Quick Links

Tomer Filiba Tomer Filiba

Plumbum: Shell Combinators and More

Ever wished the compactness of shell scripts be put into a real programming language? Say hello to Plumbum Shell Combinators. Plumbum (Latin for lead, which was used to create pipes back in the day) is a small yet feature-rich library for shell script-like programs in Python. The motto of the library is “Never write shell scripts again”, and thus it attempts to mimic the shell syntax (shell combinators) where it makes sense, while keeping it all Pythonic and cross-platform.

Apart from shell-like syntax and handy shortcuts, the library provides local and remote command execution (over SSH), local and remote file-system paths, easy working-directory and environment manipulation, quick access to ANSI colors, and a programmatic Command-Line Interface (CLI) application toolkit. Now let’s see some code!


Cheat Sheet


>>> from plumbum import local
>>> ls = local["ls"]
>>> ls
LocalCommand(<LocalPath /bin/ls>)
>>> ls()
>>> notepad = local["c:\\windows\\notepad.exe"]
>>> notepad()                                   # Notepad window pops up
''                                              # Notepad window is closed by user, command returns

Instead of writing xxx = local["xxx"] for every program you wish to use, you can also import commands:

>>> from plumbum.cmd import grep, wc, cat, head
>>> grep
LocalCommand(<LocalPath /bin/grep>)

Or, use the local.cmd syntactic-sugar:

>>> local.cmd.ls
LocalCommand(<LocalPath /bin/ls>)
>>> local.cmd.ls()

See Local Commands.


>>> chain = ls["-a"] | grep["-v", "\\.py"] | wc["-l"]
>>> print(chain)
/bin/ls -a | /bin/grep -v '\.py' | /usr/bin/wc -l
>>> chain()

See Pipelining.


>>> ((cat < "setup.py") | head["-n", 4])()
'#!/usr/bin/env python3\nimport os\n\ntry:\n'
>>> (ls["-a"] > "file.list")()
>>> (cat["file.list"] | wc["-l"])()

See Input/Output Redirection.

Working-directory manipulation

>>> local.cwd
<Workdir /home/tomer/workspace/plumbum>
>>> with local.cwd(local.cwd / "docs"):
...     chain()

A more explicit, and thread-safe way of running a command in a different directory is using the .with_cwd() method:

.. code-block:: python
>>> ls_in_docs = local.cmd.ls.with_cwd("docs")
>>> ls_in_docs()

See Paths and The Local Object.

Foreground and background execution

>>> from plumbum import FG, BG
>>> (ls["-a"] | grep["\\.py"]) & FG         # The output is printed to stdout directly
>>> (ls["-a"] | grep["\\.py"]) & BG         # The process runs "in the background"
<Future ['/bin/grep', '\\.py'] (running)>

See Background and Foreground.

Command nesting

>>> from plumbum.cmd import sudo
>>> print(sudo[ifconfig["-a"]])
/usr/bin/sudo /sbin/ifconfig -a
>>> (sudo[ifconfig["-a"]] | grep["-i", "loop"]) & FG
lo        Link encap:Local Loopback
          UP LOOPBACK RUNNING  MTU:16436  Metric:1

See Command Nesting.

Remote commands (over SSH)

Supports openSSH-compatible clients, PuTTY (on Windows) and Paramiko (a pure-Python implementation of SSH2):

>>> from plumbum import SshMachine
>>> remote = SshMachine("somehost", user = "john", keyfile = "/path/to/idrsa")
>>> r_ls = remote["ls"]
>>> with remote.cwd("/lib"):
...     (r_ls | grep["0.so.0"])()

See Remote.

CLI applications

import logging
from plumbum import cli

class MyCompiler(cli.Application):
    verbose = cli.Flag(["-v", "--verbose"], help = "Enable verbose mode")
    include_dirs = cli.SwitchAttr("-I", list = True, help = "Specify include directories")

    @cli.switch("-loglevel", int)
    def set_log_level(self, level):
        """Sets the log-level of the logger"""

    def main(self, *srcfiles):
        print("Verbose:", self.verbose)
        print("Include dirs:", self.include_dirs)
        print("Compiling:", srcfiles)

if __name__ == "__main__":

Sample output

$ python3 simple_cli.py -v -I foo/bar -Ispam/eggs x.cpp y.cpp z.cpp
Verbose: True
Include dirs: ['foo/bar', 'spam/eggs']
Compiling: ('x.cpp', 'y.cpp', 'z.cpp')

See Command-Line Interface (CLI).

Colors and Styles

from plumbum import colors
with colors.red:
    print("This library provides safe, flexible color access.")
    print(colors.bold | "(and styles in general)", "are easy!")
print("The simple 16 colors or",
      colors.orchid & colors.underline | '256 named colors,',
      colors.rgb(18, 146, 64) | "or full rgb colors" ,
      'can be used.')
print("Unsafe " + colors.bg.dark_khaki + "color access" + colors.bg.reset + " is available too.")

Sample output

This library provides safe color access.
Color (and styles in general) are easy!
The simple 16 colors, 256 named colors, or full hex colors can be used.
Unsafe color access is available too.

See Colors.

Development and Installation

The library is developed on GitHub, and will happily accept patches from users. Please use the GitHub’s built-in issue tracker to report any problem you encounter or to request features. The library is released under the permissive MIT license.


Plumbum supports Python 3.6-3.10 and PyPy and is continually tested on Linux, Mac, and Windows machines through GitHub Actions. Any Unix-like machine should work fine out of the box, but on Windows, you’ll probably want to install a decent coreutils environment and add it to your PATH, or use WSL(2). I can recommend mingw (which comes bundled with Git for Windows), but cygwin should work too. If you only wish to use Plumbum as a Popen-replacement to run Windows programs, then there’s no need for the Unix tools.

Note that for remote command execution, an openSSH-compatible client is required (also bundled with Git for Windows), and a bash-compatible shell and a coreutils environment is also expected on the host machine.

This project uses setuptools to build wheels; and setuptools_scm is required for building SDists. These dependencies will be handled for you by PEP 518 compatible builders, like build and pip 10+.


You can download the library from the Python Package Index (in a variety of formats), or run pip install plumbum directly. If you use Anaconda, you can also get it from the conda-forge channel with conda install -c conda-forge plumbum.

User Guide

The user guide covers most of the features of Plumbum, with lots of code-snippets to get you swimming in no time. It introduces the concepts and “syntax” gradually, so it’s recommended you read it in order. A quick reference guide is available.

API Reference

The API reference (generated from the docstrings within the library) covers all of the exposed APIs of the library. Note that some “advanced” features and some function parameters are missing from the guide, so you might want to consult with the API reference in these cases.


The local object is an instance of a machine.


The original purpose of Plumbum was to enable local and remote program execution with ease, assuming nothing fancier than good-old SSH. On top of this, a file-system abstraction layer was devised, so that working with local and remote files would be seamless.

I’ve toyed with this idea for some time now, but it wasn’t until I had to write build scripts for a project I’ve been working on that I decided I’ve had it with shell scripts and it’s time to make it happen. Plumbum was born from the scraps of the Path class, which I wrote for the aforementioned build system, and the SshContext and SshTunnel classes that I wrote for RPyC. When I combined the two with shell combinators (because shell scripts do have an edge there) the magic happened and here we are.


The project has been inspired by PBS (now called sh) of Andrew Moffat, and has borrowed some of his ideas (namely treating programs like functions and the nice trick for importing commands). However, I felt there was too much magic going on in PBS, and that the syntax wasn’t what I had in mind when I came to write shell-like programs. I contacted Andrew about these issues, but he wanted to keep PBS this way. Other than that, the two libraries go in different directions, where Plumbum attempts to provide a more wholesome approach.

Plumbum also pays tribute to Rotem Yaari who suggested a library code-named pyplatform for that very purpose, but which had never materialized.