If, hypothetically speaking, I wanted to execute a bash script on my server through the UI of Backdrop CMS (implemented with a custom module). 

  • Can I do this securely?
  • Can anyone point me at any existing sample code?
  • How would I do this?

Comments

> My first reaction when I read this was "You shouldn't" 😅...

* this was my first reaction as well; in many years of BD and Drupal dev I've never had occasion for this; so i'm curios as to what your use case is.

 

  • > Can I do this securely?
    • depends on a lot
      • what is the bash script doing? is just `ls` then prolly fine
      • who can access it; admin only; everyone?
  • > Can anyone point me at any existing sample code?
    • @bwpanda has shown how `drush` is doing it; should work in backdrop too, but it should be pointed out a `drush` user is a cli user and already has access to cli and bash scripting ... that is a big difference
  • > How would I do this?
    • the how is in the `drush` example.

Might be better to tell us about your use case and find an alternate method to what you are trying to accomplish.

Ignoring "If you should..."

A Backdrop module is just PHP, so, first result from Google:

https://stackoverflow.com/questions/11052162/run-bash-command-from-php

# # #

But I agree with everyone else, however that gets called needs to be completely locked down.  My off the cuff suggestion is to not have this in a module, but have it embedded into a page and use the normal roles/permissions to lock the page down.  Basically KISS.

Best, Michael

I did some experimenting and was able to run a bash script located within the Drupal directory using the php function: https://www.php.net/manual/en/function.exec.php

However, it's not clear to me that this is a good option and I understand the concerns. 

We are currently experimenting with another option. We have set up a script that runs on periodic basis. Each time the script runs, it checks the database for a new request. If there is a new request for a specific action, it runs another script on the server to fulfill that request. 

In this case, the script is only looking for specific values and if that specific value is present, it takes the action requested. 

This seems like it's a lot safer. Any thoughts?

Happy to discuss the specifics with anyone in Zulip. :-)