NAME Test::DBIx::Class - Easier test cases for your DBIx::Class applications SYNOPSIS The following is example usage for this module. Assume you create a standard Perl testing script, such as "MyApp/t/schema/01-basic.t" which is run from the shell like "prove -l t/schema/01-basic.t" or during "make test". That test script could contain: use Test::More; { use strict; use warnings; use Test::DBIx::Class { schema_class => 'MyApp::Schema', connect_info => ['dbi:SQLite:dbname=:memory:','',''], fixture_class => '::Populate', }, 'Person', 'Person::Employee' => {-as => 'Employee'}, 'Job', 'Phone'; ## Your testing code below ## ## Your testing code above ## done_testing(); } Yes, it looks like a lot of boilerplate, but sensible defaults are in place (the above code example shows most of the existing defaults) and configuration data can be loaded from a central file. So your 'real life' example is going to look closer to (assuming you put all your test configuration in the standard place, "t/etc/schema.conf": use Test::More; { use strict; use warnings; use Test::DBIx::Class qw(:resultsets); ## Your testing code below ## ## Your testing code above ## done_testing(); } Then, assuming the existance of a DBIx::Class::Schema subclass called, "MyApp::Schema" and some DBIx::Class::ResultSources named like "Person", "Person::Employee", "Job" and "Phone", will automatically deploy a testing schema in the given database / storage (or auto deploy to an in-memory based DBD::SQLite database), install fixtures and let you run some test cases, such as: ## Your testing code below ## fixtures_ok 'basic' => 'installed the basic fixtures from configuration files'; fixtures_ok { Job => [ [qw/name description/], [Programmer => 'She who writes the code'], ['Movie Star' => 'Knows nothing about the code'], ], }, 'Installed some custom fixtures via the Populate fixture class', ok my $john = Person->find({email=>'jjnapiork@cpan.org'}) => 'John has entered the building!'; is_fields $john, { name => 'John Napiorkowski', email => 'jjnapiork@cpan.org', age => 40, }, 'John has the expected fields'; is_fields ['job_title'], $john->jobs, [ {job_title => 'programmer'}, {job_title => 'administrator'}, ], is_fields 'job_title', $john->jobs, [qw/programmer administrator/], 'Same test as above, just different compare format; is_fields [qw/job_title salary/], $john->jobs, [ ['programmer', 100000], ['administrator, 120000], ], 'Got expected fields from $john->jobs'; is_fields [qw/name age/], $john, ['John Napiorkowski', 40], => 'John has expected name and age'; is_fields_multi 'name', [ $john, ['John Napiorkowski'], $vanessa, ['Vanessa Li'], $vincent, ['Vincent Zhou'], ] => 'All names as expected'; is_fields 'fullname', ResultSet('Country')->find('USA'), 'United States of America', 'Found the USA'; is_deeply [sort Schema->sources], [qw/ Person Person::Employee Job Country Phone /], 'Found all expected sources in the schema'; fixtures_ok my $first_album = sub { my $schema = shift @_; my $cd_rs = $schema->resultset('CD'); return $cd_rs->create({ name => 'My First Album', track_rs => [ {position=>1, title=>'the first song'}, {position=>2, title=>'yet another song'}, ], cd_artist_rs=> [ {person_artist=>{person => $vanessa}}, {person_artist=>{person => $john}}, ], }); }, 'You can even use a code reference for custom fixtures'; ## Your testing code above ## Please see the test cases for more examples. DESCRIPTION The goal of this distribution is to make it easier to write test cases for your DBIx::Class based applications. It does this in three ways. First, it trys to make it easy to deploy your Schema. This can be to your dedicated testing database, or a simple SQLite database. This allows you to run tests without interfering with your development work and having to stop and set up a testing database instance. Second, we allow you to load test fixtures via several different tools. Last we create some helper functions in your test script so that you can reduce repeated or boilerplate code. Overall, we attempt to reduce the amount of code you have to write before you can begin writing tests. IMPORTED METHODS The following methods are automatically imported when you use this module. Schema You probably won't need this directly in your tests unless you have some application logic methods in it. ResultSet ($source, ?{%search}, ?{%conditions}) Although you can import your sources as local keywords, sometimes you might need to get a particular resultset when you don't wish to import it globally. Use like ok ResultSet('Job'), "Yeah, some jobs in the database"; ok ResultSet( Job => {hourly_pay=>{'>'=>100}}), "Good paying jobs available!"; Since this returns a normal DBIx::Class::ResultSet, you can just call the normal methods against it. ok ResultSet('Job')->search({hourly_pay=>{'>'=>100}}), "Good paying jobs available!"; This is the same as the test above. fixtures_ok This is used to install and verify installation of fixtures, either inlined, from a fixture set in a file, or through a custom sub reference. Accept three argument styles: coderef Given a code reference, execute it against the currently defined schema. This is used when you need a lot of control over installing your fixtures. Example: fixtures_ok sub { my $schema = shift @_; my $cd_rs = $schema->resultset('CD'); return $cd_rs->create({ name => 'My First Album', track_rs => [ {position=>1, title=>'the first song'}, {position=>2, title=>'yet another song'}, ], cd_artist_rs=> [ {person_artist=>{person => $vanessa}}, {person_artist=>{person => $john}}, ], }); }, 'Installed fixtures'; The above gets executed at runtime and if there is an error it is trapped, reported and we move on to the next test. hashref Given a hash reference, attempt to process it via the default fixtures loader or through the specified loader. fixtures_ok { Person => [ ['name', 'age', 'email'], ['John', 40, 'john@nowehere.com'], ['Vincent', 15, 'vincent@home.com'], ['Vanessa', 35, 'vanessa@school.com'], ], }, 'Installed fixtures'; This is a good option to use while you are building up your fixture sets or when your sets are going to be small and not reused across lots of tests. This will get you rolling without messing around with configuration files. fixture set name Given a fixture name, or array reference of names, install the fixtures. fixtures_ok 'core'; fixtures_ok [qw/core extra/]; Fixtures are installed in the order specified. All different types can be mixed and matched in a given test file. is_result ($result, ?$result) Quick test to make sure $result does inherit from DBIx::Class or that it inherits from a subclass of DBIx::Class. is_resultset ($resultset, ?$resultset) Quick test to make sure $resultset does inherit from DBIx::Class::ResultSet or from a subclass of DBIx::Class::ResultSet. eq_resultset ($resultset, $resultset, ?$message) Given two ResultSets, determine if the are equal based on class type and data. This is a true set equality that ignores sorting order of items inside the set. eq_result ($resultset, $resultset, ?$message) Given two row objects, make sure they are the same. hri_dump ($resultset) Not a test, just returns a version of the ResultSet that has its inflator set to DBIx::Class::ResultClass::HashRefInflator, which returns a set of hashes and makes it easier to stop issues. This return value is suitable for dumping via Data::Dump, for example. reset_schema Wipes and reloads the schema. dump_settings Returns the configuration and related settings used to initialize this testing module. This is mostly to help you debug trouble with configuration and to help the authors find and fix bugs. At some point this won't be exported by default so don't use it for your real tests, just to help you understand what is going on. You've been warned! is_fields A 'Swiss Army Knife' method to check your results or resultsets. Tests the values of a Result or ResultSet against expected via a pattern. A pattern is automatically created by instrospecting the fields of your ResultSet or Result. Example usage for testing a result follows. ok my $john = Person->find('john'); is_fields 'name', $john, ['John Napiorkowski'], 'Found name of $john'; is_fields [qw/name age/], $john, ['John Napiorkowski', 40], 'Found $johns name and age'; is_fields $john, { name => 'John Napiorkowski', age => 40, email => 'john@home.com'}; # Assuming $john has only the three columns listed In the case were we need to infer the match pattern, we get the columns of the given result but remove the primary key. Please note the following would also work: is_fields [qw/name age/] $john, { name => 'John Napiorkowski', age => 40}, 'Still got the name and age correct'; You should choose the method that makes most sense in your tests. Example usage for testing a resultset follows. is_fields 'name', Person, [ 'John', 'Vanessa', 'Vincent', ]; is_fields ['name'], Person, [ 'John', 'Vanessa', 'Vincent', ]; is_fields ['name','age'], Person, [ ['John',40], ['Vincent',15], ['Vanessa',35], ]; is_fields ['name','age'], Person, [ {name=>'John', age=>40}, {name=>'Vanessa',age=>35}, {name=>'Vincent', age=>15}, ]; I find the array version is most consise. Please note that the match is not ordered. If you need to test that a given Resultset is in a particular order, you will currently need to write a custom test. If you have a big need for this I'd be willing to write a test for it, or gladly accept a patch to add it. You should examine the test cases for more examples. is_fields_multi TBD: Not yet written. SETUP AND INITIALIZATION The generic usage for this would look like one of the following: use Test::DBIx::Class \%options, @sources use Test::DBIx::Class %options, @sources Where %options are key value pairs and @sources an array as specified below. Initialization Options The only difference between the hash and hash reference version of %options is that the hash version requires its keys to be prepended with "-". If you are inlining a lot of configuration the hash reference version may look neater, while if you are only setting one or two options the hash version might be more readable. For example, the following are the same: use Test::DBIx::Class -config_path=>[qw(t etc config)], 'Person', 'Job'; use Test::DBIx::Class {config_path=>[qw(t etc config)]}, 'Person', 'Job'; The following options are currently standard and always available. Depending on your storage engine (such as SQLite or mysql) you will have other options. config_path These are the relative paths searched for configuration file information. See "Initialization Sources" for more. In the case were we have both inlined and file based configurations, the inlined is merged last (that is, has highest authority to override configuration files. When the final merging of all configurations (both anything inlined at 'use' time, and anything found in any of the specified config_paths, we do a single 'post' config_path check. This allows you to add in a configuration file from inside a configuration file. For safety and sanity you can only do this once. This feature makes it easier to globalize any additional configuration files. For example, I often store user specific settings in "~/etc/conf.*". This feature allows me to add that into my standard "t/etc/schema.*" so it's available to all my test cases. schema_class Required. This must be your subclass of DBIx::Class::Schema that defines your database schema. connect_info Required. This will accept anything you can send to "connect" in DBIx::Class. Defaults to: ['dbi:SQLite:dbname=:memory:','',''] if left blank (but see 'traits' below for more) fixture_path These are a list of relative paths search for fixtures. Each item should be a directory that contains files loadable by Config::Any and suitable to be installed via one of the fixture classes. fixture_class Command class that installs data into the database. Must provide a method called 'install_fixtures' that accepts a perl data structure and installs it into the database. Must capture and report errors. Default value is "::Populate", which loads Test::DBIx::Class::FixtureClass::Populate, which is a command class based on "populate" in DBIx::Class::Schema. resultsets Lets you add in some result source definitions to be imported at test script runtime. See "Initialization Sources" for more. force_drop_table When deploying the database this option allows you add a 'drop table' statement before the create ddl. Since this will return an error if you attempt to drop a table that doesn't exist, this is off by default for SQLite storage engines. You may need to enble it you you are using the following 'keep_db' option. keep_db By default your testing database is 'cleaned up' after you are finished. This drops all the created tables (but currently doesn't delete any related files or database users, if any). If you want to keep your testing database after all the tests are run, you can set this to true. If so, you may also need to set the previously mentioned option 'force_drop_table' to true as well, or we will attempt to create tables and populate them when they are already populated and created. traits Traits are Moose::Roles that are applied to the class managing the connection to your database. If you leave this option blank and you don't specify anything for 'connect_info' (above), we automatically load the SQLite trait (which can be reviewed at Test::DBIx::Class::SchemaManager::Trait::SQLite). This trait installs the ability to automatically discover and deploy to an in memory or a filesystem SQLite database. If you are just getting started with testing, this is probably your easiest option. Currently there are only two traits, the SQLite trait just described (and since it get's automatically loaded you never need to load it yourself) and the Test::DBIx::Class::SchemaManager::Trait::Testmysqld trait, which is built on top of Test::mysqld and allows you the ability to deploy to and run tests against a temporary instance of Mysql. For this trait Mysql and DBD::mysql needs to be installed, but Mysql does not need to be running, nor do you need to create a test database or user. See "TRAITS" for more. Please note that although all initialization options can be set inlined or in a configuration file, some options can also be set via %ENV variables. %ENV settings will only apply IF there are no existing values for the option in any configuration file. As of this time we don't merge %ENV settings, they only provider overrides to the default settings. Example use (assumes you are using the default SQLite database) DBPATH=test.db KEEP_DB=1 prove -lv t/schema/check-person.t After running the test there will be a new file called 'test.db' in the home directory of your distribution. You can use: sqlite3 test.db to open and view the tables and their data as loaded by any fixtures or create statements. See Test::DBIx::Class::SchemaManager::Trait::SQLite for more. Note that you can specify both 'dbpath' and 'keep_db' in your configuration files if you prefer. I tried to expose a subset of configuration to %ENV that I thought the most useful. Patches and suggestions welcomed. Initialization Sources The @sources are a list of result sources that you want helper methods injected into your test script namespace. This is the 'Source' part of: $schema->resultset('Source'); Injecting methods are optional since you can also use the 'ResultSet' keyword Imported Source keywords use Sub::Exporter so you have quite a few options for controling how the keywords are imported. For example: use Test::DBIx::Class 'Person', 'Person::Employee' => {-as => 'Employee'}, 'Person' => {search => {age=>{'>'=>55}}, -as => 'OlderPerson'}; This would import three local keywork methods, "Person", "Employee" and "OlderPerson". For "OlderPerson", the search parameter would automatically be resolved via $resultset->search and the correct resultset returned. You may wish to preconfigure all your test result set cases in one go at the top of your test script as a way to promote reusability. In addition to the 'search' parameter, there is also an 'exec' parameter which let's you process your resultset programatically. For example: 'Person' => {exec => sub { shift->older_than(55) }, -as => 'OlderPerson'}; This code reference gets passed the resultset object. So you can use any method on $resultset. For example: 'Person' => {exec => sub { shift->find('john') }, -as => 'John'}; is_result John; is John->name, 'John Napiorkowski', "Got Correct Name"; Although since fixtures will not yet be installed, the above is probably not going to be a normally working example :) Additionally, since you can also initialize sources via the 'resultsets' configuration option, which can be placed into your global configuration files this means you can predefine and result resultsets across all your tests. Here is an example 't/etc/schema.pl' file where I initialize pretty much everything in one file: { 'schema_class' => 'Test::DBIx::Class::Example::Schema', 'resultsets' => [ 'Person', 'Job', 'Person' => { '-as' => 'NotTeenager', search => {age=>{'>'=>18}}}, ], 'fixture_sets' => { 'basic' => { 'Person' => [ [ 'name', 'age', 'email' ], [ 'John', '40', 'john@nowehere.com' ], [ 'Vincent', '15', 'vincent@home.com' ], [ 'Vanessa', '35', 'vanessa@school.com' ] ] } }, }; In this case you can simple do "use Test::DBIx::Class" and everything will happen automatically. CONFIGURATION BY FILE By default, we try to load configuration fileis from the following locations: ./t/etc/schema.* ./t/etc/[test file path].* Where "." is the root of the distribution and "*" is any of the configuration file types supported by Config::Any configuration loader. This allows you to store configuration in the format of your choice. "[test file path]" is the relative path part under the "t" directory of the calling test script. For example, if your test script is "t/mytest.t" we add the path "./t/etc/mytest.*" to the path. Additionally, we do a a merge using Hash::Merge of all the matching found configurations. This allows you to do 'cascading' configuration from the most global to the most local settings. You can override this search path with the "-config_path" key in options. For example, the following searches for "t/etc/myconfig.*" (or whatever is the correct directory separator for your operating system): use Test::DBIx::Class -config_path => [qw/t etc myconfig/]; Relative paths are rooted to the distribution home directory (ie, the one that contains your 'lib' and 't' directories). Full paths are searched without modification. You can specify multiply paths. The following would search for both "schema.*" and "share/schema". use Test::DBIx::Class -config_path => [[qw/share schema/], [qw/schema/]]; Lastly, you can use the special symbol "+" to indicate that your custom path adds to or prepends to the default search path. Since as indicated we merge all the configurations found, this means it's easy to create user level configuration settings mixed with global settings, as in: use Test::DBIx::Class -config_path => [ [qw(/ etc myapp test-schema)], '+', [qw(~ etc test-schema)], ]; Which would search and combine "/etc/myapp/test-schema.*", "./t/etc/schema.*", "./etc/[test script name].*" and "~/etc/test-schema.*". This would let you set up server level global settings, distribution level settings and finally user level settings. Please note that in all the examples given, paths are written as an array reference of path parts, rather than as a string with delimiters (i.e. we do [qw(t etc)] rather than "t/etc"). This is not required but recommended. All arguments, either string or array references, are passed to Path::Class so that we can maintain better compatibility with non unix filesystems. If you are writing for CPAN, please consider our non Unix filesystem friends :) Lastly, there is an %ENV variable named '' which, if it exists, can be used to further customize your configuration path. If we find that $ENV{TEST_DBIC_CONFIG_SUFFIX} is set, we attempt to find configuration files with the suffix appended to each of the items in the config_path option. So, if you have: use Test::DBIx::Class -config_path => [ [qw(/ etc myapp test-schema)], '+', [qw(~ etc test-schema)], ]; and $ENV{TEST_DBIC_CONFIG_SUFFIX} = '-mysql' we will check the following paths for valid and loading configuration files (assuming unix filesystem conventions) /etc/myapp/test-schema.* /etc/myapp/test-schema-mysql.* ./t/etc/schema.* ./t/etc/schema-mysql.* ./etc/[test script name].* ./etc/[test script name]-mysql.* ~/etc/test-schema.* ~/etc/test-schema-mysql.* Each path is testing in turn and all found configurations are merged from top to bottom. This feature is intended to make it easier to switch between sets of configuration files when developing. For example, you can create a test suite intended for a mysql database, but allow a failback to the default Sqlite should certain enviroment variables not exist. CONFIGURATION SUBSTITUTIONS Similarly to Catalyst::Plugin::ConfigLoader, there are some macro style keyword inflators available for use within your configuration files. This allows you to set the value of a configuration setting from an external source, such as from %ENV. There are currently two macro substitutions: ENV Given a value in %ENV, substitute the keyword for the value of the named substitution. For example, if you had: email = 'vanessa__ENV(TEST_DBIC_LAST_NAME)__@school.com' in your configuration filem your could: TEST_DBIC_LAST_NAME=_lee prove -lv t/schema-your-test.t and then: is $vanessa->email, 'vanessa_lee@school.com', 'Got expected email'; You might find this useful for configuring localized username and passwords although personally I'd rather set that via configuration in the user home directory. TRAITS As described, a trait is a Moose::Role that is applied to the class managing your database and test instance. Currently we only have the default 'SQLite' trait and the 'Testmysqld' trait, but we eventually intend to have traits to add easy support for creating Postgresql databases and supporting testing on replicated systems. Traits are installed by the 'traits' configuration option, which expects an ArrayRef as its input (however will also normalize a scalar to an ArrayRef). Available traits are as follows. SQLite This is the default trait which will be loaded if no other traits are installed and there is not 'connect_info' in the configuration. In this case we assume you want us to go and create a tempory SQLite database for testing. Please see Test::DBIx::Class::SchemaManager::Trait::SQLite for more. Testmysqld If Mysql is installed on the testing machine, and DBD::mysql, we try to auto create an instance of Mysql and deploy our tests to that. Similarly to the way the SQLite trait works, we attempt to create the database without requiring any other using effort or setup. See Test::DBIx::Class::SchemaManager::Trait::Testmysqld for more. Testpostgresql If Postgresql is install on the testing machine, along with DBD::Pg, we try to auto create an instance of Postgresql in a testing area and deploy our tests and fixtures to it. See Test::DBIx::Class::SchemaManager::Trait::Testpostgresql for more. SEE ALSO The following modules or resources may be of interest. DBIx::Class, DBIx::Class::Schema::PopulateMore, DBIx::Class::Fixtures AUTHOR John Napiorkowski "" COPYRIGHT & LICENSE Copyright 2009, John Napiorkowski "" This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.