停止在Python中读取进程输出而不挂起?

时间:2010-12-11 17:27:08

标签: python subprocess multiprocessor hung

我的Linux程序几乎就像这样:

import os
import time

process = os.popen("top").readlines()

time.sleep(1)

os.popen("killall top")

print process

程序挂起在这一行:

process = os.popen("top").readlines()

这种情况发生在保持更新输出的工具中,如“Top”

我最好的考验:

import os
import time
import subprocess

process = subprocess.Popen('top')

time.sleep(2)

os.popen("killall top")

print process

它比第一个(它被砍掉)效果更好,但它返回:

<subprocess.Popen object at 0x97a50cc>

第二次审判:

import os
import time
import subprocess

process = subprocess.Popen('top').readlines()

time.sleep(2)

os.popen("killall top")

print process

与第一个相同。它由于“readlines()”而被绞死

它的回归应该是这样的:

top - 05:31:15 up 12:12,  5 users,  load average: 0.25, 0.14, 0.11
Tasks: 174 total,   2 running, 172 sleeping,   0 stopped,   0 zombie
Cpu(s):  9.3%us,  3.8%sy,  0.1%ni, 85.9%id,  0.9%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   1992828k total,  1849456k used,   143372k free,   233048k buffers
Swap:  4602876k total,        0k used,  4602876k free,  1122780k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND            
31735 Barakat   20   0  246m  52m  20m S 19.4  2.7  13:54.91 totem              
 1907 root      20   0 91264  45m  15m S  1.9  2.3  38:54.14 Xorg               
 2138 Barakat   20   0 17356 5368 4284 S  1.9  0.3   3:00.15 at-spi-registry    
 2164 Barakat    9 -11  164m 7372 6252 S  1.9  0.4   2:54.58 pulseaudio         
 2394 Barakat   20   0 27212 9792 8256 S  1.9  0.5   6:01.48 multiload-apple    
 6498 Barakat   20   0 56364  30m  18m S  1.9  1.6   0:03.38 pyshell            
    1 root      20   0  2880 1416 1208 S  0.0  0.1   0:02.02 init               
    2 root      20   0     0    0    0 S  0.0  0.0   0:00.02 kthreadd           
    3 root      RT   0     0    0    0 S  0.0  0.0   0:00.12 migration/0        
    4 root      20   0     0    0    0 S  0.0  0.0   0:02.07 ksoftirqd/0        
    5 root      RT   0     0    0    0 S  0.0  0.0   0:00.00 watchdog/0         
    9 root      20   0     0    0    0 S  0.0  0.0   0:01.43 events/0           
   11 root      20   0     0    0    0 S  0.0  0.0   0:00.00 cpuset             
   12 root      20   0     0    0    0 S  0.0  0.0   0:00.02 khelper            
   13 root      20   0     0    0    0 S  0.0  0.0   0:00.00 netns              
   14 root      20   0     0    0    0 S  0.0  0.0   0:00.00 async/mgr          
   15 root      20   0     0    0    0 S  0.0  0.0   0:00.00 pm

并保存在变量“process”中。任何我想到的人,我现在真的被困住了吗?

5 个答案:

答案 0 :(得分:25)

#!/usr/bin/env python
"""Start process; wait 2 seconds; kill the process; print all process output."""
import subprocess
import tempfile
import time

def main():
    # open temporary file (it automatically deleted when it is closed)
    #  `Popen` requires `f.fileno()` so `SpooledTemporaryFile` adds nothing here
    f = tempfile.TemporaryFile() 

    # start process, redirect stdout
    p = subprocess.Popen(["top"], stdout=f)

    # wait 2 seconds
    time.sleep(2)

    # kill process
    #NOTE: if it doesn't kill the process then `p.wait()` blocks forever
    p.terminate() 
    p.wait() # wait for the process to terminate otherwise the output is garbled

    # print saved output
    f.seek(0) # rewind to the beginning of the file
    print f.read(), 
    f.close()

if __name__=="__main__":
    main()

仅打印输出部分的尾部解决方案

您可以在另一个线程中读取进程输出并将所需的最后一行数保存在队列中:

import collections
import subprocess
import time
import threading

def read_output(process, append):
    for line in iter(process.stdout.readline, ""):
        append(line)

def main():
    # start process, redirect stdout
    process = subprocess.Popen(["top"], stdout=subprocess.PIPE, close_fds=True)
    try:
        # save last `number_of_lines` lines of the process output
        number_of_lines = 200
        q = collections.deque(maxlen=number_of_lines) # atomic .append()
        t = threading.Thread(target=read_output, args=(process, q.append))
        t.daemon = True
        t.start()

        #
        time.sleep(2)
    finally:
        process.terminate() #NOTE: it doesn't ensure the process termination

    # print saved lines
    print ''.join(q)

if __name__=="__main__":
    main()

此变体要求q.append()为原子操作。否则输出可能已损坏。

signal.alarm()解决方案

您可以使用signal.alarm()在指定的超时后调用process.terminate(),而不是在另一个线程中读取。虽然它可能与subprocess模块没有很好地交互。基于@Alex Martelli's answer

import collections
import signal
import subprocess

class Alarm(Exception):
    pass

def alarm_handler(signum, frame):
    raise Alarm

def main():
    # start process, redirect stdout
    process = subprocess.Popen(["top"], stdout=subprocess.PIPE, close_fds=True)

    # set signal handler
    signal.signal(signal.SIGALRM, alarm_handler)
    signal.alarm(2) # produce SIGALRM in 2 seconds

    try:
        # save last `number_of_lines` lines of the process output
        number_of_lines = 200
        q = collections.deque(maxlen=number_of_lines)
        for line in iter(process.stdout.readline, ""):
            q.append(line)
        signal.alarm(0) # cancel alarm
    except Alarm:
        process.terminate()
    finally:
        # print saved lines
        print ''.join(q)

if __name__=="__main__":
    main()

此方法仅适用于* nix系统。如果process.stdout.readline()没有返回,它可能会阻止。

threading.Timer解决方案

import collections
import subprocess
import threading

def main():
    # start process, redirect stdout
    process = subprocess.Popen(["top"], stdout=subprocess.PIPE, close_fds=True)

    # terminate process in timeout seconds
    timeout = 2 # seconds
    timer = threading.Timer(timeout, process.terminate)
    timer.start()

    # save last `number_of_lines` lines of the process output
    number_of_lines = 200
    q = collections.deque(process.stdout, maxlen=number_of_lines)
    timer.cancel()

    # print saved lines
    print ''.join(q),

if __name__=="__main__":
    main()

此方法也适用于Windows。在这里,我使用process.stdout作为可迭代的;它可能会引入额外的输出缓冲,如果不需要,可以切换到iter(process.stdout.readline, "")方法。如果进程没有在process.terminate()上终止,则脚本会挂起。

没有线程,没有信号解决方案

import collections
import subprocess
import sys
import time

def main():
    args = sys.argv[1:]
    if not args:
        args = ['top']

    # start process, redirect stdout
    process = subprocess.Popen(args, stdout=subprocess.PIPE, close_fds=True)

    # save last `number_of_lines` lines of the process output
    number_of_lines = 200
    q = collections.deque(maxlen=number_of_lines)

    timeout = 2 # seconds
    now = start = time.time()    
    while (now - start) < timeout:
        line = process.stdout.readline()
        if not line:
            break
        q.append(line)
        now = time.time()
    else: # on timeout
        process.terminate()

    # print saved lines
    print ''.join(q),

if __name__=="__main__":
    main()

此变体既不使用线程也不使用信号,但会在终端中产生乱码输出。如果process.stdout.readline()阻止,它将阻止。

答案 1 :(得分:3)

我建议不要使用“顶部”,而是建议使用“ps”来提供相同的信息,但只能使用一次而不是每秒一次。

你还需要使用ps的一些标志,我倾向于使用“ps aux”

答案 2 :(得分:0)

我会做的,而不是这种方法,是检查您尝试从中获取信息的程序,并确定该信息的最终来源。它可以是API调用或设备节点。然后,编写一些从同一个源获取它的python。这消除了“刮”“熟”数据的问题和开销。

答案 3 :(得分:0)

(J.F. Sebastian你的代码很棒,我觉得它比我的解决方案更好=))

我用另一种方式解决了它。

不是直接在终端上输出,而是将其变成文件“tmp_file”:

top >> tmp_file

然后我使用工具“cut”使其输出“是最高输出”作为过程的值

cat tmp_file

它完成了我想要它做的事。这是最终的代码:

import os
import subprocess
import time

subprocess.Popen("top >> tmp_file",shell = True)

time.sleep(1)

os.popen("killall top")

process = os.popen("cat tmp_file").read()

os.popen("rm tmp_file")

print process

# Thing better than nothing =)

非常感谢你们的帮助

答案 4 :(得分:0)

事实上,如果你填写输出缓冲区,你会得到一些答案。因此,一种解决方案是使用大垃圾输出填充缓冲区(〜6000字符,bufsize = 1)。

让我们说,你有一个在sys.stdout上写的python脚本而不是top:

GARBAGE='.\n'
sys.stdout.write(valuable_output)
sys.stdout.write(GARBAGE*3000)

在启动器端,而不是简单的process.readline():

GARBAGE='.\n'
line=process.readline()
while line==GARBAGE:
   line=process.readline()

确定它有点脏,因为2000依赖于子进程实现,但它工作正常并且非常简单。设置除了bufsize = 1以外的任何东西都会让事情变得更糟。