如何读取子进程的stdout的第一个字节,然后在Python中丢弃其余的？

Question

我想读一个子进程'stdout的第一个字节,知道它已经开始运行了.之后,我想丢弃所有进一步的输出,这样我就不用担心缓冲区了.

做这个的最好方式是什么？

澄清:我希望子进程继续与我的程序一起运行,我不想等待它终止或类似的东西.理想情况下,有一些简单的方法可以做到这一点,而不是诉诸threading,fork或multiprocessing.

如果我忽略输出流,或者.close()它,如果发送的数据多于它可以容纳在其缓冲区中的数据,则会导致错误.

Answer 1

如果你正在使用Python 3.3+,您可以使用DEVNULL特殊的价值stdout和stderr丢弃子输出.

from subprocess import Popen, DEVNULL

process = Popen(["mycmd", "myarg"], stdout=DEVNULL, stderr=DEVNULL)

或者,如果您使用的是Python 2.4+,则可以使用以下方法模拟:

import os
from subprocess import Popen

DEVNULL = open(os.devnull, 'wb')
process = Popen(["mycmd", "myarg"], stdout=DEVNULL, stderr=DEVNULL)

但是,这并没有让您有机会读取stdout的第一个字节.

Answer 2

这似乎可行，但感觉不惯用。

#!/usr/bin/env python3.1
import threading
import subprocess

def discard_stream_while_running(stream, process):
    while process.poll() is None:
        stream.read(1024)

def discard_subprocess_pipes(process, out=True, err=True, in_=True):
    if out and process.stdout is not None and not process.stdout.closed:
        t = threading.Thread(target=discard_stream_while_running, args=(process.stdout, process))
        t.start()

    if err and process.stderr is not None and not process.stderr.closed:
        u = threading.Thread(target=discard_stream_while_running, args=(process.stderr, process))
        u.start()

    if in_ and process.stdin is not None and not process.stdin.closed:
        process.stdin.close()

示例/测试用法

if __name__ == "__main__":
    import tempfile
    import textwrap
    import time

    with tempfile.NamedTemporaryFile("w+t", prefix="example-", suffix=".py") as f:
        f.write(textwrap.dedent("""
            import sys
            import time

            sys.stderr.write("{} byte(s) read through stdin.\\n"
                             .format(len(sys.stdin.read())))

            # Push a couple of MB/s to stdout, messages to stderr.
            while True:
                sys.stdout.write("Hello Parent\\n" * 1000000)
                sys.stderr.write("Subprocess Writing Data\\n")
                time.sleep(0.5)
        """))
        f.flush()

        p = subprocess.Popen(["python3.1", f.name],
                             stdout=subprocess.PIPE,
                             stdin=subprocess.PIPE)

        p.stdin.write("Hello Child\n".encode())

        discard_subprocess_pipes(p) # <-- Here

        for s in range(16, 0, -1):
            print("Main Process Running For", s, "More Seconds")
            time.sleep(1)