Unit Testing With MATLAB

Link

http://xtargets.com/cms/Downloads/cat_view-2.html

Version

2.4

Note

MUnit requires MTemplate, so you have to download both packages.

Link

http://sourceforge.net/project/showfiles.php?group_id=166995

Version

1.4

1. Run setups:

   MTemplate

   Start mtemplate-install-v1.4.exe and install MTemplate to $HOME/mtemplate.

   MUnit

   Start munit-install-v2.4.exe and install MUnit to $HOME/munit.

2. Add directories to MATLAB path:

   addpath('$HOME/mtemplate');
   addpath('$HOME/munit');

1. Unzip mlunit_1.3.zip to $HOME.
2. Add directory to MATLAB path:

   addpath('$HOME/mlunit/src');

File test_munit.m

function test = test_munit

test = munit_testcase;

    function setup
    end

    function teardown
    end

    function test_success
        x = 10;
        y = 10;
        
        test.assert(x == y);
    end

    function test_failure
        x = 10;
        y = 20;
        
        test.assert(x == y);
    end
end

File test_mlunit.m

function test = test_mlunit

test = load_tests_from_mfile(test_loader);

    function set_up
    end
    
    function tear_down
    end

    function test_success
        x = 10;
        y = 10;
        
        assert(x == y);
    end

    function test_failure
        x = 10;
        y = 20;
        
        assert(x == y);
    end
end

Type at the MATLAB command prompt:

suite = munit_testsuite('test_munit');
r = suite.run();
r.web();

The command r.web(); opens the MATLAB web browser with the following page:

With edit the user can open the MATLAB editor with the respective test (file and line), with run (and debug) execute the test again (in debugging mode) and with Hide Details hide as well as display the details of failed tests.

Option A

Type at the MATLAB command prompt:

run(text_test_runner(1, 2), test_mlunit);

The results are written to the standard output or the given file handle (the first parameter of text_test_runner):

test_mlunit/test_success(function_test_case) ... OK
test_mlunit/test_failure(function_test_case) ... FAIL

======================================================================
FAIL: test_mlunit/test_failure(function_test_case)
----------------------------------------------------------------------
Traceback (most recent call first): 
  In p:\temp\test_mlunit.m at line 22
  In p:\temp\mlunit\src\@function_test_case\run_test.m at line 11
  In p:\temp\mlunit\src\@test_case\run.m at line 32
  In p:\temp\mlunit\src\@test_suite\run.m at line 14
  In p:\temp\mlunit\src\@text_test_runner\run.m at line 14
AssertionError: no message.
----------------------------------------------------------------------
Ran 2 tests in 0.031s

FAILED (errors=0, failures=1)

Option B (new with Version 1.4)

Type at the MATLAB command prompt:

run(gui_test_runner, 'test_mlunit');

The command opens the following MATLAB figure, executes the tests and shows the result:

With Show the user can open the MATLAB editor with the respective failed test (file and line of failed assertion). With Run the test case can be executed again.

Bookmarking a Test Suite

MUnit allows on the front page of its user interface the creation of bookmarks to test suites.

Type at the MATLAB command prompt:

munit;

The user interface of MUnit is opened:

The bottom half presents a list of bookmarks to existing test suites, each bookmark with a link edit, run and remove. To add a new test suite to the list, simply type in the name of the test suite and press add.

Defining Assertions With Anonymous Functions or Constraints

The standard way of defining an assertion to compare two values is (nearly) the same with both frameworks:

test.assert(x == y);

The assert command of MUnit furthermore accepts an anonymous function handle as the input value:

test.assert(@() x == y);

The difference lies in the test report of a failed assertion. The following figure shows the test report of the above defined test method test_failure:

Now the assert statement of the test method is changed in that way, that it uses an anonymous function:

As you can see, the test report includes much more details about the reasons of the failed test. Therefore this is now the preferred way of defining assertions with MUnit.

With prior versions the recommended way of defining assertions was the use of constraints:

c = constraint;
test.assert(c.eq(x, y));

The constraint object supports a number of operators, eg. eq, ne, and, or, isa or ask. The last one allows the definition of interactive tests, as the argument is shown in a dialog box with two buttons Yes and No. A click to No lets the test fail.

The Green Bar

As most popular testing frameworks the user interface of mlUnit provides a progress bar, which shows how many tests already have been executed and even more important, whether all of them were successful (green bar) or one of them failed (red bar). You can watch the progress bar in action, if you for example execute the unit tests of mlUnit itself:

addpath('$HOME/mlunit/test');
run(gui_test_runner, 'mlunit_all_tests');

Class-Based Testing

In addition to the above shown example, which defines all test methods within one .m-file, mlUnit allows also the traditional class-based definition of tests. The following steps explain the definition of the example with this:

1. Create a folder for the class:

   mkdir('@test_mlunit_cb');

2. Change dir to the new folder:

   cd('@test_mlunit_cb');

3. Create an .m-file for the constructor of the class, which is inherited of test_case:

   File test_mlunit.m

   function self = test_mlunit_cb(name)
   
   tc = test_case(name);
   self = class(struct([]), 'test_mlunit_cb', tc);

4. Create an .m-file for each test method:

   File test_success.m

   function self = test_success(self)
   
   x = 10;
   y = 10;
   
   assert(x == y);

   File test_failure.m

   function self = test_failure(self)
   
   x = 10;
   y = 20;
   
   assert(x == y);

5. Execute the test case:

   cd('..');
   run(gui_test_runner, 'test_mlunit_cb');