Correctly reading stdout/stderr from subprocess

[Christoph Haas]

On Wed, Jun 14, 2006 at 03:56:16PM +0200, Maric Michaud wrote:
> I did it just to validate my point and because i don't use threads very often 
> in python, some exercises can't hurt :)
 
Since you are familiar with threads I believe that my excercises are a
tad bit more low-level compared to yours. :)

> def run(command):
> 
>    import subprocess
> 
>    run = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, 
> stderr=subprocess.PIPE)
> 
>    # Wait for the process to return
>    import thread, threading
>    out, err = [], []
>    out_ended, err_ended = threading.Event(), threading.Event()
> 
>    def getOutput(output, lines, ended_event) :
>        for i in output.readlines() : lines.append(i.rstrip('\n'))
>        ended_event.set()
> 
>    out_thread = thread.start_new_thread(getOutput, (run.stdout, out, 
> out_ended))
>    err_thread = thread.start_new_thread(getOutput, (run.stderr, err, 
> err_ended))
> 
>    out_ended.wait()
>    err_ended.wait()
> 
>    returncode = run.wait()
> 
>    return returncode, out, err

Quite interesting. I dived into /usr/lib/python2.4/subprocess.py and
found out that the "communicate()" method actually works similarly. I
might even prefer your version because I get the returncode from
run.wait() what "communicate()" does not.

Great work. Thanks for your time.

Kindly
 Christoph