NoseJS 0.9.1

Install

First, you need to download Rhino. More on that below.

You can get NoseJS with easy_install

$ easy_install NoseJS

Or you can clone the source using Mercurial from http://bitbucket.org/kumar303/nosejs/ and install it with

$ python setup.py develop

Usage

Here is the most basic way to invoke NoseJS (assumes you’ve downloaded Rhino into ~/src)

$ nosetests --with-javascript --rhino-jar ~/src/rhino1_7R1/js.jar

This command would look for any files along Nose’s path ending in .js that match Nose’s current test pattern, collect them all, then execute them using Rhino in a single Java subprocess at the end of all other tests. By default, files looking like test*.js will be collected and run.

The idea behind NoseJS is that you might have a Python web application that relies on JavaScript for some of its functionality and you want to run both Python and JavaScript tests with one command, nosetests. You can put these JavaScript tests wherever you want in your project.

Here is a more realistic example that shows how the Fudge project is tested simultaneously for Python and JavaScript functionality. Its project layout looks roughly like this:

|-- fudge
|   |-- __init__.py
|   |-- patcher.py
|   |-- tests
|   |   |-- __init__.py
|   |   |-- test_fudge.py
|   |   |-- test_patcher.py
|-- javascript
|   |-- fudge
|   |   |-- fudge.js
|   |   |-- tests
|   |   |   |-- test_fudge.html
|   |   |   `-- test_fudge.js
`-- setup.py

Both Python and JavaScript tests can be run with this command:

$ nosetests  --with-javascript \
             --rhino-jar ~/src/rhino1_7R1/js.jar \
             --with-dom \
             --js-test-dir javascript/fudge/tests/
......................................................
----------------------------------------------------------------------
Test Fake
  can find objects
  can create objects
  expected call not called
  call intercepted
  returns value
  returns fake
Test ExpectedCall
  ExpectedCall properties
  call is logged
Test fudge.registry
  expected call not called
  start resets calls
  stop resets calls
  global stop
  global clear expectations

Loaded 6 JavaScript files

OK
----------------------------------------------------------------------
Ran 54 tests in 0.392s

OK

The dots are the Python tests that were run and the output below that is what Fudge’s JavaScript test files printed out. Be sure to read the Caveats section below ;)

Specifying a path to JavaScript files

If JavaScript files are nested in a subdirectory, like the above example, specify that directory with:

$ nosetests --with-javascript --js-test-dir javascript/fudge/tests/ --js-test-dir ./another/dir

nosejs JavaScript namespace

All JavaScripts have the nosejs JavaScript namespace available for use. The following methods are available:

• nosejs.requireFile(path)

  • Load a JavaScript file from your test script. If you require the same file multiple times, it will only be loaded once. If the file does not start with a slash, then it should be a path relative to the directory of the script where requireFile() was called from. For example, here is how test_fudge.js requires the fudge library before testing:

    if (nosejs) {
        nosejs.requireFile("../fudge.js");
    }

• nosejs.requireResource(name)

  • Require a JavaScript file that is bundled with NoseJS. There are a few available resources:

    jquery-1.3.1.js

    Will load the JQuery library before loading any other tests

    jquery/qunit-testrunner.js

    Will load a very minimal set of JavaScript functions for testing. It is a partial implementation of the QUnit test runner interface. Supported methods: module(), test(), equals(), ok(), and expect()

    For example, test_fudge.js uses jquery and the testrunner

    if (nosejs) {
        nosejs.requireResource("jquery-1.3.1.js");
        nosejs.requireResource("jquery/qunit-testrunner.js");
        nosejs.requireFile("../fudge.js");
    }

Using the DOM

If your JavaScript under test relies on a browser-like DOM environment, it might still work! Just run:

$ nosetests --with-javascript --with-dom

This will load a copy of John Resig’s env.js script to simulate a DOM before loading any other JavaScript. There are a few very minor patches made to env.js, marked with @@nosejs in the NoseJS source.

Caveats

• Currently if JavaScript tests fail, nosetests will not indicate failure and its exit status will be unaffected. I can’t figure out a clean way to do this. Please get in touch if you’d like to help.

• In general, JavaScript tests are not very well integrated into Nose.

Wait … Python and Java?

Rhino is pretty much the only stable, command line oriented implementation of JavaScript I know of and it’s well supported by Mozilla.

Alternatively, John J. Lee created a Python binding to the fast Spider Monkey engine, Atul Varma revived it once in a fork on Google Code, and now Paul J. Davis is working on it (in git and it is released on PyPI). I have only had success in building Atul’s fork and unfortunately the error messages leave a lot to be desired. However, I have a basic implementation of NoseJS that supports a --spidermonkey option, which switches out the JavaScript engine. The code is here in a fork: http://bitbucket.org/kumar303/nosejs-spidermonkey/.

Once Python Spidermonkey becomes a bit more stable it would be nicer than Rhino because tests could be executed inline alongside other Python tests without the overhead of starting up a Java process.

Contributing

Please submit bugs and patches. All contributors will be acknowledged. Thanks!

Changelog

• 0.9.1

  • Fixed distribution problem

• 0.9

  • Initial release