Self-contained test file

A single test file could be used for simple tests where you want to use Moo attributes for fixtures that get used by test blocks.

Here is an example that requires a "corpus" attribute, stores lines from that file in the "lines" attribute and makes it available to all test blocks:

    # examples/cookbook/single_file.t
    use Test::Roo;
    use MooX::Types::MooseLike::Base qw/ArrayRef/;
    use Path::Tiny;
    has corpus => (
        is       => 'ro',
        isa      => sub { -f shift },
        required => 1,
    );
    has lines => (
        is  => 'lazy',
        isa => ArrayRef,
    );
    sub _build_lines {
        my ($self) = @_;
        return [ map { lc } path( $self->corpus )->lines ];
    }
    test 'sorted' => sub {
        my $self = shift;
        is_deeply( $self->lines, [ sort @{$self->lines} ], "alphabetized");
    };
    test 'a to z' => sub {
        my $self = shift;
        my %letters = map { substr($_,0,1) => 1 } @{ $self->lines };
        is_deeply( [sort keys %letters], ["a" .. "z"], "all letters found" );
    };
    run_me( { corpus => "/usr/share/dict/words" } );
    # ... test other corpuses ...
    done_testing;

Standalone test class

You don't have to put the test class into the .t file. It's just a class.

Here is the same corpus checking example as before, but now as a class:

    # examples/cookbook/lib/CorpusCheck.pm
    package CorpusCheck;
    use Test::Roo;
    use MooX::Types::MooseLike::Base qw/ArrayRef/;
    use Path::Tiny;
    has corpus => (
        is       => 'ro',
        isa      => sub { -f shift },
        required => 1,
    );
    has lines => (
        is  => 'lazy',
        isa => ArrayRef,
    );
    sub _build_lines {
        my ($self) = @_;
        return [ map { lc } path( $self->corpus )->lines ];
    }
    test 'sorted' => sub {
        my $self = shift;
        is_deeply( $self->lines, [ sort @{$self->lines} ], "alphabetized");
    };
    test 'a to z' => sub {
        my $self = shift;
        my %letters = map { substr($_,0,1) => 1 } @{ $self->lines };
        is_deeply( [sort keys %letters], ["a" .. "z"], "all letters found" );
    };
    1;

Running it from a .t file doesn't even need Test::Roo:

    # examples/cookbook/standalone.t
    use strictures;
    use Test::More;
    use lib 'lib';
    use CorpusCheck;
    CorpusCheck->run_tests({ corpus => "/usr/share/dict/words" });
    done_testing;

Standalone Test Roles

The real power of Test::Roo is decomposing test behaviors into roles that can be reused.

Imagine we want to test a file-finder module like Path::Iterator::Rule. We could put tests for it into a role, then run the tests from a file that composes that role. For example, here would be the test file:

    # examples/cookbook/test-pir.pl
    use Test::Roo;
    use lib 'lib';
    with 'IteratorTest';
    run_me(
        {
            iterator_class => 'Path::Iterator::Rule',
            result_type    => '',
        }
    );
    done_testing;

Then in the distribution for Path::Class::Rule, the same role could be tested with a test file like this:

    # examples/cookbook/test-pcr.pl
    use Test::Roo;
    use lib 'lib';
    with 'IteratorTest';
    run_me(
        {
            iterator_class => 'Path::Class::Rule',
            result_type    => 'Path::Class::Entity',
        },
    );
    done_testing;

What is the common role that they are consuming? It sets up a test directory, creates files and runs tests:

    # examples/cookbook/lib/IteratorTest.pm
    package IteratorTest;
    use Test::Roo::Role;
    use MooX::Types::MooseLike::Base qw/:all/;
    use Class::Load qw/load_class/;
    use Path::Tiny;
    has [qw/iterator_class result_type/] => (
        is       => 'ro',
        isa      => Str,
        required => 1,
    );
    has test_files => (
        is      => 'ro',
        isa     => ArrayRef,
        default => sub {
            return [
                qw(
                aaaa
                bbbb
                cccc/dddd
                eeee/ffff/gggg
                )
            ];
        },
    );
    has tempdir => (
        is  => 'lazy',
        isa => InstanceOf ['Path::Tiny']
    );
    has rule_object => (
        is      => 'lazy',
        isa     => Object,
        clearer => 1,
    );
    sub _build_description { return shift->iterator_class }
    sub _build_tempdir {
        my ($self) = @_;
        my $dir = Path::Tiny->tempdir;
        $dir->child($_)->touchpath for @{ $self->test_files };
        return $dir;
    }
    sub _build_rule_object {
        my ($self) = @_;
        load_class( $self->iterator_class );
        return $self->iterator_class->new;
    }
    sub test_result_type {
        my ( $self, $file ) = @_;
        if ( my $type = $self->result_type ) {
            isa_ok( $file, $type, $file );
        }
        else {
            is( ref($file), '', "$file is string" );
        }
    }
    test 'find files' => sub {
        my $self = shift;
        $self->clear_rule_object; # make sure have a new one each time
        $self->tempdir;
        my $rule = $self->rule_object;
        my @files = $rule->file->all( $self->tempdir, { relative => 1 } );
        is_deeply( \@files, $self->test_files, "correct list of files" )
        or diag explain \@files;
        $self->test_result_type($_) for @files;
    };
    # ... more tests ...
    1;

Skipping all tests

If you need to skip all tests in the .t file because some prerequisite isn't available or some fixture couldn't be built, use a "BUILD" method and call "plan skip_all => $reason".

    use Class::Load qw/try_load_class/;
    has fixture => (
        is => 'lazy',
    );
    sub _build_fixture {
        # ... something that might die if unavailable ...
    }
    sub BUILD {
        my ($self) = @_;
        try_load_class('Class::Name')
            or plan skip_all => "Class::Name required to run these tests";
        eval { $self->fixture }
            or plan skip_all => "Couldn't build fixture";
    }

Setting a test description

You can override "_build_description" to create a test description based on other attributes. For example, the "IteratorTest" package earlier had these lines:

    has [qw/iterator_class result_type/] => (
        is       => 'ro',
        isa      => Str,
        required => 1,
    );
    sub _build_description { return shift->iterator_class }

The "iterator_class" attribute is required and then the description is set to it. Or, there could be a more verbose description:

    sub _build_description {
        my $name = shift->iterator_class;
        return "Testing the $name class"
    }

Requiring a builder

A test role can specify a lazy attribute and then require the consuming class to provide a builder for it.

In the test role:

    has fixture => (
        is => 'lazy',
    );
    requires '_build_fixture';

In the consuming class:

    sub _build_fixture { ... }

Clearing fixtures

If a fixture has a clearer method, it can be easily reset during testing. This works really well with lazy attributes which get regenerated on demand.

    has fixture => (
        is => 'lazy',
        clearer => 1,
    );
    test "some test" => sub {
        my $self = shift;
        $self->clear_fixture;
        ...
    };

Setting up a fixture before testing

When you need to do some extra work to set up a fixture, you can put a method modifier on the "setup" method. In some cases, this is more intuitive than doing all the work in an attribute builder.

Here is an example that creates an SQLite table before any tests are run and cleans up afterwards:

    # example/cookbook/sqlite.t
    use Test::Roo;
    use DBI;
    use Path::Tiny;
    has tempdir => (
        is      => 'ro',
        clearer => 1,
        default => sub { Path::Tiny->tempdir },
    );
    has dbfile => (
        is      => 'lazy',
        default => sub { shift->tempdir->child('test.sqlite3') },
    );
    has dbh => ( is => 'lazy', );
    sub _build_dbh {
        my $self = shift;
        DBI->connect(
            "dbi:SQLite:dbname=" . $self->dbfile, { RaiseError => 1 }
        );
    }
    before 'setup' => sub {
        my $self = shift;
        $self->dbh->do("CREATE TABLE f (f1, f2, f3)");
    };
    after 'teardown' => sub { shift->clear_tempdir };
    test 'first' => sub {
        my $self = shift;
        my $dbh  = $self->dbh;
        my $sth  = $dbh->prepare("INSERT INTO f(f1,f2,f3) VALUES (?,?,?)");
        ok( $sth->execute( "one", "two", "three" ), "inserted data" );
        my $got = $dbh->selectrow_arrayref("SELECT * FROM f");
        is_deeply( $got, [qw/one two three/], "read data" );
    };
    run_me;
    done_testing;

Running tests during setup and teardown

You can run any tests you like during setup or teardown. The previous example could have written the setup and teardown hooks like this:

    before 'setup' => sub {
        my $self = shift;
        ok( ! -f $self->dbfile, "test database file not created" );
        ok( $self->dbh->do("CREATE TABLE f (f1, f2, f3)"), "created table");
        ok( -f $self->dbfile, "test database file exists" );
    };
    after 'teardown' => sub {
        my $self = shift;
        my $dir = $self->tempdir;
        $self->clear_tempdir;
        ok( ! -f $dir, "tempdir cleaned up");
    };

Global modifiers with "each_test"

Modifying "each_test" triggers methods before or after every test block defined with the "test" function. Because this affects all tests, whether from the test class or composed from roles, it needs to be used thoughtfully.

Here is an example that ensures that every test block is run in its own separate temporary directory.

    # examples/cookbook/with_tempd.t
    use Test::Roo;
    use File::pushd qw/tempd/;
    use Cwd qw/getcwd/;
    has tempdir => (
        is => 'lazy',
        isa => sub { shift->isa('File::pushd') },
        clearer => 1,
    );
    # tempd changes directory until the object is destroyed
    # and the fixture caches the object until cleared
    sub _build_tempdir { return tempd() }
    # building attribute will change to temp directory
    before each_test => sub { shift->tempdir };
    # clearing attribute will change to original directory
    after each_test => sub { shift->clear_tempdir };
    # do stuff in a temp directory
    test 'first test' => sub {
        my $self = shift;
        is( $self->tempdir, getcwd(), "cwd is " . $self->tempdir );
        # ... more tests ...
    };
    # do stuff in a separate, fresh temp directory
    test 'second test' => sub {
        my $self = shift;
        is( $self->tempdir, getcwd(), "cwd is " . $self->tempdir );
        # ... more tests ...
    };
    run_me;
    done_testing;

Individual test modifiers

If you want to have method modifiers on an individual test, put your Test::More tests in a method, add modifiers to that method, and use "test" to invoke it.

    # examples/cookbook/hookable_test.t
    use Test::Roo;
    has counter => ( is => 'rw', default => sub { 0 } );
    sub is_positive {
        my $self = shift;
        ok( $self->counter > 0, "counter is positive" );
    }
    before is_positive => sub { shift->counter( 1 ) };
    test 'hookable' => sub { shift->is_positive };
    run_me;
    done_testing;

Wrapping tests

As a middle ground between global and individual modifiers, if you need to call some code repeatedly for some, but not all all tests, you can create a custom test function. This might make sense for only a few tests, but could be helpful if there are many that need similar behavior, but you can't make it global by modifying "each_test".

The following example clears the fixture before tests defined with the "fresh_test" function.

    # examples/cookbook/wrapped.t
    use strict;
    use Test::Roo;
    has fixture => (
        is => 'rw',
        lazy => 1,
        builder => 1,
        clearer => 1,
    );
    sub _build_fixture { "Hello World" }
    sub fresh_test {
        my ($name, $code) = @_;
        test $name, sub {
            my $self = shift;
            $self->clear_fixture;
            $code->($self);
        };
    }
    fresh_test 'first' => sub {
        my $self = shift;
        is ( $self->fixture, 'Hello World', "fixture has default" );
        $self->fixture("Goodbye World");
    };
    fresh_test 'second' => sub {
        my $self = shift;
        is ( $self->fixture, 'Hello World', "fixture has default" );
    };
    run_me;
    done_testing;