Executing Shell Commands with Python
Repetitive tasks are ripe for automation. It is common for developers and system administrators to automate routine tasks like health checks and file backups with shell scripts. However, as those tasks become more complex, shell scripts may become harder to maintain.
Fortunately, we can use Python instead of shell scripts for automation. Python provides methods to run shell commands, giving us the same functionality of those shells scripts, with additional support from Python’s ecosystem. Learning how to run shell commands in Python opens the door for us to automate computer tasks in a structured and scalable way.
In this article, we will look at the various ways to execute shell commands in Python, and the ideal situation to use each method.
Using os.system to Run a Command
Python allows us to immediately execute a shell command that’s stored in a string using the os.system() function. Let’s start by creating a new Python file called echo_adelle.py and enter the following:
import os os.system("echo Hello from the other side!") The first thing we do in our Python file is import the os module, which contains the system function that can execute shell commands. The next line does exactly that, runs the echo command in our shell through Python.
In your Terminal, run this file with using the following command, and you should see the corresponding output:
$ python3 echo_adelle.py Hello from the other side! As the echo commands prints to our stdout , os.system() also displays the output on our stdout stream. While not visible in the console, the os.system() command returns the exit code of the shell command. An exit code of 0 means it ran without any problems and any other number means an error. This way, you ca control the flow of your commands, check the results and either retry or return an error if any call fails:
# cd_return_codes.py import os home_dir = os.system("cd ~") print("`cd ~` ran with exit code %d" % home_dir) unknown_dir = os.system("cd doesnotexist") print("`cd doesnotexis` ran with exit code %d" % unknown_dir) In this script, we create two variables that store the result of executing commands that change the directory to the home folder, and to a folder that does not exist. Running this file, we will see:
$ python3 cd_return_codes.py `cd ~` ran with exit code 0 sh: line 0: cd: doesnotexist: No such file or directory `cd doesnotexist` ran with exit code 256 The first command, which changes the directory to the home directory, executes successfully. Therefore, os.system() returns its exit code, zero, which is stored in home_dir . On the other hand, unknown_dir stores the exit code of the failed bash command to change the directory to a folder that does not exist.
The os.system() function executes a command, prints any output of the command to the console, and returns the exit code of the command. If we would like more fine grained control of a shell command’s input and output in Python, we should use the subprocess module.
Running a Command with subprocess
The subprocess module is Python’s recommended way to executing shell commands. It gives us the flexibility to suppress the output of shell commands or chain inputs and outputs of various commands together, while still providing a similar experience to os.system() for basic use cases.
In a new file called list_subprocess.py , let’s run a command with the subprocess module:
import subprocess list_files = subprocess.run(["ls", "-l"]) print("The exit code was: %d" % list_files.returncode) In the first line, we import the subprocess module, which is part of the Python standard library. We then use the subprocess.run() function to execute the command. Like os.system() , the subprocess.run() command returns the exit code of what was executed.
Unlike os.system() , note how subprocess.run() requires a list of strings as input instead of a single string. The first item of the list is the name of the command. The remaining items of the list are the flags and the arguments of the command.
Note: As a rule of thumb, you need to separate the arguments based on space, for example ls -alh would be [«ls», «-alh»] , while ls -a -l -h , would be [«ls», «-a», -«l», «-h»] . As another example, echo hello world would be [«echo», «hello», «world»] , whereas echo «hello world» or echo hello\ world would be [«echo», «hello world»] .
Run this file and your console’s output would be similar to:
$ python3 list_subprocess.py total 80 [email protected] 1 stackabuse staff 216 Dec 6 10:29 cd_return_codes.py [email protected] 1 stackabuse staff 56 Dec 6 10:11 echo_adelle.py [email protected] 1 stackabuse staff 116 Dec 6 11:20 list_subprocess.py The exit code was: 0 Now let’s try to use one of the more advanced features of subprocess.run() , namely ignore output to stdout . In the same list_subprocess.py file, change:
list_files = subprocess.run(["ls", "-l"]) list_files = subprocess.run(["ls", "-l"], stdout=subprocess.DEVNULL) The standard output of the command now pipes to the special /dev/null device, which means the output would not appear on our consoles. Execute the file in your shell to see the following output:
$ python3 list_subprocess.py The exit code was: 0 What if we wanted to provide input to a command? The subprocess.run() facilitates this by its input argument:
Free eBook: Git Essentials
Check out our hands-on, practical guide to learning Git, with best-practices, industry-accepted standards, and included cheat sheet. Stop Googling Git commands and actually learn it!
import subprocess useless_cat_call = subprocess.run(["cat"], stdout=subprocess.PIPE, text=True, input="Hello from the other side") print(useless_cat_call.stdout) # Hello from the other side We use subprocess.run() with quite a few commands, let’s go through them:
- stdout=subprocess.PIPE tells Python to redirect the output of the command to an object so it can be manually read later
- text=True returns stdout and stderr as strings. The default return type is bytes.
- input=»Hello from the other side» tells Python to add the string as input to the cat command.
Running this produces the following output:
Hello from the other side We can also raise an Exception without manually checking the return value. In a new file, false_subprocess.py , add the code below:
import subprocess failed_command = subprocess.run(["false"], check=True) print("The exit code was: %d" % failed_command.returncode) In your Terminal, run this file. You will see the following error:
$ python3 false_subprocess.py Traceback (most recent call last): File "false_subprocess.py", line 4, in failed_command = subprocess.run(["false"], check=True) File "/usr/local/python/3.7.5/Frameworks/Python.framework/Versions/3.7/lib/python3.7/subprocess.py", line 512, in run output=stdout, stderr=stderr) subprocess.CalledProcessError: Command '['false']' returned non-zero exit status 1. By using check=True , we tell Python to raise any exceptions if an error is encountered. Since we did encounter an error, the print statement on the final line was not executed.
The subprocess.run() function gives us immense flexibility that os.system() doesn’t when executing shell commands. This function is a simplified abstraction of the subprocess.Popen class, which provides additional functionality we can explore.
Running a Command with Popen
The subprocess.Popen class exposes more options to the developer when interacting with the shell. However, we need to be more explicit about retrieving results and errors.
By default, subprocess.Popen does not stop processing of a Python program if its command has not finished executing. In a new file called list_popen.py , type the following:
import subprocess list_dir = subprocess.Popen(["ls", "-l"]) list_dir.wait() This code is equivalent to that of list_subprocess.py . It runs a command using subprocess.Popen , and waits for it to complete before executing the rest of the Python script.
Let’s say we do not want to wait for our shell command to complete execution so the program can work on other things. How would it know when the shell command has finished execution?
The poll() method returns the exit code if a command has finished running, or None if it’s still executing. For example, if we wanted to check if list_dir was complete instead of wait for it, we would have the following line of code:
To manage input and output with subprocess.Popen , we need to use the communicate() method.
In a new file called cat_popen.py , add the following code snippet:
import subprocess useless_cat_call = subprocess.Popen(["cat"], stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True) output, errors = useless_cat_call.communicate(input="Hello from the other side!") useless_cat_call.wait() print(output) print(errors) The communicate() method takes an input argument that’s used to pass input to the shell command. The communicate method also returns both the stdout and stderr when they are set.
Having seen the core ideas behind subprocess.Popen , we have now covered three ways to run shell commands in Python. Let’s re-examine their characteristics so we’ll know which method is best suited for a project’s requirements.
Best Approach to Run Shell Commands with Python?
If you need to run one or a few simple commands and do not mind if their output goes to the console, you can use the os.system() command. If you want to manage the input and output of a shell command, use subprocess.run() . If you want to run a command and continue doing other work while it’s being executed, use subprocess.Popen .
Here is a table with some usability differences that you can also use to inform your decision:
| os.system | subprocess.run | subprocess.Popen | |
|---|---|---|---|
| Requires parsed arguments | no | yes | yes |
| Waits for the command | yes | yes | no |
| Communicates with stdin and stdout | no | yes | yes |
| Returns | return value | object | object |
Conclusion
Python allows you to execute shell commands, which you can use to start other programs or better manage shell scripts that you use for automation. Depending on our use case, we can use os.system() , subprocess.run() or subprocess.Popen to run bash commands.
Using these techniques, what external task would you run via Python?