= Test::Unit Given

Author::  Dave Copeland (mailto:davetron5000 at g mail dot com)
Copyright:: Copyright (c) 2011 by Dave Copeland
License:: Distributes under the Apache License, see LICENSE.txt in the source distro

Get your Test::Unit test cases fluent, without RSpec, magic, or crazy meta-programming.

This gives you some simple tools to make your test cases readable:

* Given/When/Then to delineate which parts of your tests do what
* A method "test_that" that defines test cases with strings instead of method names (much like +test+ in Rails tests)
* A module Any that defines methods to create arbitrary values, so it's clear in your tests what values are what, but without having to use long method names like `Faker::Lorem.words(2).join(' ')`

== Install

    gem install test_unit-given

== `test_that`

    class SomeTest < Test::Unit::Given::TestCase
      test_that {
        # do a test
      }

      test_that "description of our test" do
        # do a test
      end
    end

Yes, just like `test` from Rails, but without having to have ActiveSupport

== Given/When/Then

    class Circle
      def initialize(radius)
        @radius = radius
      end

      def area
        @radius * @radius * 3.14
      end

    end

    require 'test/unit/given'

    class CircleTest < Test::Unit::Given::TestCase
      test_that {
        Given {
          @circle = Circle.new(10)
        }
        When {
          @area = @circle.area
        }
        Then {
          assert_equal 314,@area
        }
      }
    end

You can, of course, provide a description for your test if you need to:

    class CircleTest < Test::Unit::Given::TestCase
      test_that "the area is correctly calculated" do
        Given {
          @circle = Circle.new(10)
        }
        When {
          @area = @circle.area
        }
        Then {
          assert_equal 314,@area
        }
      end
    end

You can use as much, or as little, of this library as you want.  If you don't like +test_that+, it's no problem:

    class CircleTest < Test::Unit::TestCase
      include Test::Unit::Given::Simple

      def test_that_area_is_calculated
        Given {
          @circle = Circle.new(10)
        }
        When {
          @area = @circle.area
        }
        Then {
          assert_equal 314,@area
        }
      end
    end

If you just want to use the +Given+, you can:

    class CircleTest < Test::Unit::TestCase
      include Test::Unit::Given::Simple

      def test_that_area_is_calculated
        Given {
          @circle = Circle.new(10)
        }
        area = @circle.area
        assert_equal 314,@area
      end
    end

Feel a +Given+ is too verbose?

    class CircleTest < Test::Unit::TestCase
      include Test::Unit::Given::Simple

      def test_that_area_is_calculated
        When {
          @area = Circle.new(10).area
        }
        Then {
          assert_equal 314,@area
        }
      end
    end

Use whatever makes sense; this is here to make your tests readable and communicate your intent, *not* lock you into some particular way of writing your tests.

== How does it work?

+Given+/+When+/+Then+/+And+/+But+ are all the same method under the covers.  They take a block and execute it immediately.  By using instance variables, you can send information between blocks.  This is actually a feature, since it means than any instance variables are important while local variables are just there to set up your test or help evalulate things.

This means that you can make methods that return blocks as a means of re-use.

    class CircleTest < Test::Unit::Given::TestCase

      def circle_with_radius(r)
        lambda { @circle = Circle.new(r) }
      end

      def get_area
        lambda { @area = @circle.area }
      end

      def area_should_be(area)
        lambda { assert_equal area,@area }
      end

      test_that {
        Given circle_with_radius(10)
        When get_radius
        And {
          @diameter = @circle.diameter
        }
        Then area_should_be(314)
        And {
          assert_equal 20,@diameter
        }
      }
    end

I would not recommend doing this a lot, as it can make things very confusing.  You could just as easily continue to use methods.

+test_that+ works just like +test+ in Rails, except that it doesn't require a name.  If your test is short enough, naming it might make things more confusing.  I tend to always name mine, but on occasion it gets in the way.

=== What about mocks?

Mocks create an interesting issue, because the "assertions" are the mock expectations you setup before you call the method under test.  This means that the "then" side of things is out of order.

    class CircleTest < Test::Unit::Given::TestCase
      test_that "our external diameter service is being used" do
        Given {
          @diameter_service = mock()
          @diameter_service.expects(:get_diameter).with(10).returns(400)
          @circle = Circle.new(10,@diameter_service)
        }
        When  {
          @diameter = @circle.diameter
        }
        Then {
          // assume mocks were called
        }
      end
    end

This is somewhat confusing.  We could solve it using two blocks provided by this library, +the_test_runs+, and +mocks_shouldve_been_called+, like so:

    class CircleTest < Test::Unit::Given::TestCase
      test_that "our external diameter service is being used" do
        Given {
          @diameter_service = mock()
        }
        When the_test_runs
        Then {
          @diameter_service.expects(:get_diameter).with(10).returns(400)
        }
        Given {
          @circle = Circle.new(10,@diameter_service)
        }
        When  {
          @diameter = @circle.diameter
        }
        Then mocks_shouldve_been_called
      end
    end

Although both <tt>the_test_runs</tt> and <tt>mocks_shouldve_been_called</tt> are no-ops,
they allow our tests to be readable and make clear what the assertions are that we are making.

Yes, this makes our test a bit longer, but it's *much* more clear.

=== What about block-based assertions, like +assert_raises+

Again, things are a bit out of order, but if you invert Then and When, you'll still get a readable test:

    class CircleTest < Test::Unit::Given::TestCase

      test_that "there is no diameter method" do
        Given {
          @circle = Circle.new(10)
        }
        Then {
          assert_raises NoMethodError do
            When {
              @circle.diameter
            }
          end
        }
      end
    end

== Any

Our tests tend to have a lot of arbitrary or meaningless values.  They also have a lot of very important and meaningfule values.  Often both of these are expressed as literals in our code.  What the Any module gives you is the ability to hide arbitrary values behind a method call.  This will ensure that the values truly are arbitrary, and will also mean that any literals left in your tests are important.  For example, you might have a test like this:


    def test_something
      service = mock()
      service.expects(:remote_call).with({ :method => 'process', :amount => 45.6}).returns(87)
      object_under_test = ObjectUnderTest.new('foo',service)

      assert_equal 8700,object_under_test.doit
    end

What's being tested here?  What values are meaningfule and which aren't?  Let's apply Any to make it clear:

    def test_something
      service = mock()
      service_return = any_int
      service.expects(:remote_call).with({ :method => any_string, :amount => any_number}).returns(service_return)
      object_under_test = ObjectUnderTest.new(any_string,service)

      assert_equal (service_return * 100),object_under_test.doit
    end

*Now* it's clear that we're expecting our object under test to multiple the return of our service call by 100.  The only value that has any meaning to this test is the integer 100, and that's the only literal that's there.  Beauty.

=== What about Faker or Sham?

They simply don't provide a concise API to do this, nor do they really communicate this concept.  We aren't passing _fake_ strings or numbers, we're passing *arbitrary* strings and numbers.  It's worth making that clear in our tests that certainly values that must not be nil, don't matter to the test.  That they are random each time keeps us honest.

=== What if I need some sort of restriction?

Two ways to do that.  The built-in any_* methods provide limited options:

    def test_truncate
      person = Person.new
      person.name = any_string :min => 256
      person.age = any_int :positive
      person.save!
      assert_equal 255,person.name.length
    end

The second way is to create your own any:

    def setup
      new_any :age do |options|
        age = any_number % 80
        age += 18 if options == :adult
        age
      end
    end

    def test_truncate
      person = Person.new
      person.name = any_string :min => 256
      person.age = any :age, :adult
      person.save!
      assert_equal 255,person.name.length
    end


== WTF?  Why?

Just because you're using Test::Unit doesn't mean you can't write fluent, easy to understand tests.
You really don't need RSpec, and RSpec has some baggage, such as nonstandard assignment, confusing class_eval
blocks, and generally replaces stuff you can do in plain Ruby.  Here, everything is simple, plain Ruby.  No
magic, nothing to understand. 

If you like Test::Unit, and you want to make your tests a bit more readable, this is for you.