从当前任务切换到另一个任务中,以节省时间和内存
result = await
表示result就是await后面的指令运行完毕后得到的结果
import asyncio
async def fun1():
print('1')
await asyncio.sleep(2)
return '结束'
async def main():
# 创建任务
task1 = asyncio.create_task(fun1())
task2 = asyncio.create_task(fun1())
# 创建事件循环
res1 = await task1
res2 = await task2
print(res1, res2)
asyncio.run(main())
也可以这么写
import asyncio
async def fun1():
print('1')
await asyncio.sleep(2)
return '结束'
async def main():
# 创建任务
task = [asyncio.create_task(fun1()) for i in range(10)]
# 创建事件循环
res = await asyncio.gather(task)
print(res)
asyncio.run(main())
数量太少看不出效率,但是可以证明await
会等待所有任务返回结果后再继续往下运行
aiohttp
与requests
相比最大的区别就是aiohttp
支持异步操作,因此用协程编写爬虫时aiohttp
是相当重要的一个模块
aiohttp.ClientSession()
:
session.get(url)
和 session.post(url)
:
ClientSeesion
对象上调用这些方法可以发送GET/POST请求url
作为参数传递,一般就是访问的主网址response.status
和 response.text()
:
ClientResponse
对象的属性和方法response.status
返回响应状态码(如200、404等)response.status
返回响应内容的文本字符串response.json()
:
async with session.get(url) as response
和 async with session.post(url) as response
:
async with
语法结构,可以在异步上下文管理器中发送请求和处理响应response
是一个异步上下文管理器返回的响应对象,可以执行response.status
之类的操作,并且使其能够被await
包裹利用aiohttp
模块获取一个简单的浏览器响应
import asyncio
import aiohttp
from lxml import etree
url = 'https://www.baidu.com'
async def main():
connector = aiohttp.TCPConnector(ssl=False)
async with aiohttp.ClientSession(connector=connector) as session:
async with session.get(url) as response:
res = await response.text()
et = etree.HTML(res)
print(et)
if __name__ == '__main__':
asyncio.run(main())
aiofiles
与python中常用的with open操作类似,并且支持异步操作,且与asyncio
配合良好
具体操作也和with open类似
读
async def read_file():
async with aiofiles.open('file.txt', mode='r') as file:
contents = await file.read()
print(contents)
if __name__ == '__main__':
asyncio.run(read_file())
写
async def write_file():
async with aiofiles.open('file.txt', mode='w') as file:
await file.write('Hello, World!')
if __name__ == '__main__':
asyncio.run(write_file())
掏出上次我们写的线程池爬虫案例:
from concurrent.futures import ThreadPoolExecutor
import requests
from lxml import etree
url = 'https://loryx.wiki/%E6%B5%8F%E8%A7%88/%E7%89%8C%E5%BA%93'
def download(name, src):
with open(name, 'wb') as f:
f.write(requests.get(src).content)
print(f'{name}已下载')
def main():
res = requests.get(url=url)
res.encoding = 'utf-8'
et = etree.HTML(res.text)
src = et.xpath("//td[@class='col15 leftalign']/a/@href")
name = et.xpath("//td[@class='col0 leftalign']/text()")
for i, index in enumerate(name):
name[i] = index.strip()
with ThreadPoolExecutor(64) as t:
for i in range(len(src)):
file_name = f"img/{name[i]}.png"
t.submit(download, file_name, src[i])
if __name__ == '__main__':
main()
现在开始改写
首先跟requests
相关的可以全部删了换成aiohttp
,比如
res = requests.get(url=url)
res.encoding = 'utf-8'
et = etree.HTML(res.text)
替换为
async with aiohttp.ClientSession(connector=connector) as session:
async with session.get(url) as response:
res = await response.text()
et = etree.HTML(res)
然后ThreadPoolExecutor
相关的也可以全部用asyncio
替换
for i, index in enumerate(name):
name[i] = index.strip()
with ThreadPoolExecutor(64) as t:
for i in range(len(src)):
file_name = f"img/{name[i]}.png"
t.submit(download, file_name, src[i])
替换为
tasks = []
for i in range(len(src)):
file_name = f"img/{name[i]}.png"
if not os.path.exists('img'):
os.makedirs('img')
task = asyncio.create_task(spider(file_name, src[i]))
tasks.append(task)
await asyncio.gather(*tasks)
文件读写的部分也可以用aiofiles
重写
def download(name, src):
with open(name, 'wb') as f:
f.write(requests.get(src).content)
print(f'{name}已下载')
替换成
async def spider(name, src):
connector = aiohttp.TCPConnector(ssl=False)
async with aiohttp.ClientSession(connector=connector) as session:
async with session.get(src) as response:
count = await response.read()
async with aiofiles.open(name, 'wb') as f:
await f.write(count)
print(f'{name}已下载')
import asyncio
import os.path
import aiofiles
import aiohttp
from lxml import etree
url = 'https://loryx.wiki/%E6%B5%8F%E8%A7%88/%E7%89%8C%E5%BA%93'
async def spider(name, src):
# 关闭SSL证书验证
connector = aiohttp.TCPConnector(ssl=False)
# 创建图片链接对象
async with aiohttp.ClientSession(connector=connector) as session:
async with session.get(src) as response:
# 读取图片信息 准备写入本地
count = await response.read()
# 写入本地 下载时遇到io阻塞自动跳转其他任务
async with aiofiles.open(name, 'wb') as f:
await f.write(count)
print(f'{name}已下载')
async def main():
# 关闭SSL证书验证
connector = aiohttp.TCPConnector(ssl=False)
# 创建异步HTTP客户端对象
async with aiohttp.ClientSession(connector=connector) as session:
# 发送get请求
async with session.get(url) as response:
# 返回响应内容的字符串
res = await response.text()
et = etree.HTML(res)
src = et.xpath("//td[@class='col15 leftalign']/a/@href") # 图片链接
temp_name = et.xpath("//td[@class='col0 leftalign']/text()") # 图片名称
for i, index in enumerate(temp_name):
temp_name[i] = index.strip()
# 任务列表
tasks = []
for i in range(len(src)):
# 下载到本地的名称
file_name = f"img/{temp_name[i]}.png"
if not os.path.exists('img'):
os.makedirs('img')
# 批量创建asyncio异步任务 执行spider函数
task = asyncio.create_task(spider(file_name, src[i]))
tasks.append(task)
# 启动
await asyncio.wait(tasks)
if __name__ == '__main__':
asyncio.run(main())
(src)):
# 下载到本地的名称
file_name = f"img/{temp_name[i]}.png"
if not os.path.exists('img'):
os.makedirs('img')
# 批量创建asyncio异步任务 执行spider函数
task = asyncio.create_task(spider(file_name, src[i]))
tasks.append(task)
# 启动
await asyncio.wait(tasks)
if __name__ == '__main__':
asyncio.run(main())