[Python 多线程] Barrier (十一)

Barrier

栅栏，也叫屏障。可以想象成路障、道闸。

Python 3.2引入的新功能。

构造方法：

threading.Barrier(parties, action=None, timeout=None)

构建Barrier对象，parties 指定参与方数目，timeout是wait方法未指定时超时的默认值。

n_waiting    当前在栅栏中等待的线程数

parties        通过栅栏所需的线程数

wait(timeout=None) 等待通过栅栏，返回0到线程数-1的整数(barrier_id)，每个线程返回不同。如果wait方法设置了超时，并超时发送，栅栏将处于broken状态。

例1：

#Barrier 栅栏
import threading,logging
logging.basicConfig(level=logging.INFO,format="[-] %(threadName)s %(message)s")

def work(barrier:threading.Barrier):
    logging.info("n_waiting = {}".format(barrier.n_waiting))   # 等待的线程数
    bid = barrier.wait()   # 参与者的id，返回0到线程数减1的数值
    logging.info("after barrier {}".format(bid))  # 栅栏之后

barrier = threading.Barrier(3) # 3个参与者，每3个开闸放行，0,1,2  4,5,6

for x in range(1,4):  # 所有参数者个数，4,5,6,10,15
    threading.Event().wait(1)
    threading.Thread(target=work,args=(barrier,),name="Barrier-{}".format(x)).start()

运行结果：
[-] Barrier-1 n_waiting = 0
[-] Barrier-2 n_waiting = 1
[-] Barrier-3 n_waiting = 2
[-] Barrier-3 after barrier 2
[-] Barrier-2 after barrier 1
[-] Barrier-1 after barrier 0

　　每一个进来就等待，不够3个就阻塞，直到够3个就开闸放行。

Barrier实例的方法：

broken  检测栅栏是否处于打破的状态，返回True或False

abort()  将栅栏置于broken状态，等待中的线程或者调用等待方法的线程都会抛出threading.BrokenBarrieError异常，直到reset方法来恢复栅栏

reset()  恢复栅栏，重新开始拦截

例2：

#Barrier 栅栏
import threading,logging
logging.basicConfig(level=logging.INFO,format="[-] %(threadName)s %(message)s")

def work(barrier:threading.Barrier):
    logging.info("n_waiting = {}".format(barrier.n_waiting))
    try:
        bid = barrier.wait()
        logging.info("after barrier {}".format(bid))
    except threading.BrokenBarrierError:
        logging.info("Broken Barrier in {}".format(threading.current_thread()))

barrier = threading.Barrier(3)

for x in range(1,12): #12个
    if x == 3:
        barrier.abort() #有一个人坏了规矩
    elif x == 6:
        barrier.reset()
    threading.Event().wait(1)
    threading.Thread(target=work,args=(barrier,),name="Barrier-{}".format(x)).start()

运行结果：
[-] Barrier-1 n_waiting = 0 #0,1
[-] Barrier-2 n_waiting = 1
[-] Barrier-2 Broken Barrier in <Thread(Barrier-2, started 3124)>
[-] Barrier-1 Broken Barrier in <Thread(Barrier-1, started 8036)>
[-] Barrier-3 n_waiting = 0
[-] Barrier-3 Broken Barrier in <Thread(Barrier-3, started 7428)>
[-] Barrier-4 n_waiting = 0
[-] Barrier-4 Broken Barrier in <Thread(Barrier-4, started 1828)>
[-] Barrier-5 n_waiting = 0
[-] Barrier-5 Broken Barrier in <Thread(Barrier-5, started 7416)>
[-] Barrier-6 n_waiting = 0 #6,7,8
[-] Barrier-7 n_waiting = 1
[-] Barrier-8 n_waiting = 2
[-] Barrier-8 after barrier 2
[-] Barrier-7 after barrier 1
[-] Barrier-6 after barrier 0
[-] Barrier-9 n_waiting = 0  #9,10,11
[-] Barrier-10 n_waiting = 1
[-] Barrier-11 n_waiting = 2
[-] Barrier-11 after barrier 2
[-] Barrier-9 after barrier 0
[-] Barrier-10 after barrier 1

　　一共有12个参与者，依次开始，1和2处于等待状态，到达第3的时候，进入了broken状态，则直到第6个，才恢复栅栏，从6开始继续拦截，达到3个(6,7,8)就放行，9,10,11也放行。

例3：

wait方法

#Barrier 栅栏
import threading,logging
logging.basicConfig(level=logging.INFO,format="[-] %(threadName)s %(message)s")

def work(barrier:threading.Barrier,i:int):
    logging.info("n_waiting = {}".format(barrier.n_waiting))
    try:
        if i < 3:
            bid = barrier.wait(1)  #超时1秒就将栅栏置于broken状态，抛出异常后续语句不会执行
        else:
            if i == 6:
                barrier.reset() #恢复栅栏
            bid = barrier.wait()
        # logging.info("broken status = {}".format(barrier.broken))  #是否处于broken状态
        logging.info("after barrier {}".format(bid))
    except threading.BrokenBarrierError:
        logging.info("Broken Barrier in {}".format(threading.current_thread()))

barrier = threading.Barrier(3)

for i in range(1,11): #10个
    threading.Event().wait(2) #强制延迟2秒,让出时间片
    threading.Thread(target=work,args=(barrier,i),name="Barrier-{}".format(i)).start()

运行结果：
[-] Barrier-1 n_waiting = 0
[-] Barrier-1 Broken Barrier in <Thread(Barrier-1, started 3100)>
[-] Barrier-2 n_waiting = 0
[-] Barrier-2 Broken Barrier in <Thread(Barrier-2, started 8836)>
[-] Barrier-3 n_waiting = 0
[-] Barrier-3 Broken Barrier in <Thread(Barrier-3, started 8428)>
[-] Barrier-4 n_waiting = 0
[-] Barrier-4 Broken Barrier in <Thread(Barrier-4, started 1204)>
[-] Barrier-5 n_waiting = 0
[-] Barrier-5 Broken Barrier in <Thread(Barrier-5, started 1556)>
[-] Barrier-6 n_waiting = 0
[-] Barrier-7 n_waiting = 1
[-] Barrier-8 n_waiting = 2
[-] Barrier-8 after barrier 2
[-] Barrier-7 after barrier 1
[-] Barrier-6 after barrier 0
[-] Barrier-9 n_waiting = 0
[-] Barrier-10 n_waiting = 1
阻塞中

　　wait方法在等待超时1秒后，就强制将栅栏置于broken状态，直到第6个的时候才reset恢复，然后6,7,8放行，9,10,继续阻塞。如果此时有第11个，就会9,10,11放行。

应用场景：

并发初始化

所有线程都必须初始化完成后，才能继续工作，例如运行前加载数据，检查，如果这些工作没完成就不能正常工作运行。

10个线程做10种工作准备，每个线程负责一种工作，只有10个线程都完成后，才能继续工作，先完成的要等待后完成的线程。

例如，启动一个程序，需要先加载磁盘文件、缓存预热、初始化连接池等工作，这些工作可以齐头并进，不过只有都满足了，程序才能继续向后执行。假设数据库链接失败，则初始化工作失败，就要abort，栅栏broken，所有线程收到异常退出。

工作量

有10个计算任务，完成6个，就算工作完成。

PPTBYG