Main Content

Extend Function-Based Tests

Typically, with function-based tests, you create a test file and pass the file name to the runtests function without explicitly creating a suite of Test objects. However, if you create an explicit test suite, additional features are available in function-based testing. These features include:

  • Test logging and verbosity

  • Test selection

  • Plugins to customize the test runner

For additional functionality, consider using class-based unit tests. For more information, see Ways to Write Unit Tests.

Fixtures for Setup and Teardown Code

When writing tests, use the applyFixture method to handle setup and teardown code for actions such as:

  • Changing the current working folder

  • Adding a folder to the path

  • Creating a temporary folder

  • Suppressing the display of warnings

These fixtures take the place of manually coding the actions in the setupOnce, teardownOnce, setup, and teardown functions of your function-based test.

For example, if you manually write setup and teardown code to set up a temporary folder for each test, and then you make that folder your current working folder, your setup and teardown functions could look like this.

function setup(testCase)
% store current folder
testCase.TestData.origPath = pwd;

% create temporary folder
testCase.TestData.tmpFolder = ['tmpFolder' datestr(now,30)];
mkdir(testCase.TestData.tmpFolder)

% change to temporary folder
cd(testCase.TestData.tmpFolder)
end

function teardown(testCase)
% change to original folder
cd(testCase.TestData.origPath)

% delete temporary folder
rmdir(testCase.TestData.tmpFolder)
end

However, you also can use a fixture to replace both of those functions with just a modified setup function. The fixture stores the information necessary to restore the initial state and performs the teardown actions.

function setup(testCase)
% create temporary folder
f = testCase.applyFixture(matlab.unittest.fixtures.TemporaryFolderFixture);

% change to temporary folder
testCase.applyFixture(matlab.unittest.fixtures.CurrentFolderFixture(f.Folder));
end

Test Logging and Verbosity

Your test functions can use the log method. By default, the test runner reports diagnostics logged at verbosity level 1 (Terse). Use the matlab.unittest.plugins.LoggingPlugin.withVerbosity method to respond to messages of other verbosity levels. Construct a TestRunner object, add the LoggingPlugin, and run the suite with the run method. For more information on creating a test runner, see Test Runner Customization.

Test Suite Creation

Calling your function-based test returns a suite of Test objects. You also can use the testsuite function or the matlab.unittest.TestSuite.fromFile method. If you want a particular test and you know the test name, you can use matlab.unittest.TestSuite.fromName. If you want to create a suite from all tests in a particular folder, you can use matlab.unittest.TestSuite.fromFolder.

Test Selection

With an explicit test suite, use selectors to refine your suite. Several of the selectors are applicable only for class-based tests, but you can select tests for your suite based on the test name:

Use these approaches in a suite generation method, such as matlab.unittest.TestSuite.fromFile, or create a suite and filter it using the selectIf method. For example, in this listing, the four values of suite are equivalent.

import matlab.unittest.selectors.HasName
import matlab.unittest.constraints.ContainsSubstring
import matlab.unittest.TestSuite.fromFile

f = 'rightTriTolTest.m';
selector = HasName(ContainsSubstring('Triangle'));

% fromFile, name-value pair
suite = TestSuite.fromFile(f,'Name','*Triangle*')

% fromFile, selector
suite = TestSuite.fromFile(f,selector)

% selectIf, name-value pair
fullSuite = TestSuite.fromFile(f);
suite = selectIf(fullSuite,'Name','*Triangle*')

% selectIf, selector
fullSuite = TestSuite.fromFile(f);
suite = selectIf(fullSuite,selector)

If you use one of the suite creation methods with a selector or name-value pair, the testing framework creates the filtered suite. If you use the selectIf method, the testing framework creates a full test suite and then filters it. For large test suites, this approach can have performance implications.

Test Running

There are several ways to run a function-based test.

To Run All TestsUse Function
In a fileruntests with the name of the test file
In a suiterun with the suite
In a suite with a custom test runnerrun. (See Test Runner Customization.)

For more information, see Run Tests for Various Workflows.

Programmatic Access of Test Diagnostics

In certain cases, the testing framework uses a DiagnosticsRecordingPlugin plugin to record diagnostics on test results. The framework uses the plugin by default if you do any of these:

  • Run tests using the runtests function.

  • Run tests using the testrunner function with no input.

  • Run tests using the run method of the TestSuite or TestCase classes.

  • Run performance tests using the runperf function.

  • Run performance tests using the run method of the TimeExperiment class.

After you run tests, you can access recorded diagnostics using the DiagnosticRecord field in the Details property on the TestResult object. For example, if your test results are stored in the variable results, then result(2).Details.DiagnosticRecord contains the recorded diagnostics for the second test in the suite.

The recorded diagnostics are DiagnosticRecord objects. To access particular types of test diagnostics for a test, use the selectFailed, selectPassed, selectIncomplete, and selectLogged methods of the DiagnosticRecord class.

By default, the DiagnosticsRecordingPlugin plugin records qualification failures and logged events at the matlab.automation.Verbosity.Terse level of verbosity. For more information, see DiagnosticsRecordingPlugin and DiagnosticRecord.

Test Runner Customization

Use a TestRunner object to customize the way the framework runs a test suite. With a TestRunner object you can:

  • Produce no output in the command window using the withNoPlugins method.

  • Run tests in parallel using the runInParallel method.

  • Add plugins to the test runner using the addPlugin method.

For example,use test suite, suite, to create a silent test runner and run the tests with the run method of TestRunner.

runner = matlab.unittest.TestRunner.withNoPlugins;
results = runner.run(suite);

Use plugins to customize the test runner further. For example, you can redirect output, determine code coverage, or change how the test runner responds to warnings. For more information, see Add Plugin to Test Runner and the plugins classes.

See Also

| | | | |

Related Topics