Skip to content

Latest commit

 

History

History
142 lines (95 loc) · 6.4 KB

CONTRIBUTING.md

File metadata and controls

142 lines (95 loc) · 6.4 KB
Raw

Contribution Guide

This guide is for those who want to have a better understanding of the library internals and how to contribute. The code and the tests are the best documentation you will find, so the guide only aims at introducing the higher picture of the library mechanisms.

  1. Differences between 2.x and 3.x
  2. Architecture
    1. FixtureBuilder
    2. Generator
  3. Expression Language
  4. Contributing
    1. Testing
    2. Profiling

Differences between 2.x and 3.x

Main differences between 2.x and 3.x:

  • The persistence layer has been removed
  • Rewrite of the API to make it more extensible and more robust
  • Change in the architecture to address some limitations found in 2.x

Architecture

The two entry points of the library are the DataLoader and FileLoader:

Alice's goal is to render a PHP array containing a description of objects and parameters (with support for injected external objects and parameters) into a set of objects called ObjectSet.

In DataLoader, the main components are:

  • FixtureBuilder: denormalizes the given PHP array into a comprehensible set of data called FixtureSet
  • Generator: generates the objects and resolve the parameters described in FixtureSet to generate an ObjectSet

FixtureBuilder

FixtureBuilder is composed of two components: Denormalizer and ExpressionLanguage.

Denormalizer transforms the input data array with the injected values into a collection of Fixture, which describes an object to generate with a set of parameters.

ExpressionLanguage is the component used by the Denormalizer to parse the values such as @user<current()>, i.e. the Alice DSL interpreter.

The result FixtureSet is always invariant: reloading the same set of data will always result in the same result, hence is cacheable.

Generator

The Generator transforms a FixtureSet (composed of Fixture, parameters, injected objects, and injected parameters) into an ObjectSet, which is a collection of PHP objects and parameters. Because some data are generated "randomly" thanks to Faker, a FixtureSet will always give a different ObjectSet. If there is no dynamic data, the result will always be the same. If Faker is seeded with null, the data will be generated more or less randomly. For a given seed, though, the data generated will always be the same.

The data is generated in the following order:

  1. Fixture Resolution (FixtureSet):
    1. Resolve the parameters: it is assumed the injected parameters are already resolved. Existing parameters are overriden by the local ones if they conflict.
    2. Resolve the fixtures: a fixture may have flags (for example, to support templates). This resolution step is where it is possible to alter the collection of fixtures depending on the flags. In the templates example, this is the step where a dummy fixture inheriting from a base_dummy will inherit the properties of base_dummy, and base_dummy will be removed of the list of fixtures (templates should not be generated).
  2. Instantiation: This when all fixtures will be instantiated. The instantiation attempts to account for the order. For example, if second_dummy is being instantiated but requires first_dummy to exist, the instantiator will instantiate first_dummy before resuming the instantiation of second_dummy.
  3. Hydration: This is when all the property values will be set.
  4. Calling: Additional function calls can be made on the PHP objects after instantiation and hydration.

During the instantiation, hydration or calling process, values may be passed. Those values may be a reference to another fixture, a parameter, a static value, but or a value that must be generated via faker. For example, <name()> will result in a random name. Those values are always resolved on the fly by a ValueResolver.

To see more about the fixture lifecycle, please check #388.

For more details regarding a class, the easiest way is to check the code itself and the tests.

Expression Language

As already mentioned, Alice ships with an Expression Language, which interprets values such as @user* or <current()>. The complete list of supported features can be found in ParserIntegrationTest with the original RFC.

Contributing

Testing

The project uses PHPUnit for tests. The library also includes framework bridges (like for Symfony), which amounts to registering the right services with the right properties (like tags and configuration) to the framework Dependency Injection Container. Any other framework special features should be handled in another library, bundle, module etc.

To avoid any conflicts, the framework dependencies used by the bridges are installed in dedicated folders thanks to bamarni bin composer plugin and theofidry inheritance composer plugin. As a result, if you want to run the tests for Symfony, you must run the tests with phpunit_symfony.xml.dist instead of phpunit.xml.dist.

To run the tests use make test.

The tests should be descriptive and are "testdox friendly" i.e. if you are using the testdox option, you will get something like:

Profiling

A Profiling of different scenarii is done with Blackfire. The scenario can be found under profiling. If you wish to run them, you can try to simply run make tp (for "test performances"). You may however have to touch some elements of the blackfire configuration (see the profiling/scenarioX/blackfire.php files).


« Advanced Guide • Table of Contents »