将子流程的stdout重定向到2个或更多子流程的stdin [英] Redirect stdout of a subprocess to stdin of 2 or more subprocesses

问题描述

基本上，我想学习如何使用一个subprocess(例如proc1)的stdout作为2个或更多其他subprocess es(例如proc2& proc3)的stdin的用法.在python中.

Basically I want to learn how to use the stdout of one subprocess (say proc1) as stdin of 2 or more other subprocesses (say proc2 & proc3) in python.

我需要zcat .gz文件，并将发送到subprocess.PIPE的输出用于cksum(unix实用程序)和行数.

Hi, I need to zcat a .gz file and use the output sent to subprocess.PIPE for both cksum (unix utility) and to line count.

我可以像这样用bash来做...

I can do it in bash like this...

[hashroot@dev_server 12]$ zcat ABC_C_TPM_26122014.data.gz | tee >(wc -l) >(cksum)| tail -2
2020090579 112180
586

我想在python中做同样的事情.

I want to do the same in python.

我一旦这样做...

>>> import subprocess
>>> import os
>>> fl123 = 'ABC_C_TPM_26122014.data.gz'
>>> pqr123 = subprocess.Popen(['zcat', fl123], stdout=subprocess.PIPE)
>>> subprocess.check_output(['cksum'], stdin=pqr123.stdout)
b'4286000649 256100 \n'

现在PIPE为空，所以如何获得行数，直到不再执行zcat.

Now the PIPE is empty so how will I get line count till I don't do zcat again.

通过在子进程中运行zcat两次，然后将第一个zcat输出重定向到wc -l并将第二个zcat输出重定向到cksum，我可以很好地做到这一点.但是zcat是基于磁盘IO的，速度很慢.所以我想避免它.

I can very well do it by running zcat twice in subprocess, and redirecting the first zcat output to wc -l and the second zcat's output to cksum. But zcat is disk IO based and is slow. So I want to avoid it.

推荐答案

在Python中实现tee命令的一种简单方法是手动写入子过程:

A simple way to implement the tee command in Python is to write to the subprocesses manually:

import gzip
from subprocess import Popen, PIPE

# zcat ABC_C_TPM_26122014.data.gz | tee >(wc -l) >(cksum)
with gzip.open("ABC_C_TPM_26122014.data.gz", "rb") as input_file:
    wc = Popen(['wc', '-l'], stdin=PIPE, bufsize=1, close_fds=True)
    cksum = Popen(['cksum'], stdin=PIPE, bufsize=1, close_fds=True)

    line_count = 0
    for line_count, line in enumerate(input_file, start=1):
        wc.stdin.write(line)
        cksum.stdin.write(line)
    wc.stdin.close()
    cksum.stdin.close()
wc.wait()
cksum.wait()
print("Line count in the parent: %d" % line_count)

如果输入中的行可能很大，那么您可以分块读取输入:chunk = input_file.read(chunk_size)而不是逐行(b'\n').

If the lines in the input can be large then you could read the input in chunks: chunk = input_file.read(chunk_size) instead of line by line (b'\n').

这篇关于将子流程的stdout重定向到2个或更多子流程的stdin的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！