nose-testconfig 0.4

Test Configuration plugin for nosetests.

• Project hosting: <http://code.google.com/p/nose-testconfig/>

Contents

• About

• What It Does

• Test Usage

• Defining a configuration file

• Command line options

• Changes & News

About

Written by Jesse Noller Licensed under the Apache Software License, 2.0

You can install it with easy_install nose-testconfig

What It Does

nose-testconfig is a plugin to the nose test framework which provides a faculty for passing test-specific (or test-run specific) configuration data to the tests being executed.

Currently configuration files in the following formats are supported:

• YAML (via PyYAML)

• INI (via ConfigParser)

• Pure Python (via Exec)

The plugin is meant to be flexible, ergo the support of exec’ing arbitrary python files as configuration files with no checks. The default format is assumed to be ConfigParser ini-style format.

The plugin provides a method of overriding certain parameters from the command line (assuming that the main “config” object is a dict) and can easily have additional parsers added to it.

Test Usage

For now (until something better comes along) tests can import the “config” singleton from testconfig:

from testconfig import config

By default, YAML files parse into a nested dictionary, and ConfigParser ini files are also collapsed into a nested dictionary for foo[bar][baz] style access. Tests can obviously access configuration data by referencing the relevant dictionary keys:

from testconfig import config
def test_foo():
    target_server_ip = config['servers']['webapp_ip']

Warning: Given this is just a dictionary singleton, tests can easily write into the configuration. This means that your tests can write into the config space and possibly alter it. This also means that threaded access into the configuration can be interesting.

When using pure python configuration - obviously the “sky is the the limit” - given that the configuration is loaded via an exec, you could potentially modify nose, the plugin, etc. However, if you do not export a config{} dict as part of your python code, you obviously won’t be able to import the config object from testconfig.

When using YAML-style configuration, you get a lot of the power of pure python without the danger of unprotected exec() - you can obviously use the pyaml python-specific objects and all of the other YAML creamy goodness.

Defining a configuration file

Simple ConfigParser style:

[myapp_servers]
main_server = 10.1.1.1
secondary_server = 10.1.1.2

So your tests access the config options like this:

from testconfig import config
def test_foo():
    main_server = config['myapp_servers']['main_server']

YAML style configuration::

myapp:

servers:

main_server: 10.1.1.1 secondary_server: 10.1.1.2

And your tests can access it thus:

from testconfig import config
def test_foo():
    main_server = config['myapp']['servers']['main_server']

Python configuration file:

import socket

global config
config = {}
possible_main_servers = ['10.1.1.1', '10.1.1.2']

for srv in possible_main_servers:
    try:
        s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
        s.connect((srv, 80))
    except:
        continue
    s.close()
    config['main_server'] = srv
    break

And lo, the config is thus:

from testconfig import config
def test_foo():
    main_server = config['main_server']

If you need to put python code into your configuration, you either need to use the python-config file faculties, or you need to use the !!python tags within PyYAML/YAML - raw ini files no longer have any sort of eval magic.

Command line options

After it is installed, the plugin adds the following command line flags to nosetests:

--tc-file=TESTCONFIG  Configuration file to parse and pass to tests
                      [NOSE_TEST_CONFIG_FILE]

--tc-format=TESTCONFIGFORMAT  Test config file format, default is
                              configparser ini format
                              [NOSE_TEST_CONFIG_FILE_FORMAT]

--tc=OVERRIDES        Option:Value specific overrides.

--tc-exact            Optional: Do not explode periods in override keys to
                      individual keys within the config dict, instead treat
                      them as config[my.toplevel.key] ala sqlalchemy.url in
                      pylons.

Passing in an INI configuration file:

$ nosetests -s --tc-file example_cfg.ini

Passing in a YAML configuration file:

$ nosetests -s --tc-file example_cfg.yaml --tc-format yaml

Passing in a Python configuration file:

$ nosetests -s --tc-file example_cfg.py --tc-format python

Overriding a configuration value on the command line:

$ nosetests -s --tc-file example_cfg.ini --tc=myvalue.sub:bar

Overriding multiple key:value pairs:

$ nosetests -s --tc-file example_cfg.ini --tc=myvalue.sub:bar \
    --tc=myvalue.sub2:baz --tc=myvalue.sub3:bar3

Warning: When using the –tc= flag, you can pass it in as many times as you want to override as many keys/values as needed, however you can not use it to add in new keys: The configuration key must already be defined. The format is in parent.child.child = value format - the periods are translated into keys within the config dict, for example:

myvalue.sub2:baz = config[myvalue][sub2] = baz

You can override the explosion of the periods by passing in the –tc-exact argument on the command line.

Changes & News

0.4:

• Per feedback from Kumar and others, the eval()’ing of ini-file values has been removed: allowing arbitrary python in the values was more annoying less standard then was worth it.

• Added the –tc-exact command line flag, to block the exploding of name.name values into dicts-within-dicts

• Updated the docs to parse right.

0.3:

Fix documentation examples per Kumar’s feedback.

0.2:

Fix pypi packaging issues

0.1:

Initial release. May contain bits of glass.