Python：并行执行 cat 子进程

Question

我正在远程服务器上运行多个cat | zgrep 命令并分别收集它们的输出以进行进一步处理：

class MainProcessor(mp.Process):
    def __init__(self, peaks_array):
        super(MainProcessor, self).__init__()
        self.peaks_array = peaks_array

    def run(self):
        for peak_arr in self.peaks_array:
            peak_processor = PeakProcessor(peak_arr)
            peak_processor.start()

class PeakProcessor(mp.Process):
    def __init__(self, peak_arr):
        super(PeakProcessor, self).__init__()
        self.peak_arr = peak_arr

    def run(self):
        command = 'ssh remote_host cat files_to_process | zgrep --mmap "regex" '
        log_lines = (subprocess.check_output(command, shell=True)).split('\n')
        process_data(log_lines)

然而，这会导致 subprocess('ssh ... cat ...') 命令的顺序执行。第二个高峰等待第一个完成，依此类推。

如何修改此代码以使子进程调用并行运行，同时仍能单独收集每个子进程的输出？

Answer 1

您不需要multiprocessing 或threading 来并行运行子进程。例如：

#!/usr/bin/env python
from subprocess import Popen

# run commands in parallel
processes = [Popen("echo {i:d}; sleep 2; echo {i:d}".format(i=i), shell=True)
             for i in range(5)]
# collect statuses
exitcodes = [p.wait() for p in processes]

它同时运行 5 个 shell 命令。注意：这里既不使用线程也不使用multiprocessing 模块。在 shell 命令中添加与符号 & 是没有意义的：Popen 不会等待命令完成。您需要显式调用.wait()。

很方便，但不需要使用线程来收集子进程的输出：

#!/usr/bin/env python
from multiprocessing.dummy import Pool # thread pool
from subprocess import Popen, PIPE, STDOUT

# run commands in parallel
processes = [Popen("echo {i:d}; sleep 2; echo {i:d}".format(i=i), shell=True,
                   stdin=PIPE, stdout=PIPE, stderr=STDOUT, close_fds=True)
             for i in range(5)]

# collect output in parallel
def get_lines(process):
    return process.communicate()[0].splitlines()

outputs = Pool(len(processes)).map(get_lines, processes)

相关：Python threading multiple bash subprocesses?.

以下代码示例在同一线程中同时从多个子进程获取输出：

#!/usr/bin/env python3
import asyncio
import sys
from asyncio.subprocess import PIPE, STDOUT

@asyncio.coroutine
def get_lines(shell_command):
    p = yield from asyncio.create_subprocess_shell(shell_command,
            stdin=PIPE, stdout=PIPE, stderr=STDOUT)
    return (yield from p.communicate())[0].splitlines()

if sys.platform.startswith('win'):
    loop = asyncio.ProactorEventLoop() # for subprocess' pipes on Windows
    asyncio.set_event_loop(loop)
else:
    loop = asyncio.get_event_loop()

# get commands output in parallel
coros = [get_lines('"{e}" -c "print({i:d}); import time; time.sleep({i:d})"'
                    .format(i=i, e=sys.executable)) for i in range(5)]
print(loop.run_until_complete(asyncio.gather(*coros)))
loop.close()

Answer 2

另一种方法（而不是其他将 shell 进程置于后台的建议）是使用multithreading.

您拥有的 run 方法会执行以下操作：

thread.start_new_thread ( myFuncThatDoesZGrep)

要收集结果，您可以执行以下操作：

class MyThread(threading.Thread):
   def run(self):
       self.finished = False
       # Your code to run the command here.
       blahBlah()
       # When finished....
       self.finished = True
       self.results = []

按照上面关于多线程的链接中的说明运行线程。当您的线程对象具有 myThread.finished == True 时，您可以通过 myThread.results 收集结果。