多进程multiprocessing
multiprocessing is a package that supports spawning processes using an API similar to the threading module. The multiprocessing package offers both local and remote concurrency（并发）, effectively side-stepping（有效的避过） the Global Interpreter Lock by using subprocesses instead of threads. Due to this, the multiprocessing module allows the programmer to fully leverage（充分利用） multiple processors on a given machine. It runs on both Unix and Windows.
多进程的用法和多线程差不多
import multiprocessing
def run():
print(__name__)
print("prcess is running")
if __name__ == "__main__": #On Windows the subprocesses will import (i.e. execute) the main module at start. You need to protect the main code like this to avoid creating subprocesses recursively
p = multiprocessing.Process(target=run)
p.start()
结果：
__mp_main__
prcess is running
To show the individual process IDs involved, here is an expanded example:
import multiprocessing,threading
import os
def info(title):
print(title)
print("module name:",__name__)
print("parent process:",os.getppid()) #父进程
print("process id:",os.getpid()) #当前进程id
print(threading.current_thread()) #打印当前线程id
def f(name):
info(" 33[31;1m function f 33[0m")
print("hello",name)
if __name__ == "__main__":
info(" 33[32;1mmain process line 33[0m")
p = multiprocessing.Process(target=f,args=("lxj",))
p.start()
multiprocessing.Queue()
结果:
main process line
module name: __main__
parent process: 7388
process id: 13352
<_MainThread(MainThread, started 5828)>
function f
module name: __mp_main__
parent process: 13352
process id: 608
<_MainThread(MainThread, started 2948)>
hello lxj
我们直接运行info，也有主进程id，那这个主进程pid是怎么产生的呢？查看windows任务管理器，可以看到，7388pid号即pycharm
这里验证了我们之前进程是由线程创建的，我们称为主线程。
进程间通信
先来看下线程间通讯
#子线程共享所属进程的内存空间
def f():
q.put(1)
if __name__ == "__main__":
q = queue.Queue() #线程的队列
p = threading.Thread(target=f)
p.start()
print(q.get())
p.join()
结果:
1
j
进程间是不共享内存空间的，看下进程间是如何通信的
Queue
from multiprocessing import Queue,Process
import threading
import queue
#换成多进程
def f():
q.put(1)
if __name__ == "__main__":
q = queue.Queue()
p = Process(target=f)
p.start()
print(q.get())
p.join()
结果：
报错
NameError: name 'q' is not defined
因为进程间的内存是不共享的
对上面的程序做修改
def f(q):
q.put(1)
if __name__ == "__main__":
q = Queue() #换成进程间的Queue
p = Process(target=f,args=(q,)) #把q传到子进程
p.start()
print(q.get())
p.join()
结果：
1
进程间通讯Pipe
The Pipe() function returns a pair of connection objects connected by a pipe which by default is duplex (two-way). For example:
from multiprocessing import Process,Pipe
import os
def f(conn):
conn.send([1,2,3])
data = conn.recv()
print(os.getpid(),os.getppid(),data)
if __name__ == "__main__":
parent_conn,child_conn = Pipe() #两端数据
p = Process(target=f,args=(parent_conn,)) #这里传入一端，随便哪个都行
p.start()
data = child_conn.recv() #使用另外一个来接收，有点类似socket客户端服务器模型,可以发多条，收多条
print(os.getpid(), os.getppid(), data)
child_conn.send(["asd"])
p.join()
结果：
5176 11420 [1, 2, 3]
7304 5176 ['asd']
Queue和Pipe是实现进程间数据的传递，那么如何能使数据共享呢？答案使使用Mangers
A manager object returned by Manager() controls a server process which holds Python objects and allows other processes to manipulate them using proxies.
A manager returned by Manager() will support types list, dict, Namespace, Lock, RLock, Semaphore, BoundedSemaphore, Condition, Event, Barrier, Queue, Value and Array. For example
from multiprocessing import Process,Manager
import os
def func(d,l):
d[os.getpid()] = os.getppid()
l.append(os.getpid())
if __name__ == "__main__":
with Manager() as manger:
d = manger.dict() #生成一个字典
l = manger.list(range(3))
p_list = []
for i in range(10):
p = Process(target=func,args=(d,l))
p.start()
p_list.append(p)
for p in p_list: #等待进程结束
p.join()
print(d)
print(l)
结果：
{1232: 4656, 13656: 4656, 11668: 4656, 9368: 4656, 2516: 4656, 13772: 4656, 12316: 4656, 6764: 4656, 11788: 4656, 8756: 4656}
[0, 1, 2, 2516, 11668, 12316, 9368, 13656, 13772, 11788, 6764, 8756, 1232]
可以发现实现了进程间的数据共享
进程同步
当要用到同一块屏幕时，需要用到锁
Without using the lock output from the different processes is liable to get all mixed up.
from multiprocessing import Process, Lock
def f(l, i):
l.acquire()
try:
print('hello world', i)
finally:
l.release()
if __name__ == '__main__':
lock = Lock()
for num in range(10):
Process(target=f, args=(lock, num)).start()
进程池
进程池内部维护一个进程序列，当使用时，则去进程池中获取一个进程，如果进程池序列中没有可供使用的进进程，那么程序就会等待，直到进程池中有可用进程为止。
进程池中有两个方法：
- apply
- apply_async
apply（同步）用法
from multiprocessing import Process,Pool
import os,time
def foo(i):
time.sleep(1)
print("in process",os.getpid())
return i +100
if __name__ == "__main__":
pool = Pool(5)
for i in range(10):
pool.apply(func=foo,args=(i,))
print("end")
pool.close()
pool.join()
结果：是串行的执行方式
apply_async（异步）用法
from multiprocessing import Process,Pool
import os,time
def foo(i):
time.sleep(1)
print("in process",os.getpid())
return i +100
if __name__ == "__main__":
pool = Pool(5)
for i in range(10):
# pool.apply(func=foo,args=(i,))
pool.apply_async(func=foo,args=(i,))
pool.close()
pool.join() #这里必须先close再join（可以看源码），如果这里只close，不join则主进程不会等待进程池执行完，会看到结果直接打印end
print("end")
#结果会看到5个进程，5个进程的执行效果
from multiprocessing import Process,Pool
import os,time
def foo(i):
time.sleep(1)
print("in process",i+100,os.getpid())
return i +100
def bar(arg): #arg参数为foo返回的结果
print("-->exex done:",arg,os.getpid())
if __name__ == "__main__":
pool = Pool(5)
print(os.getpid())
for i in range(10):
# pool.apply(func=foo,args=(i,))
pool.apply_async(func=foo,args=(i,),callback=bar) #回调函数，当foo函数执行完成，执行回调函数bar
pool.close()
pool.join() #这里必须先close再join（可以看源码），如果这里close，不join则主进程不会等待进程池执行完
print("end")
#回调函数的用法
结果：
12328
in process 100 4160
-->exex done: 100 12328
in process 101 1304
-->exex done: 101 12328
in process 102 1988
-->exex done: 102 12328
in process 103 4744
-->exex done: 103 12328
in process 104 9460
-->exex done: 104 12328
in process 105 4160
-->exex done: 105 12328
in process 106 1304
-->exex done: 106 12328
in process 107 1988
-->exex done: 107 12328
in process 108 4744
-->exex done: 108 12328
in process 109 9460
-->exex done: 109 12328
end
每次执行的结果都不一样，上面的结果我们可以得到回调函数是主进程执行的