param_test

Parameterized unit tests for Ruby/ActiveSupport

This gem extends ActiveSupport::TestCase with a simple param_test class method that generates multiple tests with different parameters from a single code block.

In unit tests, you'll often want to run a set of the same assertions on multiple inputs. One way to do this is by looping over the inputs, for example:

class StringTest < ActiveSupport::TestCase

  test "strings are ASCII only" do
    ["foo", "bar", "baz"].each do |string|
      # Bad programmer! Don't do this
      assert string.ascii_only?
    end
  end

end

Don't do this! This is almost always a bad idea. If this test fails you won't know at which of the 3 input parameters it failed. A failure will also prevent subsequent inputs from getting tested, masking other failures that you won't discover until later.

param_test addresses these problems. The above test should be rewritten like this:

require 'param_test'

class StringTest < ActiveSupport::TestCase

  param_test "string %s is ASCII only",
  ["foo", "bar", "baz"] do |string|
    assert string.ascii_only?
  end

end

This will generate three separate tests, one for each of the three parameters ("foo", "bar" and "baz") passed in the second argument. Each test is independent and can fail independently.

The tests will be named after the description template passed as the first argument. Standard Ruby string formatting is used to substitute the parameters into the description. Generally you'll want to just use string substitution with %s.

You can have multiple parameters per test:

param_test "%s is uppercase %s",
[["FOO", "foo"], ["BAR", "bar"]] do |expected, param|
  assert_equal expected, param.upcase
end

Each parameter set must have the same number of parameters. The description template must have a corresponding number of substitutions.

Ruby's underlying unit test framework requires tests to be defined as methods beginning with test, so the description is actually converted to a valid method name. In the last example, the methods generated are :test_FOO_is_uppercase_foo and :test_BAR_is_uppercase_bar.

If string substitution would create the same method names for two different sets of parameters (this can happen because for example any all-whitespace string converts to the same single underscore), a counting variable will be added to the method name to keep them unique.

For better readability, any nil parameters will be string substituted as a "nil" string (the default string substitution for nil values is the empty string).

If using Rails 5.x, add to your Gemfile:

gem 'param_test', '~> 0.2.0'

If using Rails 4.x, add to your Gemfile:

gem 'param_test', '~> 0.1.1'

If using Rails 3.2, add to your Gemfile:

gem 'param_test', '~> 0.0.2'

(Since you generally only use param_test in tests, you probably want to add this to the group :test in your Gemfile.)

Distributed under an MIT license.