Download the PHP package michaldudek/knit without Composer

Knit

Knit ties your PHP objects with your database of choice in a simple way.

Knit is not a full-blown ORM/ODM, but rather an extended "data mapper" that abstracts away a persistent store behind a "repository" pattern.

Features overview:

• Repository pattern (with methods: find, findOne, add, update, remove, save as well as magic findByPropertyName and findOneByPropertyName).
• Building queries in form of "criteria" arrays.
• Swappable data stores.
• Various events dispatched (will read, will write, will update, will remove, did read, did update, did remove).
• Easy programmatic joins (which means joins between stores).
• Swappable data mappers.

Check KnitBundle for Symfony2 integration.

Quick Demo:

<?php

// gets a repository to manage object of class MyApp\News
$repository = $knit->getRepository(MyApp\News::class, 'news');

// gets top 20 news from the last week that have a cover photo set
// and were in one of the three country sections
$news = $repository->find([
    'cover_photo:not' => null,
    'published_at:gt' => strtotime('1 week ago'),
    'country' => ['US','UK','PL']
], [
    'orderBy' => 'views',
    'orderDir' => Knit\Knit::ORDER_DESC,
    'limit' => 20
]);

// joins photos to the news items based on `news.id = photo.news_id`
// into `$photos` property on news object (by calling `::setPhotos()` method)
$news = $repository->joinMany(
    $news,
    $knit->getRepository(MyApp\Photo::class, 'photos'),
    'id',
    'news_id',
    'photos'
);

foreach ($news as $i => $item) {
    $featured = $i < 5 && count($item->getPhotos()) > 3;
    $item->setFeatured($featured);

    $repository->save($item);
}

Criteria expressions

As you can see in the quick demo above, Knit repositories allow simple yet quite powerful filtering through criteria arrays. You don't need to worry about building database queries and if you ever want to switch the underlying database you can do so without changing any of your existing code.

The pattern to build criteria is as follows:

'[property_name]:[operator]' => [value_to_match]

You can put multiple criteria in an array and by default all will need to match.

For more complex queries you can use logical operators like this:

[
    Knit\Knit::LOGIC_OR => [
        'active' => 1,
        Knit\Knit::LOGIC_AND => [
            'active' => 0,
            'registered_at:gt' => strtotime('1 day ago')
        ]
    ]
]

This will translate to an SQL WHERE statement WHERE active = 1 OR (active = 0 AND registered_at > 1234567890) or MongoDB query {"$or":[{"active":1},{"$and":[{"active":0},{"registered_at":{"$gt":1234567890}}]}]}.

Available operators

The following operators are available. Please note that there may be some small inconsistencies in how they work between various stores.

• :eq (default, can be ommitted) - equals (for array values it will switch to :in operator - but only in DoctrineDBAL Store)
• :not - not equals (for array values it will switch to :not_in operator - but only in DoctrineDBAL Store)
• :gt - greater than
• :gte - greater than or equal
• :lt - lower than
• :lte - lower than or equal
• :regex - matches a regular expression
• :like - allows % to escape one or more characters
• :not_like - allows % to escape one or more characters (only in DoctrineDBAL Store)
• :exists - property exists (only in MongoDB)

Usage

Knit is available through Packagist:

$ composer require michaldudek/knit:0.2.*

The main core class of Knit is Knit\Repository and you will interact with it the most. Knit\Knit class serves as a convenient factory for repositories.

Knit consists of three main building blocks:

• Repository - which is the facade for working with objects. Each object class has its own repository instance and all objects are managed through it.
• Store - "driver" for a persistent (or not necesserily) store like MySQL or MongoDB.
• DataMapper - responsible for mapping arrays to objects (hydration) and vice versa (extraction).

In order to work, Knit requires default store and default data mapper, as well as an event dispatcher. For example:

<?php

// this will be used as the default store for all repositories
$store = new Knit\Store\DoctrineDBAL\Store([
    'driver' => 'pdo_mysql',
    'user' => '[username]',
    'password' => '[password]',
    'host' => '[database_host]',
    'port' => [database_port],
    'dbname' => '[database_name]'
]);

$knit = new Knit\Knit(
    $store,
    // this will be used as the default data mapper for all repositories
    new Knit\DataMapper\ArraySerializable\ArraySerializer(),
    // this will be injected to all repositories
    new Symfony\Component\EventDispatcher\EventDispatcher()
);

To get a repository for an object you can now use

$repository = $knit->getRepository(MyApp\News::class, 'news');

where first argument is the class name of objects that will be managed by the repository and the second argument is collection name within the underlying store in which the objects are stored (tables in MySQL). Additional arguments are optional, but you can pass a custom store for the created repository as 3rd argument, a custom data mapper as 4th argument and even a custom repository class as 5th argument.

By default, all created repositories will be instances of Knit\Repository, but you can your custom class if you create it right next to object you want to manage and add a *Repository suffix. The lookup is done this way:

// for given class:
$objectClass = MyApp\MyObject::class;

// do the lookup
if (class_exists($objectClass .'Repository')) {
    $repositoryClass = $objectClass .'Repository';
}

Or you can pass your custom class name as 5th argument to ::getRepository() as mentioned above.

Note that all repositories must extend the core Knit\Repository class.

Available Repository Methods:

• ::find(array $criteria = [], array $params = []): array
• ::findByIdentifiers(array $identifiers, array $params = []): array
• ::findOne(array $criteria = [], array $params = []): object|null
• ::findOneByIdentifier($identifier): object|null
• ::findBy*($value, array $params = []): array
• ::findOneBy*($value, array $params = []): object|null
• ::count(array $criteria = [], array $params = []): integer
• ::save($object) - proxies to ::add($object) or ::update($object)
• ::add($object)
• ::update($object)
• ::delete($object)
• ::joinOne(array $objects, Repository $withRepository, $leftProperty, $rightProperty, $targetProperty, array $criteria = [], $excludeEmpty = false): array
• ::public function joinMany(array $objects, Repository $withRepository, $leftProperty, $rightProperty, $targetProperty, array $criteria = [], array $params = [], $excludeEmpty = false): array
• ::createWithData(array $data = []): object
• ::updateWithData($object, array $data = [])

Events

You can listen on various events dispatched by a repository:

• knit.will_read with Knit\Events\CriteriaEvent allows to alter criteria before read
• knit.did_read with Knit\Events\ResultsEvent allows to alter results after read
• knit.will_save with Knit\Events\ObjectEvent allows to alter object before save
• knit.did_save with Knit\Events\ObjectEvent allows to alter object after save
• knit.will_add with Knit\Events\ObjectEvent allows to alter object only before insert
• knit.did_add with Knit\Events\ObjectEvent allows to alter object only after insert
• knit.will_update with Knit\Events\ObjectEvent allows to alter object only before update
• knit.did_update with Knit\Events\ObjectEvent allows to alter object only after update
• knit.will_delete with Knit\Events\ObjectEvent allows to prevent delete
• knit.did_delete with Knit\Events\ObjectEvent

You can get a reference to the event dispatcher either on repository: $repository->getEventDispatcher() or on Knit instance $knit->getEventDispatcher().

Note that the event dispatcher is global for all repositories so your listeners should always check what type of data is included in the event.

Available Stores

At the moment the following stores are available:

DoctrineDBALStore

Knit\Store\DoctrineDBAL\Store

This store is based on Doctrine DBAL and therefore supports all the databases Doctrine supports, namely MySQL or PgSQL and more.

It accepts the same configuration options as DoctrineDBAL Connection class as the first argument. See Configuration for details.

You can access Doctrine Query Builder in a repository that uses this class by calling

$queryBuilder = $this->store->getConnection()->createQueryBuilder();

but please remember to hydrate all the results later by calling

$results = $this->mapResults($results);

MongoDBStore

Knit\Store\MongoDb\Store

Store that uses MongoDB PHP extension to communicate with MongoDB.

First constructor argument is a configuration array and it must contain at least hostname and database keys. It also accepts the following options:

• username
• password
• replica_set
• port
• hosts - an array of additional hosts to connect to (elements must be arrays with key hostname and optionally port)

Available Data Mappers

At the moment there is only one simple data mapper available: Knit\DataMapper\ArraySerializable\ArraySerializer.

It simply hydrates objects by calling ::fromArray(array $data) and ::toArray(): array methods on them and therefore requires all objects to implement Knit\DataMapper\ArraySerializable\ArraySerializableInterface.

Contributing

PR's and issues are most welcome.

Before submitting a Pull Request please run

$ make qa

in order to run all unit and lint tests.

Please make sure your changes are fully tested and covered.

If you need to test against a real database (like the currently imlemented stores are), there is a Vagrant VM config in resources/tests. It will create a box with PHP 5.6, MySQL and MongoDB and some env vars which guide the tests that they can be ran (see ::setUp() methods in DoctrineDBALTest and [MongoDbTest] (tests/Store/MongoDbTest)). Feel free to customize / extend this setup.

$ cd resources/tests
$ vagrant up
$ vagrant ssh
$ cd /knit
$ make qa

The above will run the full tests suite.

To generate a code coverage report run

$ make report

It will save the HTML report in resources/coverage directory.

License

MIT, see LICENSE.md.

Copyright (c) 2015 Michał Pałys-Dudek