如何从“子进程”流式输出（一次一行）？

Question

I'm really struggling to successfully stream output from a bash command in my Python script using the subprocess package. The command is just a simple AWS CLI command to upload an object to S3:

aws s3 cp --profile MY-PROFILE UPLOADED_FILE s3://BUCKET/PREFIX/

When I run this in bash , it shows the progress of the upload. With subprocess , I can only get it working in the following ways:

• Final output only at the end.

   upload: ./OBJECT to s3://BUCKET/PREFIX/OBJECT

• Line by line output

  Completed 512.0 KiB/12.2 MiB (697.8 KiB/s) with 1 file(s) remaining Completed 768.0 KiB/12.2 MiB (980.0 KiB/s) with 1 file(s) remaining Completed 1.0 MiB/12.2 MiB (1.2 MiB/s) with 1 file(s) remaining Completed 1.2 MiB/12.2 MiB (1.5 MiB/s) with 1 file(s) remaining Completed 1.5 MiB/12.2 MiB (1.6 MiB/s) with 1 file(s) remaining Completed 1.8 MiB/12.2 MiB (1.4 MiB/s) with 1 file(s) remaining

• Similar thing without line separation:

   b'Completed 256.0 KiB/12.2 MiB (297.5 KiB/s) with 1 file(s) remaining\\rCompleted 512.0 KiB/12.2 MiB (506.7 KiB/s) with 1 file(s) remaining\\rCompleted 768.0 KiB/12.2 MiB (544.4 KiB/s) with 1 file(s) remaining\\rCompleted 1.0 MiB/12.2 MiB (672.0 KiB/s) with 1 file(s) remaining \\rCompleted 1.2 MiB/12.2 MiB (707.0 KiB/s) with 1 file(s) remaining \\rCompleted 1.5 MiB/12.2 MiB (780.3 KiB/s) with 1 file(s) remaining \\rCompleted 1.8 MiB/12.2 MiB (849.1 KiB/s) with 1 file(s) remaining \\rCompleted 2.0 MiB/12.2 MiB (865.6 KiB/s) with 1 file(s) remaining \\rCompleted 2.2 MiB/12.2 MiB (971.0 KiB/s) with 1 file(s) remaining \\rCompleted 2.5 MiB/12.2 MiB (992.4 KiB/s) with 1 file(s) remaining \\rCompleted 2.8 MiB/12.2 MiB (1.0 MiB/s) with 1 file(s) remaining

What I need is to just use one line and show the live status, just as it would do in bash.

Here are a number of different functions I've tried, all with no luck:

from subprocess import Popen, PIPE
import subprocess
import os
import shlex
from time import sleep


def stream_process(process):
    go = process.poll() is None
    for line in process.stdout:
        print(line)
    return go

process = subprocess.Popen(cmd, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
while stream_process(process):
    sleep(0.1)


def run(command):
  with subprocess.Popen(command, text=True, shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT) as process:
    for line in process.stdout:
      print(line)

run(command)

process = Popen(command, text=True, shell=True, stdout=PIPE, bufsize=1, universal_newlines=True)
while True:
    output = process.stdout.readline()
    if process.poll() is not None and output == b'':
        break
    if output:
        print (output.strip())
retval = process.poll()


with Popen(command, text=True, shell=True, stdout=PIPE, bufsize=1, universal_newlines=True) as p:

    while True: 
        line = p.stdout.readline()
        print(line.strip())


    for line in p.stdout:
        print(line.strip(), "\n", flush=True, end='')
        # print(line.strip())


print(result.stdout.readlines())

print("args", result.args)
print("returncode", result.returncode)
print("stdout", result.stdout)
print("stderr", result.stderr)
print("check_returncode()", result.check_returncode())

def run_command(cmd):
    p = Popen(shlex.split(cmd), bufsize=1, universal_newlines=True)
    return p.poll()

run_command(command)
# invoke process
process = subprocess.Popen(shlex.split(command),shell=False,stdout=subprocess.PIPE)

# Poll process.stdout to show stdout live
while True:
  output = process.stdout.readline()
  if process.poll() is not None:
    break
  if output:
    print(output.strip())
rc = process.poll()


print(result.stdout.readlines())
while True:
    line = result.stdout.readline()
    if not line:
        break
    print(line, flush=True)

FAQ

Why don't you use boto3 instead of subprocess + bash ?

I'm developing a program for non-Python devs, so I'd like to limit dependencies and only use standard packages as much as possible.

Answer 1

You simply should to:

import subprocess

p = subprocess.Popen(("ls"), stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
for line in p.stdout:
    print(line.decode("utf-8").rstrip())
p.wait()
status = p.poll()
print("process terminate with code: %s" % status)

Result in:

my-file.txt
...
process terminate with code: 0

When the subprocess exit, stdout is closed, which break the loop.

subprocess stdout is a bytes stream. If you need str objects, you should decode it: line.decode("utf-8") .

But I think this is not the problem. Some command have different output in case of stdout is a terminal or a pipe. Like with wget . I you want a progress bar, you have to parse output and implement your own progress bar.

Answer 2

for line in process.stdout and process.stdout.readline() are not fit for the task because they don't return before a whole line has ended, ie a \\n is received, while a progress part only ends with \\r . A way to read output independently of any line termination is to use os.read() :

from os         import read

process = Popen(shlex.split(cmd), stdout=PIPE)
while buffer := read(process.stdout.fileno(), 4096):
    print(buffer.decode(), end='')

As xrisk noted, using text=True would allow lines ending with \\r to be read with readline() , but that's of no use here because therewith all line endings in the output will be converted to '\\n' .

But there's indeed a way to use a text stream without loosing the original line endings by means of io.TextIOWrapper - therewith we can do this:

import io

process = Popen(shlex.split(cmd), stdout=PIPE)
for line in io.TextIOWrapper(process.stdout, newline=""): print(line, end="")