Curtis "Ovid" Poe > Test-Class-Moose > Test::Class::Moose::Tutorial

Download:
Test-Class-Moose-0.54.tar.gz

Dependencies

Annotate this POD

Website

View/Report Bugs
Module Version: 0.54   Source  

NAME ^

Test::Class::Moose::Tutorial - A starting guide for Test::Class::Moose

VERSION ^

version 0.54

Getting Started ^

Automated testing is wonderful. Verifying your program's correctness in all possible ways is a good thing that will save you time (and programmer time is money).

Procedural tests like Test::More are a good, general way to write tests for all kinds of things. However, it is not very good when you're trying describe relationships between tests. For this, a class-based test would work better, because you could use the standard OO-techniques for describing object relationships like inheritance.

When testing objects, it's good for code re-use to have test classes that match the relationships between the regular objects. By creating test classes with the same relationships, you can quickly increase test coverage by testing the base class, and all the child classes can inherit those tests!

A Test Class

The first and most crucial part of using Test::Class::Moose is a class that runs some tests. Test::Class::Moose loads a few modules for you automatically, so the boilerplate is, at minimum:

    package TestsFor::My::Test::Class;
    use Test::Class::Moose;

Test::Class::Moose loads strict, warnings, Moose, and Test::Most (which includes Test::More, Test::Deep, Test::Exception, and Test::Differences).

I put my test classes in the t/lib/TestsFor directory, to keep the separated from my other classes that help testing (t/lib) and my other test scripts. This is just a convention; the directory can be anything you want it to be, but it is a good idea to keep your test classes separate from your other test-related modules.

Now, we need a method that runs our test. Test::Class::Moose test methods start with test_. Any method that starts with test_ will be run as a test.

    use My::Module;

    sub test_construction {
        my $test = shift;
        my $obj = My::Module->new;
        isa_ok $obj, 'My::Module';
    }

Every test_ method is run as a subtest, and no plan is required. We can have as many test_ methods as we want.

A Test Runner

Now that we have a test class, we need a way for prove to load and run them. Test::Class::Moose can load our test modules from a given directory, and it has a runtests() method that will run any test modules that have already been loaded.

    # t/test_class_tests.t
    use File::Spec::Functions qw( catdir );
    use FindBin qw( $Bin );
    use Test::Class::Moose::Load qw( catdir( $Bin, 't', 'lib' ) );
    Test::Class::Moose->new->runtests;

Or if you're not worried about the portability of that directory:

    use Test::Class::Moose::Load 't/lib';
    Test::Class::Moose->new->runtests;

This test script will load all of the Test::Class::Moose modules inside t/lib/ and then run them. All your test modules get run by this one script, but since they're run as subtests, you will get a report on how many test classes failed.

We can run our test script using prove. I'll turn on verbose output (-v) to show you what the TAP output looks like

    prove -v t/test_class_tests.t
    t/test_class_tests.t ..
    1..1
    #
    # Running tests for TestsFor::My::Class
    #
        1..1
        # TestsFor::My::Class->test_something()
            ok 1 - I tested something!
            1..1
        ok 1 - test_something
    ok 1 - TestsFor::My::Class
    ok
    All tests successful.
    Files=1, Tests=1,  0 wallclock secs ( 0.03 usr  0.01 sys +  0.34 cusr  0.01 csys =  0.39 CPU)
    Result: PASS

Event Hooks ^

There are various points in the test script where we might want to perform some actions: Reset a test database, create a temp file, or otherwise set up prerequisites for a test. Test::Class::Moose provides some hooks that allow us to perform actions at these points.

test_startup / test_shutdown

test_startup is run as the very first thing in our test class, and is run only once per test class. This allows us to set up some global things, like a database that will be used throughout the entire test.

test_shutdown is run once as the very last thing in our test class, and is run only once per test class. This allows us to clean up things from test_startup, and also test to verify that anything from test_startup looks exactly as it should before we clean it up.

test_setup / test_teardown

What test_startup and test_shutdown are for the entire test class, test_setup and test_teardown are for every single test_* method.

test_setup is run before every test method. For canonical unit testing, this is where you can create the things you need for each test, such as a log file, rebuilding fixtures, or starting a database transaction.

test_teardown happens after every test, and is where you can clean up the things created in test_setup, such as ending the database transaction. Note that some developers actually prefer their cleanup to happen in their test_setup method, prior to setting up the test. That sounds odd, but it means that if a test method fails, you haven't yet cleaned up and can easily inspect your test environment.

Test Class Composition ^

The most important reason to choose a class test over a procedural test (using only Test::More) is class composition.

Inheritance

Since we're using Moose, inheritance is as easy as:

    package TestsFor::My::Test::Class;
    use Test::Class::Moose;
    extends 'My::Test::Base';

Test::Class::Moose even provides a shortcut:

    package TestsFor::My::Text::Class;
    use Test::Class::Moose extends => 'My::Test::Base';

If My::Test::Base will not be testing anything itself, we do not put it in t/lib/TestsFor, instead we put it in lib or t/lib (depending on if we want it to be part of the public set of modules or not). This will make sure our test runner does not try to run our base class that doesn't test anything concrete.

Roles

If your distribution uses roles, so should your tests. Like inheritance, roles are added in the regular Moose way:

    package TestsFor::My::Test::Class;
    use Test::Class::Moose;
    with 'My::Test::Role';

If you want your role to also provide tests, make sure you use Test::Class::Moose::Role instead of Moose::Role.

Organizing Your Tests

Test code should be held to the same standard as the rest of the code in your distribution:

Don't Repeat Yourself

Copy/paste code isn't okay in your module code, and it should not be okay in your test code either! Refactor your tests to use roles or inheritance.

Advanced Features ^

plan

If you need to prepare a plan for your tests, you can do so using the plan() method:

    sub test_constructor {
        my $test = shift;
        $test->test_report->plan( 1 ); # 1 test in this sub
        isa_ok My::Module->new, 'My::Module';
    }

Using the plan() method, we can know exactly how many tests did not run if the test method ends prematurely, or how many extra tests were run if we had too many tests.

Alternately, you can use the Test (a single test) or Tests attributes to set the plan. If you do this, the method is marked as a test method even if it does not begin with test_.

    # 'Test' asserts a plan of 1 test
    sub test_constructor : Test {
        my $test = shift;
        isa_ok My::Module->new, 'My::Module';
    }

    # 'Tests' means multiple tests with no plan (note the test name)
    sub a_test_method : Tests {
        # many tests here
    }

    # 'Tests($integer) means a plan of $integer
    sub this_is_another_test : Tests(3) {
        # 3 tests
    }

skip

We can use the test_startup and test_setup methods to skip tests that we can't or don't want to run for whatever reason.

If we don't want to run a single test method, we can use the test_setup method and call the test_skip method with the reason we're skipping the test.

    sub test_will_fail {
        my ( $test ) = @_;
        fail "This doesn't work!";
    }
    
    sub test_setup {
        my $test = shift;
        if ( $test->test_report->current_method->name eq 'test_will_fail' ) {
            $test->test_skip( 'It doesn't work' );
        }
    }

If we don't want to run an entire class, we can use the test_startup method and the same test_skip method with the reason we're skipping the test.

    sub test_startup {
        my $test = shift;
        $test->test_skip( "The entire class doesn't work" );
    }

Run Specific Test Classes

One of the problems with having only one test script to run all the test classes is when we're working directly with one test class we still have to run all the other test classes.

To fix this problem, Test::Class::Moose allows us to specify which specific classes we want to run in its constructor:

    # t/test_class_tests.t
    use File::Spec::Functions qw( catdir );
    use FindBin qw( $Bin );
    use Test::Class::Moose::Load catdir( $Bin, 't', 'lib' );
    Test::Class::Moose->new(
        classes => [ 'TestsFor::My::Test::Class' ],
    )->runtests;

Now, we only run TestsFor::My::Test::Class instead of all the tests found in TestsFor::.

This isn't very elegant though, since we have to edit t/test_class_tests.t every time we want to run a new test. So, Test::Class::Moose can also accept which test classes to run via @ARGV:

    # t/test_class_tests.t
    use File::Spec::Functions qw( catdir );
    use FindBin qw( $Bin );
    use Test::Class::Moose::Load catdir( $Bin, 't', 'lib' );
    Test::Class::Moose->new( classes => \@ARGV )->runtests;

If @ARGV is empty, Test::Class::Moose will run all classes. To give arguments while running prove, we use the arisdottle :::

    prove -lb t/test_class_tests.t :: My::Test::Class

Now we can choose which test class we want to run right on the command line.

Tags ^

Tags are a way of organizing your test methods into groups. Later you can choose to only execute the test methods from one or more tags. You can add tags like "online" for tests that require a network, or "database" for tests that require a database, and then include or exclude those tags when you execute your tests.

You add tags to your test methods using attributes. A test method may have one or more tags:

    sub test_database : Tags( database )            { ... }
    sub test_network  : Tests(7) Tags( online api ) { ... }

Then, if your database goes down, you can exclude those tests from the t/test_class_tests.t script:

    # t/test_class_tests.t
    use File::Spec::Functions qw( catdir );
    use FindBin qw( $Bin );
    use Test::Class::Moose::Load catdir( $Bin, 't', 'lib' );
    Test::Class::Moose->new(
        classes      => \@ARGV,
        exclude_tags => [qw( database )],
    )->runtests;

By adding tags to your tests, you can run only those tests that you absolutely need to, increasing your productivity.

Boilerplate ^

Here is the bare minimum you need to get started using Test::Class::Moose

Test Class

    # t/lib/TestsFor/My/Class.pm
    package TestsFor::My::Class;
    use Test::Class::Moose;

    sub test_something {
        pass "I tested something!";
    }

    1;

Test Runner

    # t/test_class_tests.t
    use File::Spec::Functions qw( catdir );
    use FindBin qw( $Bin );
    use Test::Class::Moose::Load catdir( $Bin, 't', 'lib' );
    Test::Class::Moose->new(
        classes => \@ARGV,
    )->runtests;

AUTHOR ^

Doug Bell: https://github.com/preaction

AUTHOR ^

Curtis "Ovid" Poe <ovid@cpan.org>

COPYRIGHT AND LICENSE ^

This software is copyright (c) 2014 by Curtis "Ovid" Poe.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.

syntax highlighting: