使用Python的Twisted框架编写非阻塞程序的代码示例

时间:2022-10-16 20:06:16

先来看一段代码:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
# ~*~ Twisted - A Python tale ~*~
 
from time import sleep
 
# Hello, I'm a developer and I mainly setup Wordpress.
def install_wordpress(customer):
  # Our hosting company Threads Ltd. is bad. I start installation and...
  print "Start installation for", customer
  # ...then wait till the installation finishes successfully. It is
  # boring and I'm spending most of my time waiting while consuming
  # resources (memory and some CPU cycles). It's because the process
  # is *blocking*.
  sleep(3)
  print "All done for", customer
 
# I do this all day long for our customers
def developer_day(customers):
  for customer in customers:
    install_wordpress(customer)
 
developer_day(["Bill", "Elon", "Steve", "Mark"])

运行一下,结果如下所示:

?
1
$ ./deferreds.py 1
?
1
2
3
4
5
6
------ Running example 1 ------
Start installation for Bill
All done for Bill
Start installation
...
* Elapsed time: 12.03 seconds

这是一段顺序执行的代码。四个消费者,为一个人安装需要3秒的时间,那么四个人就是12秒。这样处理不是很令人满意,所以看一下第二个使用了线程的例子:

?
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
import threading
 
# The company grew. We now have many customers and I can't handle the
# workload. We are now 5 developers doing exactly the same thing.
def developers_day(customers):
  # But we now have to synchronize... a.k.a. bureaucracy
  lock = threading.Lock()
  #
  def dev_day(id):
    print "Goodmorning from developer", id
    # Yuck - I hate locks...
    lock.acquire()
    while customers:
      customer = customers.pop(0)
      lock.release()
      # My Python is less readable
      install_wordpress(customer)
      lock.acquire()
    lock.release()
    print "Bye from developer", id
  # We go to work in the morning
  devs = [threading.Thread(target=dev_day, args=(i,)) for i in range(5)]
  [dev.start() for dev in devs]
  # We leave for the evening
  [dev.join() for dev in devs]
 
# We now get more done in the same time but our dev process got more
# complex. As we grew we spend more time managing queues than doing dev
# work. We even had occasional deadlocks when processes got extremely
# complex. The fact is that we are still mostly pressing buttons and
# waiting but now we also spend some time in meetings.
developers_day(["Customer %d" % i for i in xrange(15)])

运行一下:

?
1
$ ./deferreds.py 2
?
1
2
3
4
5
6
7
------ Running example 2 ------
Goodmorning from developer 0Goodmorning from developer
1Start installation forGoodmorning from developer 2
Goodmorning from developer 3Customer 0
...
from developerCustomer 13 3Bye from developer 2
* Elapsed time: 9.02 seconds

这次是一段并行执行的代码,使用了5个工作线程。15个消费者每个花费3s意味着总共45s的时间,不过用了5个线程并行执行总共只花费了9s的时间。这段代码有点复杂,很大一部分代码是用于管理并发,而不是专注于算法或者业务逻辑。另外,程序的输出结果看起来也很混杂,可读性也天津市。即使是简单的多线程的代码同样也难以写得很好,所以我们转为使用Twisted:

?
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
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
# For years we thought this was all there was... We kept hiring more
# developers, more managers and buying servers. We were trying harder
# optimising processes and fire-fighting while getting mediocre
# performance in return. Till luckily one day our hosting
# company decided to increase their fees and we decided to
# switch to Twisted Ltd.!
 
from twisted.internet import reactor
from twisted.internet import defer
from twisted.internet import task
 
# Twisted has a slightly different approach
def schedule_install(customer):
  # They are calling us back when a Wordpress installation completes.
  # They connected the caller recognition system with our CRM and
  # we know exactly what a call is about and what has to be done next.
  #
  # We now design processes of what has to happen on certain events.
  def schedule_install_wordpress():
      def on_done():
        print "Callback: Finished installation for", customer
    print "Scheduling: Installation for", customer
    return task.deferLater(reactor, 3, on_done)
  #
  def all_done(_):
    print "All done for", customer
  #
  # For each customer, we schedule these processes on the CRM
  # and that
  # is all our chief-Twisted developer has to do
  d = schedule_install_wordpress()
  d.addCallback(all_done)
  #
  return d
 
# Yes, we don't need many developers anymore or any synchronization.
# ~~ Super-powered Twisted developer ~~
def twisted_developer_day(customers):
  print "Goodmorning from Twisted developer"
  #
  # Here's what has to be done today
  work = [schedule_install(customer) for customer in customers]
  # Turn off the lights when done
  join = defer.DeferredList(work)
  join.addCallback(lambda _: reactor.stop())
  #
  print "Bye from Twisted developer!"
# Even his day is particularly short!
twisted_developer_day(["Customer %d" % i for i in xrange(15)])
 
# Reactor, our secretary uses the CRM and follows-up on events!
reactor.run()

运行结果:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
------ Running example 3 ------
Goodmorning from Twisted developer
Scheduling: Installation for Customer 0
....
Scheduling: Installation for Customer 14
Bye from Twisted developer!
Callback: Finished installation for Customer 0
All done for Customer 0
Callback: Finished installation for Customer 1
All done for Customer 1
...
All done for Customer 14
* Elapsed time: 3.18 seconds

这次我们得到了完美的执行代码和可读性强的输出结果,并且没有使用线程。我们并行地处理了15个消费者,也就是说,本来需要45s的执行时间在3s之内就已经完成。这个窍门就是我们把所有的阻塞的对sleep()的调用都换成了Twisted中对等的task.deferLater()和回调函数。由于现在处理的操作在其他地方进行,我们就可以毫不费力地同时服务于15个消费者。
前面提到处理的操作发生在其他的某个地方。现在来解释一下,算术运算仍然发生在CPU内,但是现在的CPU处理速度相比磁盘和网络操作来说非常快。所以给CPU提供数据或者从CPU向内存或另一个CPU发送数据花费了大多数时间。我们使用了非阻塞的操作节省了这方面的时间,例如,task.deferLater()使用了回调函数,当数据已经传输完成的时候会被激活。
另一个很重要的一点是输出中的Goodmorning from Twisted developer和Bye from Twisted developer!信息。在代码开始执行时就已经打印出了这两条信息。如果代码如此早地执行到了这个地方,那么我们的应用真正开始运行是在什么时候呢?答案是,对于一个Twisted应用(包括Scrapy)来说是在reactor.run()里运行的。在调用这个方法之前,必须把应用中可能用到的每个Deferred链准备就绪,然后reactor.run()方法会监视并激活回调函数。
注意,reactor的主要一条规则就是,你可以执行任何操作,只要它足够快并且是非阻塞的。
现在好了,代码中没有那么用于管理多线程的部分了,不过这些回调函数看起来还是有些杂乱。可以修改成这样:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
# Twisted gave us utilities that make our code way more readable!
@defer.inlineCallbacks
def inline_install(customer):
  print "Scheduling: Installation for", customer
  yield task.deferLater(reactor, 3, lambda: None)
  print "Callback: Finished installation for", customer
  print "All done for", customer
 
def twisted_developer_day(customers):
  ... same as previously but using inline_install() instead of schedule_install()
 
twisted_developer_day(["Customer %d" % i for i in xrange(15)])
reactor.run()

运行的结果和前一个例子相同。这段代码的作用和上一个例子是一样的,但是看起来更加简洁明了。inlineCallbacks生成器可以使用一些一些Python的机制来使得inline_install()函数暂停或者恢复执行。inline_install()函数变成了一个Deferred对象并且并行地为每个消费者运行。每次yield的时候,运行就会中止在当前的inline_install()实例上,直到yield的Deferred对象完成后再恢复运行。
现在唯一的问题是,如果我们不止有15个消费者,而是有,比如10000个消费者时又该怎样?这段代码会同时开始10000个同时执行的序列(比如HTTP请求、数据库的写操作等等)。这样做可能没什么问题,但也可能会产生各种失败。在有巨大并发请求的应用中,例如Scrapy,我们经常需要把并发的数量限制到一个可以接受的程度上。在下面的一个例子中,我们使用task.Cooperator()来完成这样的功能。Scrapy在它的Item Pipeline中也使用了相同的机制来限制并发的数目(即CONCURRENT_ITEMS设置):

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
@defer.inlineCallbacks
def inline_install(customer):
  ... same as above
 
# The new "problem" is that we have to manage all this concurrency to
# avoid causing problems to others, but this is a nice problem to have.
def twisted_developer_day(customers):
  print "Goodmorning from Twisted developer"
  work = (inline_install(customer) for customer in customers)
  #
  # We use the Cooperator mechanism to make the secretary not
  # service more than 5 customers simultaneously.
  coop = task.Cooperator()
  join = defer.DeferredList([coop.coiterate(work) for i in xrange(5)])
  #
  join.addCallback(lambda _: reactor.stop())
  print "Bye from Twisted developer!"
 
twisted_developer_day(["Customer %d" % i for i in xrange(15)])
reactor.run()
 
# We are now more lean than ever, our customers happy, our hosting
# bills ridiculously low and our performance stellar.
# ~*~ THE END ~*~

运行结果:

?
1
2
3
4
5
6
7
8
9
10
11
12
13
$ ./deferreds.py 5
------ Running example 5 ------
Goodmorning from Twisted developer
Bye from Twisted developer!
Scheduling: Installation for Customer 0
...
Callback: Finished installation for Customer 4
All done for Customer 4
Scheduling: Installation for Customer 5
...
Callback: Finished installation for Customer 14
All done for Customer 14
* Elapsed time: 9.19 seconds

从上面的输出中可以看到,程序运行时好像有5个处理消费者的槽。除非一个槽空出来,否则不会开始处理下一个消费者的请求。在本例中,处理时间都是3秒,所以看起来像是5个一批次地处理一样。最后得到的性能跟使用线程是一样的,但是这次只有一个线程,代码也更加简洁更容易写出正确的代码。

PS:deferToThread使同步函数实现非阻塞
wisted的defer.Deferred (from twisted.internet import defer)可以返回一个deferred对象.

注:deferToThread使用线程实现的,不推荐过多使用
***把同步函数变为异步(返回一个Deferred)***
twisted的deferToThread(from twisted.internet.threads import deferToThread)也返回一个deferred对象,不过回调函数在另一个线程处理,主要用于数据库/文件读取操作

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
..
 
# 代码片段
 
  def dataReceived(self, data):
    now = int(time.time())
 
    for ftype, data in self.fpcodec.feed(data):
      if ftype == 'oob':
        self.msg('OOB:', repr(data))
      elif ftype == 0x81: # 对服务器请求的心跳应答(这个是解析 防疲劳驾驶仪,发给gps上位机的,然后上位机发给服务器的)
        self.msg('FP.PONG:', repr(data))
      else:
        self.msg('TODO:', (ftype, data))
      d = deferToThread(self.redis.zadd, "beier:fpstat:fps", now, self.devid)
      d.addCallback(self._doResult, extra)

下面这儿完整的例子可以给大家参考一下

?
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
# -*- coding: utf-8 -*-
 
from twisted.internet import defer, reactor
from twisted.internet.threads import deferToThread
 
import functools
import time
 
# 耗时操作 这是一个同步阻塞函数
def mySleep(timeout):
  time.sleep(timeout)
 
  # 返回值相当于加进了callback里
  return 3
 
def say(result):
  print "耗时操作结束了, 并把它返回的结果给我了", result
 
# 用functools.partial包装一下, 传递参数进去
cb = functools.partial(mySleep, 3)
d = deferToThread(cb)
d.addCallback(say)
 
print "你还没有结束我就执行了, 哈哈"
 
reactor.run()