On Windows, I am terminating Popen subprocess if it runs too long.

Is there any way to collect output and error stream before kill is invoked?

I create a sample child process which prints number 0 to 5 and takes 5 second.

child.py

import time

i = 1
while True:
    print(i)
    if i == 5:
        break
    i += 1
    time.sleep(1)

Then I create a parent which terminates child after 2 seconds.

parent.py

import subprocess
import time

p = subprocess.Popen(
    ["C:\..\python.exe", "C:\child.py"],
    stdout=subprocess.PIPE,
    stderr=subprocess.PIPE,
)
time.sleep(2)
p.kill()
output, error = p.communicate()
print(output)

My expectation would be that p.communicate() returns what child has managed to write into output within 2 seconds, i.e. something like b'0rn1rn. However, the output is empty.

When I reverse the order of methods, then parent waits for child to finish which is not desired.

time.sleep(2)
output, error = p.communicate()  # takes 3 seconds
print(output)  # returns b'0rn1rn2rn3rn4rn'
p.kill()