Testing shell commands from Python

如何测试shell命令?最近，我遇到了一些情况，我想运行shell命令进行测试，Python称为万能胶水语言,一些自动化测试都可以完成,目前手头的工作都是用python完成的.但是无法从Python中找到有关如何进行测试的教程。在google上搜索了很久之后，我找到了一个适合我的解决方案，也许它也适合你!

为什么使用Python进行测试?

您可以使用专用工具来测试shell命令。为什么选择Python而不是其他语言或者工具?但是如果您经常使用Python，那么使用Python是有意义的，因为Python已经包含了健壮的测试功能，这些功能很容易与其他工具集成。通过测试Python中的shell命令，您可以利用这些工具，并避免跟踪不同地方的测试。另外，如果您已经熟悉用Python编写测试，那么为shell命令编写测试就变得轻而易举了。

Use cases

For the Netherlands eScience Center Python package template, I wanted tests to verify that the generated package can be installed, that the tests can be run, and that the documentation can be generated without errors. My Python text processing package nlppln contains CWL specifications of text mining tools, that can be validated by running them using a command line tool called cwltool. Another use case would be testing your package’s console scripts(although in this case it might be more convenient to use a package for creating command line interfaces that comes with built-in testing functionality, such as Click).

The sh package

You can run shell commands from Python using the subprocess module from the Python standard library. However, using this module is a hassle, because you have to do all the error handling yourself. Sh is a Python package that takes care of all that and allows you to run shell commands using a single line of code. If you want to run python setup.py install, all you have to do is:

import sh

sh.python(['setup.py', 'install'])

If you want to run foo and it is installed on your system, you can just do sh.foo().

Writing a test

So how can we use this for testing? Let’s look at an example. For the Python template, I want to test whether a project generated from the cookiecuttertemplate can be installed without errors. The goal of the template is to help users write high quality code with less effort, and having an installable empty project is a good first step. The code for the test that tests the installation is:

import pytest
import os
import sh

def test_install(cookies):
  # generate a temporary project using the cookiecutter
  # cookies fixture
  project = cookies.bake()                                                  

  # remember the directory where tests should be run from
  cwd = os.getcwd()
  # change directories to the generated project directory 
  # (the installation command must be run from here)
  os.chdir(str(project.project))

  try:
    # run the shell command
    sh.python(['setup.py', 'install'])
  except sh.ErrorReturnCode as e:
    # print the error, so we know what went wrong
    print(e)
    # make sure the test fails
    pytest.fail(e)
  finally:
    # always change directories to the test directory
    os.chdir(cwd)

That is all there is to it!

More examples

Of course there is a lot more you can do, e.g., checking whether files existafter running a shell command, or verifying the contents of generated files. What use cases can you come up with?

On Windows: use subprocess

Sh does not work on Windows. If you need to test shell commands on Windows, you are stuck with subprocess. Provenance tracking package recipycontains some nice examples of tests using subprocess that might help you on your way.

安装sh

https://amoffat.github.io/sh/