Morelia Viridis

Morelia viridis is a PythonLanguage BehaviorDrivenDevelopment platform, conceptually derived from RubyLanguage's CucumberFramework.

It is available here: ; and is forkable here:


Install it with:

  sudo pip install Morelia
To use it, first write a project.feature file, in ordinary prose, like this:

Note that "In order", "As a", and "I want" are not Morelia keywords. They are part of Feature's "predicate"; its text payload.

The other blue words are keywords. The black words are executable test specifications.

Now create a _standard_ PythonUnit test suite, like this:

  class MoreliaTest(TestCase):

def setUp(self): self.stack = []

def step_I_have_entered_a_number_into_the_calculator(self, number): r'I have entered (\d+) into the calculator' self.stack.append(int(number))

def step_I_press_add(self): # matched by method name self.result = sum(self.stack)

def step_the_result_should_be_on_the_screen(self, number): "the result should be (\d+) on the screen" assert int(number) == self.result
Note that Morelia does not waste anyone's time inventing a new testing back-end just to add a layer of literacy over our testage. Steps are miniature TestCases. Your OnsiteCustomer need never know, and your UnitTests and CustomerTests can share their support methods. The same OneTestButton? can run all TDD and BDD tests.

Next, note that Morelia matches Steps in your Feature file to either the names or doc-strings of step_ methods in your test case. And it expands regular expressions, such as (\d+), into step arguments, such as number. Remember to use tight expressions, such as (\d+), not loosey-goose expressions like (\d*) or (.*), to validate your input.

When you run your TestCase, hook into all your feature files, like this:

      def test_evaluate_file(self):
        from morelia import Parser
The passing steps will appear as passing test cases in your test run.

And note that Morelia calls setUp() and tearDown() around your Scenario. Each step calls within one TestCase, so self. can store variables between each step. This implies, to use a database, you must have a DjangoProject fixtures=[] line in your test suite, to roll its transactions back. Otherwise, your second test batch will use the data records left over from the first batch, breaking TestIsolation.


To DRY up a series of redundant scenarios, varying by only "payload" variables, roll the Scenarios up into a table, using <angles> around the payload variable names:

 Scenario: orders above $100.00 to the continental US get free ground shipping
   When we send an order totaling $<total>, with a 12345 SKU, to our warehouse
    And the order will ship to <destination>
   Then the ground shipping cost is $<cost>
    And <rapid> delivery might be available

| total | destination | cost | rapid |

| 98.00 | Rhode Island | 8.25 | yes | | 101.00 | Rhode Island | 0.00 | yes | | 99.00 | Kansas | 8.25 | yes | | 101.00 | Kansas | 0.00 | yes | | 99.00 | Hawaii | 8.25 | yes | | 101.00 | Hawaii | 8.25 | yes | | 101.00 | Alaska | 8.25 | yes | | 99.00 | Ontario, Canada | 40.00 | no | | 99.00 | Brisbane, Australia | 55.00 | no | | 99.00 | London, United Kingdom | 55.00 | no | | 99.00 | Kuantan, Malaysia | 55.00 | no | | 101.00 | Tierra del Fuego | 55.00 | no |
That Scenario will unroll into a series of scenarios, each with one value from the table inserted into their placeholders <total>, <destination>, and <rapid>. So this step method will receive each line in the "destination" column:

    def step_the_order_will_ship_to_(self, location):
        r'the order will ship to (.*)'
(And observe that naming the placeholder the same as the method argument is a reeeally good idea, but naturally unenforceable.)

MV will take each line of the table, and construct a complete test case out of the Scenario steps, running setUp() and tearDown() around them.


The other step keywords (Given, And, Then, etc.) are cosmetic, to permit good grammar. They are all aliases for Step. The committee may eventually find specific uses for them.

The When keyword, however, is special. When a Scenario contains more than one When, Morelia splits it up into one Scenario for each When block, and runs each one separately. So the following two Feature details are equivalent...

 Scenario: Split When Blocks
     Given some setup
       And some condition
      When a first trigger occurs
      Then something good happens

Scenario: Split When Blocks again Given some setup And some condition When another trigger occurs Then something else happens

 Scenario: Split When Blocks, and again
     Given some setup
       And some condition

When a first trigger occurs Then something good happens

When another trigger occurs Then something else happens
The second version DRYs the setup conditions.

The committee does not yet know what happens if a multi-When Scenario also contains a |table|, so please don't rely on whatever the current behavior is!

Here's another sneaky snake, which might also be a Green Tree Python (a Morelia viridis):

CategoryFramework CategoryPython

View edit of January 1, 2011 or FindPage with title or text search