First, the python subprocess module is the correct answer.
As an subprocess example:
import subprocess
x = subprocess.check_output(["echo","one","two","three"])
Where x will be the output (python3 bytes class: x.decode('utf-8')
for string)
Note that this will NOT duplicate stderr. If you need stderr as well, you can do something like:
x = subprocess.check_output(["bash","-c", 'echo foo; echo bar >&2'],stderr=subprocess.STDOUT)
Of course, there are many other ways of capturing stderr, including to a different output variable.
Using direct control
However, if you are doing something tricky and need to have direct control, examine the code below:
import os
rside, wside = os.pipe()
if not os.fork():
# Child
os.close(rside)
# Make stdout go to parent
os.dup2(wside, 1)
# Make stderr go to parent
os.dup2(wside, 2)
# Optionally make stdin come from nowhere
devnull = os.open("/dev/null", os.O_RDONLY)
os.dup2(devnull, 0)
# Execute the desired program
os.execve("/bin/bash",["/bin/bash","-c","echo stdout; echo stderr >&2"],os.environ)
print("Failed to exec program!")
sys.exit(1)
# Parent
os.close(wside)
pyrside = os.fdopen(rside)
for line in pyrside:
print("Child (stdout or stderr) said: <%s>"%line)
# Prevent zombies! Reap the child after exit
pid, status = os.waitpid(-1, 0)
print("Child exited: pid %d returned %d"%(pid,status))
Note: @Beginner's answer is flawed in a few ways: os._exit(0) was included which immediately causes the child to exit, rendering everything else pointless. No os.execve() rendering the primary goal of the question pointless. No way to access the child's stdout/stderr as another question goal.