sniffer

Overview

sniffer is a autotest tool for Python_ using the nosetest_ library.

NEW: sniffer can now be customize to run anything, see 'Advanced Usage'.

Sniffer will automatically re-run tests if your code changes. And with another third-party library (see below), the CPU usage of file system monitoring is reduced in comparison to pure-python solutions. However, sniffer will still work without any of those libraries.

.. _Python: http://python.org/ .. _nosetest: http://code.google.com/p/python-nose/

Looking for maintainers

I (@jeffh) am looking for a new maintainer to carry on this project to new heights. I'm currently leaving this project on maintance mode (respond to issues, merge pull requests), but I'm not dedicating most of my free time towards this project.

Contact me on twitter (@jeffhui) or via email if you're interested in taking over the helm of this project.

Usage

To install::

pip install sniffer

Simply run sniffer in your project directory.

You can use sniffer --help for options And like autonose_, you can pass the nose arguments with -x prefix: -x--with-doctest or -x--config.

The problem with autonose_, is that the autodetect can be slow to detect changes. This is due to the pure python implementation - manually walking through the file system to see what's changed [#]_. Although the default install of sniffer shares the same problem, installing a third-party library can help fix the problem. The library is dependent on your operating system:

• If you use Linux, you'll need to install pyinotify_.
• If you use Windows, you'll need to install pywin32_.
• If you use Mac OS X 10.5+ (Leopard), you'll need to install MacFSEvents_.

If you want support for other notification systems, you can install:

• gntp_ for Growl_ support (Mac OS X).
• osxnotify_ and libosxnotify_ for native OS X notifications (Max OS X 10.9.4 and newer)
• py-notify_ for LibNotify_ support (Linux).

.. [#] This has been resolved in subsequent autonose versions, using watchdog. .. _nose: http://code.google.com/p/python-nose/ .. _easy_install: http://pypi.python.org/pypi/setuptools .. _pip: http://pypi.python.org/pypi/pip .. _autonose: http://github.com/gfxmonk/autonose .. _pyinotify: http://trac.dbzteam.org/pyinotify .. _pywin32: http://sourceforge.net/projects/pywin32/ .. _MacFSEvents: http://pypi.python.org/pypi/MacFSEvents/0.2.1 .. _gntp: https://github.com/kfdm/gntp/ .. _Growl: http://growl.info .. _py-notify: http://home.gna.org/py-notify .. _LibNotify: http://developer-next.gnome.org/libnotify/ .. _osxnotify: https://github.com/tomekwojcik/osxnotify-python .. _libosxnotify: https://github.com/tomekwojcik/libosxnotify

Advanced Usage

Don't want to run nose? You can do whatever you really want. Create a scent.py file in your current working directory. Here's an example of what you can do so far:

.. code-block:: python

from sniffer.api import * # import the really small API import os, termstyle

you can customize the pass/fail colors like this

pass_fg_color = termstyle.green pass_bg_color = termstyle.bg_default fail_fg_color = termstyle.red fail_bg_color = termstyle.bg_default

All lists in this variable will be under surveillance for changes.

watch_paths = ['.', 'tests/']

this gets invoked on every file that gets changed in the directory. Return

True to invoke any runnable functions, False otherwise.

This fires runnables only if files ending with .py extension and not prefixed

with a period.

@file_validator def py_files(filename): return filename.endswith('.py') and not os.path.basename(filename).startswith('.')

This gets invoked for verification. This is ideal for running tests of some sort.

For anything you want to get constantly reloaded, do an import in the function.

sys.argv[0] and any arguments passed via -x prefix will be sent to this function as

it's arguments. The function should return logically True if the validation passed

and logicially False if it fails.

This example simply runs nose.

@runnable def execute_nose(*args): import nose return nose.run(argv=list(args))

And that's the basic case. Nothing too fancy shmanshe. You can have multiple file_validator and runnable decorators if you want.

There is also support for selecting a runnable function by file validator. Useful if you want to run nose for Python files, mocha for JavaScript files, and csslint for CSS. Or any other combination you can come up with. For example:

.. code-block:: python

# Here we instruct the 'python_tests' runnable to be kicked off
# when a .py file is changed
@select_runnable('python_tests')
@file_validator
def py_files(filename):
    return filename.endswith('.py') and not os.path.basename(filename).startswith('.')


# Here we instruct the 'javascript_tests' runnable to be kicked off
# when a .js file is changed
@select_runnable('javascript_tests')
@file_validator
def js_files(filename):
    return filename.endswith('.js') and not os.path.basename(filename).startswith('.')


@runnable
def python_tests(*args):
    import nose
    return nose.run(argv=list(args))


@runnable
def javascript_tests(*args):
    command = "mocha tests/js-tests.js"
    return call(command, shell=True) == 0

This will run the nose for modifications to Python files and mocha when JavaScript files are changed.

Other Uses

Running with Other Test Frameworks

If you want to run another unit testing framework, you can do so by overriding sniffer.Sniffer, which is the class that handles running tests, or whatever you want. Specifically, you'll want to override the run, method to configure what you need to be done.

The property, test_args, are arguments gathered through --config=blah and -x.* configuration options. You should perform you imports inside the function instead of outside, to let the class reload the test framework (and reduce possibilities of multiple-run bugs).

After subclassing, set sniffer_instance parameter to your custom class when calling run or main.

Current Issues

For linux, there is an exception that is sometimes thrown when terminating.

Currently the program only looks for changes in the current working directory. This isn't the best solution: it doesn't understand how changes to your source code affects it.