Pythonic command runner


Keywords
subprocess, commands
License
MIT
Install
pip install commandlib==0.3.5

Documentation

CommandLib

CommandLib is a pythonic wrapper around subprocess that lets you pass around command objects and daisy-chain:

  • Arguments
  • Paths
  • Other environment variables
  • Runtime directory
  • Other runtime properties (run in shell, conceal stdout/stderr, ignore error codes, etc.)

It is somewhat inspired by amoffat's sh, Kenneth Reitz's requests, jaraco's path.py and SQLAlchemy.

To install:

$ pip install commandlib

Example:

>>> from commandlib import Command
>>> ls = Command("ls")
>>> ls("-t").in_dir("/").with_shell().run()
sys  tmp  run  dev  proc  etc  boot  sbin  root  vmlinuz  initrd.img  bin  lib  opt  vmlinuz.old  initrd.img.old  media  home  cdrom  lost+found  var  srv  usr  mnt

CommandPath example:

>>> from commandlib import CommandPath
>>> bin = CommandPath("/bin")
>>> bin.ls("-t").in_dir("/").run()
sys  tmp  run  dev  proc  etc  boot  sbin  root  vmlinuz  initrd.img  bin  lib  opt  vmlinuz.old  initrd.img.old  media  home  cdrom  lost+found  var  srv  usr  mnt

API

>>> from commandlib import Command, run

# Create command object
>>> py = Command("/usr/bin/python")

# Run with *additional* environment variable PYTHONPATH (*added* to global environment when command is run)
>>> py = py.with_env(PYTHONPATH="/home/user/pythondirectory")

# Run with additional path (appended to existing PATH environment variable when command is run)
>>> py = py.with_path("/home/user/bin")

# Run in specified directory (default is current directory)
>>> py = py.in_dir("/home/user/mydir")

# Run in shell
>>> py = py.with_shell()

# Suppress stderr
>>> py = py.silently() # Suppress stdout and stderr

# Finally run command explicitly with all of the above
>>> run(py)
>>> py.run() # alternative syntax

Why?

Commandlib is a library to make it easier to pass around immutable (sort of) command objects between different modules and classes and incrementally modify the command's behavior in a readable way - adding environment variables, paths, etc.

  • call, check_call and Popen do not have the friendliest of APIs and code that uses them a lot can get ugly fast.
  • sh does a similar thing but has a lot of magic (e.g. overriding import).
  • envoy and sarge are more focused on chaining commands rather than arguments, environment variables, etc.

Advanced API

Add trailing arguments:

>>> from commandlib import Command, run
>>> manage = Command("/usr/bin/python", "manage.py").with_trailing_arguments("--settings", "local_settings.py").in_dir("projectdir")
>>> run(manage("runserver"))
[ Runs "/usr/bin/python manage.py runserver --settings local_settings.py" inside projectdir ]

Dynamically generate command bundles from directories with executables in them:

>>> from commandlib import CommandPath, Command, run
>>> postgres94 = CommandPath("/usr/lib/postgresql/9.4/bin/")
>>> run(postgres94.postgres)
[ Runs postgres ]

>>> run(postgres94.createdb)
[ Runs createdb ]

Use with path.py (or any other library where str(object) resolves to a string:

>>> from path import Path
>>> postgres94 = CommandPath(Path("/usr/lib/postgresql/9.4/bin/"))
>>> run(postgres94.postgres)