PHPixie ORM library
$slice = new \PHPixie\Slice();
$database = new \PHPixie\Database($slice->arrayData(array(
'default' => array(
'driver' => 'pdo',
'connection' => 'sqlite::memory:'
)
)));
$orm = new \PHPixie\ORM($database, $slice->arrayData(array(
// We will add some configuration later
)));
If you are using the PHPixie Framework the ORM component is already set up for you. It is accessable via
$frameworkBuilder->components()->orm()
and can be configured in theconfig/orm.php
file of your bundle.
A Model consists of a Repository, Queries and Entities and will usually map to a table in a relational database or a document collection in MongoDB.
- Entity - a single item, stored in the database, e.g. an article
- Repository - is used to create and save Entities
- Query - is used to select, update or delete Entities
You can start using the models right away without any configuration, which will use defaults based on the name of the model.
$repository = $orm->repository('article');
$newArticle = $repository->createEntity();
$articleQuery = $repository->query();
// shorthands
$newArticle = $orm->createEntity('article');
$articleQuery = $orm->query('article');
By default ORM assumes that the table name is the plural of the name of the model, and that the name of the primary key is 'id'. For MongoDB database the default id field '_id' is assumed. You can override these settings for a particular model in your configuration file:
return array(
'models' => array(
'article' => array(
'type' => 'database',
'connection' => 'default',
'id' => 'id'
),
// you can also define embedded models
// if you are using MongoDB,
// more on that later
)
);
// Saving an entity
$article->title = 'Welcome';
$article->save();
// Getting a field with a default value
$article->getField('title', 'No Title');
// Getting a required field
// Will throw an Exception if it is not set
$article->getRequiredField('title');
// Convert to a simple PHP object
// Usefull for serializing
$object = $article->asObject();
// Deleting
$article->delete();
ORM queries share a lot of syntax with the Database queries, here are a few query examples:
// Find article by name
$article = $orm->query('article')
->where('title', 'Welcome')
->findOne();
// Find by id
$article = $orm->query('article')
->in($id)
->findOne();
// Query by multiple ids
$articles = $orm->query('article')
->in($ids)
->findOne();
// Actually the in() method can be used
// to also include subqueries and entities.
//
// This will select articles with id '1',
// or the id same as $someArticle, and
// anything matched by $subQuery
$articles = $orm->query('article')
->in(array(
1, $someArticle, $subQuery
));
// Multiple conditions
$articles = $orm->query('article')
->where('viewsTotal', '>', 2)
->or('viewsDone', '<', 5)
->find();
// Ordering
// Supports multiple fields
$articles
->orderAscendingBy('name')
->orderDescendingBy('id');
// Limit and offset
$articles
->limit(1)
->offset(2);
// It continues in the same way
// as in the Database component
$articles = $orm->query('article')
->where('name', 'Welcome')
->or(function($query) {
$querty
->where('viewsTotal', '>', 2)
->or('viewsDone', '<', 5);
})
->find();
// Alternative syntax for
// nested conditions
$articles = $orm->query('article')
->where('name', 'Welcome')
->startWhereConditionGroup('or')
->where('viewsTotal', '>', 2)
->or('viewsDone', '<', 5)
->endGroup()
->find();
If findOne
is used, a query will return either a single item or null
.
When using find
a Loader is returned which can be used as follows:
// Iterate over it
// Note: this can be done only once
foreach($articles as $article) {
// ...
}
// Convert into an array
// to allow multiple iterations
$articles = $articles->asArray();
// Convert it into plain objects,
// useful for serializing
$data = $articles->asArray(true);
// Convert it into associative array
$data = $articles->asArray(true, 'id');
If you are going to access a particular relationship for multiple items that you are selecting you should preload their relationships to avoid multiple datbase queries. This is called eager loading.
// To load relationships eagerly
// just pass their names to find() or findOne()
$articles = $query->find(array(
'author',
'tags'
));
// You can also preload nested relationships
// using the dot notation
$articles = $query->find(array(
'author.friends',
));
Queries can also be used for updating and deleting items:
// Count matched items
$count = $articleQuery->count();
// Delete matched items
$articleQuery->delete();
// Update a field in all matched items
$articleQuery
->update(array(
'status' => 'published'
));
// Some more advanced updating
$articleQuery
->getUpdateBuilder()
->increment('views', 1)
->set('status', 'published')
// Removing a field in MongoDB
->remove('isDraft')
->execute();
Extending ORM Models usually forces the developer into coupling business logic to the database, which makes them hard to test and and debug. In these cases performing any kind of testing requires inserting some dummy test data into the database. PHPixie ORM solves this problem by allowing Decorator-like behavior with Wrappers. Instead of extending a class you provide wrappers that might be used to wrap ORM Entities, Queries and Repositories and add functionality to them.
class UserEntity extends \PHPixie\ORM\Wrappers\Type\Database\Entity
{
// Get users full name
public function fullName()
{
// You can access the actual entity
// Using $this->entity;
return $this->entity->firstName.' '.$this->entity->lastName;
}
}
class UserQuery extends \PHPixie\ORM\Wrappers\Type\Database\Query
{
// Extending queries is useful
// For adding bulk conditions
public function popular()
{
// Access query with $this->query
$this->query
->where('viewsPerDay', '>', 5000)
->orWhere('friendCount' '>=', 100);
}
}
// You will rarely need to extend repositories
class UserRepository extends \PHPixie\ORM\Wrappers\Type\Database\Repository
{
// Overriding a save method
// can be used for validation
public function save($entity)
{
if($entity->getField('name') === null) {
throw new \Exception("You must provide a user name");
}
$this->repository->save($entity);
}
}
Now we have to register these classes with the ORM.
class ORMWrappers extends \PHPixie\ORM\Wrappers\Implementation
{
// Model names of database entities to wrap
protected $databaseEntities = array(
'user'
);
// Model names of queries to wrap
protected $databaseQueries = array(
'user'
);
// Model names of repositories to wrap
protected $databaseRepositories = array(
'user'
);
// Model names of embedded entities to wrap
// We cover them later in this manual
protected $embeddedEntities = array(
'post'
);
// Provide methods to build the wrappers
public function userEntity($entity)
{
return new UserEntity($entity);
}
public function userQuery($query)
{
return new UserQuery($query);
}
public function userRepository($repository)
{
return new UserRepository($repository);
}
public function postEntity($entity)
{
return new PostEntity($entity);
}
}
Pass an instance of this class when creating the ORM instance:
$wrappers = new ORMWrappers();
$orm = new \PHPixie\ORM($database, $slice->arrayData(array(
// Configuration options
)), $wrappers);
When using the PHPixie Framework, you already have the
ORMWrappers
class already registered and present in your bundle.
PHPixie supports one-to-many, one-to-one, many-to-many relationships,
as well as embeds-one and embeds-many for embedded models. Each relationship
defines a set of properties that are added to entities and queries
and can be used to access related data. You can configure them using the
relationships
key in the configuration file:
return array(
'models' => array(
// ...
),
'relationships' => array(
array(
'type' => 'manyToMany',
'left' => 'article',
'right' => 'tag'
)
)
);
It is possible to define relationships between tables in same database, different databases or even between relational databases and MongoDB
Before looking at actual relationship types, let's see how can you define conditions on a relationship:
// Lets say we have a one to many relationship
// between categories and articles
// Finding all categories with
// at lest one article
//
// Note that you use the property name,
// and not the model name here.
// You can modify property names
// in your config file, more on them later
$categoryQuery->relatedTo('articles');
// or all articles with a category
$articleQuery->relatedTo('category');
// Use logic operators
// like with where()
$articleQuery->orNotRelatedTo('category');
// Find categories related to
// particular articles
$categoryQuery->relatedTo('articles', $articles);
// $articles can be an id of an article,
// an article entity or an article query,
// or an array of them, e.g.
$categoryQuery->relatedTo('articles', $articleQuery);
$categoryQuery->relatedTo('articles', $someArticle);
$categoryQuery->relatedTo('articles', 4); //id
// Will find all categories related
// to any of the defined articles
$categoryQuery->relatedTo('articles', array(
4, 3, $articleQuery, $articleEntity
));
// Relationship conditions
// Find categories related to articles
// that have a title 'Welcome'
$categoryQuery->relatedTo('articles', function($query) {
$query
->where('title', 'Welcome');
});
// Or a shorthand
// You'll be using this a lot
$categoryQuery->where('articles.title', 'Welcome');
// You can use the '.'
// to go deeper in the relationships
// Find categories that have at least one article
// written by the author 'Dracony'
$categoryQuery->where('articles.author.name', 'Dracony');
// Or combine the verbose approach
// with the shorthand one
$categoryQuery->relatedTo('articles.author', function($query) {
$query
->where('name', 'Dracony');
});
This is the most common relationship. A category can own many articles, a topic has many replies, etc.
// Configuration
return array(
'models' => array(
// ...
),
'relationships' => array(
//...
array(
// mandatory options
'type' => 'oneToMany',
'owner' => 'category',
'items' => 'article',
// The following keys are optional
// and will fallback to the defaults
// based on model names
'ownerOptions' => array(
// the name of the property added to the owner
// e.g. $category->articles();
//
// defaults to the plural of the item model
'itemsProperty' => 'articles'
),
'itemsOptions' => array(
// the name of the property added to items
// e.g. $article->category()
//
// defaults to the owner model
'ownerProperty' => 'category',
// the field that is used to link items
// defaults to '<property>Id'
'ownerKey' => 'categoryId',
// The default behavior is to set
// the field defined in ownerKey to null
// when the owner gets deleted
//
// changed it to 'delete' to delete
// the articles when their category is deleted
'onOwnerDelete' => 'update'
)
)
)
);
Now that we have relationship properties defined we may start using them:
// Using with entities
// Getting articles category
// The property names used are
// the ones defined in the config
$category = $article->category();
// Get category articles
$articles = $category->articles();
// Add article to category
$category->articles->add($article);
// Remove article from category
$category->articles->remove($article);
// Remove categories from all articles
$category->articles->removeAll();
// Or you can do the same
// from the article side
$article->category->set($category);
$article->category->remove();
Note how using properties instead of conventional methods like
addArticle
,removeAllArticles
tidies up your entities, and doesn't result in a heap of methods added to single class when there are multiple relationships defined
// You can use queries, ids and arrays
// anywhere you can use an entity.
// This allows performing bulk operations faster
// Assign first 5 articles
// to a category
$articleQuery
->limit(5)
->offset(0);
$category->articles->add($articleQuery);
Queries also have properties, just like Entities do. This allows you to perform more operations with less calls to the database.
// Assign articles to a category
$articlesQuery->category->set($category);
// Assign articles to a category with a single
// database call, without the need to select
// rows from database
$categoryQuery->where('title', 'PHP');
$articlesQuery->category->set($category);
// Unset categories for all articles
$orm->query('aricle')
->category->remove();
// Unset only some ctaegories
$categoryQuery->where('title', 'PHP');
$orm->query('aricle')
->category->remove($category);
Using queries instead of iterating over entities provides a huge boost to your performance
// You can also use query properties
// for query building
$categoryQuery = $articleQuery
->where('title', 'Welcome')
->categories();
// is the same as
$categoryQuery = $orm->query('category')
->relatedTo('articles.title', 'Welcome');
// or
$articleQuery->where('title', 'Welcome');
$categoryQuery = $orm->query('category')
->relatedTo('articles', $articleQuery);
It may seem like too many options at this point, but you can just stick to whichever syntax you prefer best. The generated queries stay exactly the same.
One to One relationships are very similar to One To Many. The main difference is that as soon as you attach a new item to an owner, the previous item gets detached. A good example of this would be a relationship between an auction lot and the highest bidder. As soon as we set a new person as the highest bidder the old one is unset. Another example wold be tasks and workers. Where each task can be performed by a single worker.
// Configuration
return array(
'models' => array(
// ...
),
'relationships' => array(
//...
array(
// mandatory options
'type' => 'oneToOne',
'owner' => 'worker',
'item' => 'task',
// The following keys are optional
// and will fallback to the defaults
// based on model names
'ownerOptions' => array(
// the name of the property added to the owner
// e.g. $worker->task();
//
// Unlike manyToMany this uses
// a singular case by default
'itemProperty' => 'task'
),
// note it's 'itemOptions' here
// but 'itemsOptions' for One To Many
'itemOptions' => array(
// the name of the property added to items
// e.g. $task->worker()
'ownerProperty' => 'worker',
// the field that is used to link items
// defaults to '<property>Id'
'ownerKey' => 'workerId',
// The default behavior is to set
// the field defined in ownerKey to null
// when the owner gets deleted
//
// changed it to 'delete' to delete
// the task when its worker is deleted
'onOwnerDelete' => 'update'
)
)
)
);
// Using with entities
// Getting task worker
$worker = $task->worker();
// Getting worker tasks
$worker = $task->worker();
// assign worker to a task
$task->worker->set($worker);
$worker->task->set($task);
// Unset task
$task->worker->remove();
$work->task->remove();
The interface in oneToOne relationships is the same on both sides and is identical to the owner property in oneToMany relationships
The most common many-to-many relationship is between articles and tags. These relationships require a special pivot table (or a MongoDB collection) to store the links between items.
// Configuration
return array(
'models' => array(
// ...
),
'relationships' => array(
//...
array(
// mandatory options
'type' => 'manyToMany',
'left' => 'article',
'right' => 'tag',
// The following keys are optional
// and will fallback to the defaults
// based on model names
'leftOptions' => array(
'property' => 'tags'
),
'rightOptions' => array(
'property' => 'articles'
),
// depends on property names
// defaults to <rightProperty><leftProperty>
'pivot' => 'articlesTags',
'pivotOptions' => array(
// defaults to the connection
// of the left model
'connection' => 'default',
// columns in pivot table
// default to '<property>Id'
'leftKey' => 'articleId',
'rightKey' => 'tagId',
)
)
)
);
Using a many-to-many relationship is simiar to using the owner side of a one-to-many one.
// Add
$article->tags->add($tag);
// Remove a particular tag
$article->tags->remove($tag);
// Remove all tags from article
$article->tags->removeAll();
// Remove all tags from multiple articles
$orm->query('article')
->where('status', 'published')
->tags->removeAll();
// Construct a tag query from article query
$tagQuery = $orm->query('article')
->where('status', 'published')
->tags();
// Everything else can be used
// in the same way as categories
// in the one-to-many examples
Using queries for bulk operations here makes it possible to assign and remove relationships with a single database call
// Link multiple articles
// to multiple tags in one go
$articleQuery->tags->add($tagQuery);
A lot of work went into optimizing these query operations, and at the moment no other PHP ORM supports editing relationships between queries. Instead of requiring m*n queries to edit many-to-many relationships, the query approach can achieve it in one go.
MongoDB supports nested documents, which allows using embedded models. E.g. the author of an article:
{
"title" : "Welcome",
//...
"author" : {
"name" : "Dracony",
//...
}
}
Embedded models consist only of Entities and have neither Repositories nor Queries. Subdocuments and subarrays are not automatically registered as embedded relationships, you have to do it inside your configuration:
return array(
'models' => array(
// Some database models
'article' => array(),
'topic' => array(),
// Configuring an embedded model
'author' => array(
'type' => 'embedded'
),
),
'relationships' => array(
array(
'type' => 'embedsOne',
'owner' => 'article',
'item' => 'author'
),
array(
'type' => 'embedsOne',
'owner' => 'topic',
'item' => 'author'
),
)
);
There are also some usage differences between embedded and database models:
// Embedded entities cannot be saved on their own.
// Instead you just save the database entity
$article->author()->name = 'Dracony';
$article->save();
// Getting the parent model
// Conditions are specified as usual
$articleQuery
->where('author.name', 'Dracony');
// To specify a subdocument condition
// use a '>' separator.
//
// This will require the article author
// to have a stats.totalPost field 'Dracony'
$articleQuery
->where('author>stats.totalPost', 'Dracony');
This is a relationship with a subdocument like the above article author example.
// Configuration
return array(
'models' => array(
// ...
),
'relationships' => array(
//...
array(
// mandatory options
'type' => 'embedsOne',
'owner' => 'article',
'item' => 'author',
// The following keys are optional
// and will fallback to the defaults
// based on model names
'ownerOptions' => array(
// the name of the property added to the owner
// e.g. $article->author();
//
// Defaults to item model name
'property' => 'task',
),
'itemOptions' => array(
// Dot separated path
// to the document within owner
//
// Defaults to owner property name
'path' => 'author'
// You can use nested paths
// e.g. 'authors.editor'
)
)
)
);
Note how we don't define the property for accessing the article from the author. This is because accessing the owner is always done using
$entity->owner()
for embedded entities.
// get author from article
$author = $article->author();
// get author owner
$article = $author->owner();
// remove author
$article->author->remove();
// Set an author
$article->author->set($author);
// Check if an author is set
$article->author->exists();
// Create and set new author
$author = $article->author->create();
// Create with data
$author = $article->author->create($data);
Defines a relationship with an embedded array of subdocuments. E.g. forum topic replies:
{
"title" : "Welcome",
"replies" : [
{
"message" : "Hello",
"author" : "Dracony"
},
{
"message" : "World",
"author" : "Dracony"
}
]
}
// Configuration
return array(
'models' => array(
// ...
),
'relationships' => array(
//...
array(
// mandatory options
'type' => 'embedsMany',
'owner' => 'topic',
'items' => 'reply',
// The following keys are optional
// and will fallback to the defaults
// based on model names
'ownerOptions' => array(
// the name of the property added to the owner
// e.g. $topic->replies();
//
// Defaults to the plural
// of the item model name
'property' => 'replies',
),
'itemOptions' => array(
// Dot separated path
// to the document within owner
//
// Defaults to owner property name
'path' => 'replies'
// You can use nested paths
// e.g. 'content.replies'
)
)
)
);
// Get replies iterator
$replies = $topic->replies();
// Get reply count
$topic->replies->count();
// Get reply by offset
$reply = $topic->replies->get(2);
// Create and add a reply
$reply = $topic->replies->create();
// Create reply from data
$reply = $topic->replies->create($data);
//Create reply at offset
$reply = $topic->replies->create($data, 2);
// Add reply
$topic->replies->add($reply);
// Add reply by offset
$topic->replies->add($reply, 2);
// Remove reply
$topic->replies->remove($reply);
// Remove multiple replies
$topic->replies->remove($replies);
// Remove reply by offset
$topic->replies->offsetUnset(2);
// Remove all replies
$topic->replies->removeAll();
// Check if a reply exists
$exists = $topic->replies->offsetExists(2);
// array access
$reply = $topic->replies[1];
$topic->replies[2] = $reply;
$exists = isset($topic->replies[2]);
unset($topic->replies[2]);
Nested Set is an efficient approach to storing trees in SQL databases. A good example would be storing category and comment trees. PHPixie is the only PHP framework to optimize it even further by namespacing subtrees which results in much faster inserts and updates to the tree. The code behind it is more complex than the usual Nested Set implementation, but is really simple from the user perspective.
// Configuration
return array(
'models' => array(
// ...
),
'relationships' => array(
//...
array(
// mandatory options
'type' => 'nestedSet',
'model' => 'category',
// Nested Set requires additional
// fields to be present in your table.
// All of them are INTEGER
// These are their default names
'leftKey' => 'left',
'rightKey' => 'right',
'depthKey' => 'depth',
'rootIdKey' => 'rootId',
// You can also customize the
// relationship property names
'parentProperty' => 'parent',
'childrenProperty' => 'children',
// Defaults to "all<plural of parentProperty>"
'allParentsProperty' => 'allParents'
// Defaults to "all<childrenProperty>"
'allChildrenProperty' => 'allChildren'
)
)
);
This relationship defines four relationship properties instead of the usual two.
The children
property refers to immediate children, while allChildren
represents
all children of a node. In the same way parent
is the immediate parent and
allParents
relates to all parents of the node.
The basic usage is straightforward an similar to the one-to-many relationship.
// Move child to parent
$category->children->add($subcategory);
// or
$subcategory->parent->set($category);
// Remove child from parent
$subcategory->parent->remove();
// Remove all children from node
$category->children->removeAll();
// Find all root nodes and preload
// their children recursively.
$categories = $orm->query('category')
->notRelatedTo('parent') // root nodes have no parents
->find(array('children'));
Some more advanced usage:
// Get a query representing
// all children recursively
$allChildrenQuery = $category->children->allQuery();
$allChidlren = $allChildrenQuery->find();
// Same with getting all parents parents
$allParents = $category->parent->allQuery()->find();
// Find a category with name 'Trees' that is
// a descendant of 'Plants'
$query
->where('name', 'Trees')
->relatedTo('allParents', function($query) {
$query->where('name', 'Plants');
})
->find();
// or like this
$query
->where('name', 'Plants')
->allChildren()
->where('name', 'Trees')
->find();
Special caution has to made when deleting nodes. PHPixie will only allow you to delete items if they either don't have any children or their children are also being deleted. For example:
$plants->children->add($trees);
$plants->children->add($flowers);
// An exception will be thrown
$plants->delete();
// move children away from the node
$plants->children->removeAll();
//now it's safe to delete
$plants->delete();
// or delete all three,
$query->in(array($plants, $trees, $flowers))->delete();
The important part is to remember that
parent
andchildren
refer only to the immediate parent an children, whileallParents
andallChildren
refer to all related nodes recursively.