一、协程
1、本质上是一个线程
2、能够在多个任务之间切换来节省一些IO时间
3、协程中任务之间的切换也消耗时间,但是开销要远远小于进程线程之间的切换,实现并发的手段
def consumer():
while True:
x = yield # 生成器函数
print('处理了数据:', x)
def producer():
c = consumer()
next(c)
for i in range(4):
print('生产了数据:', i)
c.send(i) # 传送数据,会执行consumer()
producer()
输出结果:
生产了数据: 0
处理了数据: 0
生产了数据: 1
处理了数据: 1
生产了数据: 2
处理了数据: 2
生产了数据: 3
处理了数据: 3
2、真正的协程模块就是使用greenlet完成的切换
from greenlet import greenlet
def eat():
print('eating start')
g2.switch() # 切换到g2
print('eating end')
def play():
print('playing start')
g1 = greenlet(eat)
g2 = greenlet(play)
g1.switch() # 切换到g1
输出结果
eating start
playing start
3、进程 + 线程 + 协程
5 20 500
一个进程里面有20个线程可总共开5 * 20 * 500个协程
import time
import gevent # 感知不到time.sleep,能感知到gevent.sleep
def eat():
print('eating start')
# time.sleep(1)
gevent.sleep(1)
print('eating end')
def play():
print('playing start')
# time.sleep(1)
gevent.sleep(1)
print('playing end')
g1 = gevent.spawn(eat)
g2 = gevent.spawn(play)
g1.join()
g2.join()
输出结果
eating start
playing start
eating end
playing end
from gevent import monkey;
monkey.patch_all() # 可以将所有的阻塞操作打包,就可以感知到
import threading
import time
import gevent # 感知不到time.sleep,能感知到gevent.sleep
def eat():
print(threading.current_thread().getName())
print(threading.current_thread()) # 可以看出是一个虚拟的线程
print('eating start')
time.sleep(1)
# gevent.sleep(1)
print('eating end')
def play():
print(threading.current_thread().getName())
print(threading.current_thread())
print('playing start')
time.sleep(2)
# gevent.sleep(1)
print('playing end')
g1 = gevent.spawn(eat)
g2 = gevent.spawn(play)
g1.join()
g2.join()
输出结果:
DummyThread - 1 < _DummyThread(DummyThread - 1, started daemon 1772404630344) >
eating start
DummyThread - 2 < _DummyThread(DummyThread - 2, started daemon 1772404630600) >
playing start
eating end
playing end
进程和线程的任务切换由操作系统完成,协程任务之间切换由程序(代码)完成,只有遇到协程模块功能识别的IO操作的时候,程序才会进行任务切换,实现并发的效果
4、同步和异步
from gevent import monkey;
monkey.patch_all()
import time
import gevent
def task():
time.sleep(1)
print(12345)
def sync():
for i in range(5):
task()
def async(): # 效率很高
g_list = []
for i in range(10):
g = gevent.spawn(task)
g_list.append(g)
gevent.joinall(g_list)
sync()
async()
5、协程更适合网络操作中(在一个线程中实现并发效果的概念)能够规避一些任务中的IO操作,在任务的执行过程中,检测到IO就切换到其他任务。
(1)爬虫的例子(请求过程中的IO等待)
from gevent import monkey;
monkey.patch_all()
import gevent
import requests
url = 'http://www.baidu.com'
def get_url(urls):
res = requests.get(urls)
return len(res.content.decode('utf-8'))
g1 = gevent.spawn(get_url, 'http://www.baidu.com')
g2 = gevent.spawn(get_url, 'http://www.sogou.com')
g3 = gevent.spawn(get_url, 'http://www.taobao.com')
g4 = gevent.spawn(get_url, 'http://www.hao123.com')
g5 = gevent.spawn(get_url, 'http://www.cnblogs.com')
gevent.joinall([g1, g2, g3, g4, g5])
print(g1.value)
print(g2.value)
print(g3.value)
print(g4.value)
print(g5.value)
输出结果:
2287
23446
149538
487625
41855
(2)socket server
server端:
from gevent import monkey;
monkey.patch_all()
import socket
import gevent
def talk(conn):
conn.send(b'hello')
print(conn.recv(1024).decode('utf-8'))
conn.close()
sk = socket.socket()
sk.bind(('127.0.0.1', 8080))
sk.listen(5)
while True:
conn, addr = sk.accept()
gevent.spawn(talk, conn)
sk.close()
client端:
import socket
sk = socket.socket()
sk.connect(('127.0.0.1', 8080))
print(sk.recv(1024).decode('utf-8'))
msg = input('>>').encode('utf-8')
sk.send(msg)
sk.close()
(1)多线程被弱化了协程
(2)在一个线程上提高CPU的利用率
(3)协程相比于多线程的优势切换的效率更快了