CODEWITHSUNDEEP
pythontmux
jumpingjack
jumpingjack

Reputation: 3

How can I use python to control multiple concurrant command line programs?

I'd like to use a python program to send different videos to different devices.

My plan is to use ffmpeg to control the video and the destination (I can do this for one destination using os.system) but I'm not sure how to write concurrent ffmpeg commands so that 6 videos are playing at the same time on different devices.

Initially I thought I could use tmux but I can't find a solution for how to control/access different tmux windows within my python program. Am I missing something obvious?

Upvotes: 0

Views: 207

Answers (1)

Eelco van Vliet
Eelco van Vliet

Reputation: 1238

You can use the python subprocess module for that: https://docs.python.org/3/library/subprocess.html

As a simple example, I will run a Linux utility 'sleep' which does nothing else than waiting for a given amount of seconds. I will do this in parallel to see that we can really do this with subprocess in parallel. As a first setup, I do:

import os
import subprocess
import time

n_jobs = 5
sleeping_time_in_sec = 10
shell_command_with_arguments = ["sleep", f"{sleeping_time_in_sec}s"]

I'm importing 'subprocess' to launch jobs in parallel. The 'time'module I'm going to use to measure the total running time. The 'os' module I'm going to use to block the script until all jobs are done. The shell command looks like 'sleep 10s', but for subprocess you put the script and all the arguments in a list.

Submit the jobs like this:

time_start = time.time()
jobs = list()
for counter in range(n_jobs):
    process = subprocess.Popen(shell_command_with_arguments, shell=False)
    jobs.append(process)

Now all the 5 jobs each taking 10 sec have been submitted. In order to make your script wait until the last process is done, you can add:

print(f"Waiting for all {n_jobs} processes to finish...")
for ip, process in enumerate(jobs):
    try:
        os.waitpid(process.pid, 0)
    except ChildProcessError:
        print(f"No more: {ip} : {process.pid}")
    else:
        print(f"DONE: {ip} : {process.pid}")

Finally, I report the total running time:

time_end = time.time()
delta_time = time_end - time_start
print(f"Run {n_jobs} jobs taking {sleeping_time_in_sec} s each in {delta_time} s")

The output of the script looks like:

Waiting for all 5 processes to finish...
DONE: 0 : 88988
DONE: 1 : 88989
DONE: 2 : 88990
DONE: 3 : 88991
DONE: 4 : 88992
Finished running 5 jobs each taking 10 s in 10.022365093231201 s

As you can see, if you would run the sleep command 5 times in serie, it would take 50 s. But since the script took a little more then 10 s you can see the the jobs indeed were running in parallel.

Upvotes: 1

Related Questions