You are browsing a version that is no longer maintained. |
Working with Associations
Associations between entities are represented just like in regular object-oriented PHP code using references to other objects or collections of objects.
Changes to associations in your code are not synchronized to the
database directly, only when calling EntityManager#flush()
.
There are other concepts you should know about when working with associations in Doctrine:
- If an entity is removed from a collection, the association is removed, not the entity itself. A collection of entities always only represents the association to the containing entities, not the entity itself.
- When a bidirectional assocation is updated, Doctrine only checks on one of both sides for these changes. This is called the owning side of the association.
- A property with a reference to many entities has to be instances of the
Doctrine\Common\Collections\Collection
interface.
Association Example Entities
We will use a simple comment system with Users and Comments as entities to show examples of association management. See the PHP docblocks of each association in the following example for information about its type and if it's the owning or inverse side.
1 <?php
/** @Entity */
class User
{
/** @Id @GeneratedValue @Column(type="string") */
private $id;
/**
* Bidirectional - Many users have Many favorite comments (OWNING SIDE)
*
* @ManyToMany(targetEntity="Comment", inversedBy="userFavorites")
* @JoinTable(name="user_favorite_comments")
*/
private $favorites;
/**
* Unidirectional - Many users have marked many comments as read
*
* @ManyToMany(targetEntity="Comment")
* @JoinTable(name="user_read_comments")
*/
private $commentsRead;
/**
* Bidirectional - One-To-Many (INVERSE SIDE)
*
* @OneToMany(targetEntity="Comment", mappedBy="author")
*/
private $commentsAuthored;
/**
* Unidirectional - Many-To-One
*
* @ManyToOne(targetEntity="Comment")
*/
private $firstComment;
}
/** @Entity */
class Comment
{
/** @Id @GeneratedValue @Column(type="string") */
private $id;
/**
* Bidirectional - Many comments are favorited by many users (INVERSE SIDE)
*
* @ManyToMany(targetEntity="User", mappedBy="favorites")
*/
private $userFavorites;
/**
* Bidirectional - Many Comments are authored by one user (OWNING SIDE)
*
* @ManyToOne(targetEntity="User", inversedBy="commentsAuthored")
*/
private $author;
}
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
This two entities generate the following MySQL Schema (Foreign Key definitions omitted):
1 CREATE TABLE User (
id VARCHAR(255) NOT NULL,
firstComment_id VARCHAR(255) DEFAULT NULL,
PRIMARY KEY(id)
) ENGINE = InnoDB;
CREATE TABLE Comment (
id VARCHAR(255) NOT NULL,
author_id VARCHAR(255) DEFAULT NULL,
PRIMARY KEY(id)
) ENGINE = InnoDB;
CREATE TABLE user_favorite_comments (
user_id VARCHAR(255) NOT NULL,
favorite_comment_id VARCHAR(255) NOT NULL,
PRIMARY KEY(user_id, favorite_comment_id)
) ENGINE = InnoDB;
CREATE TABLE user_read_comments (
user_id VARCHAR(255) NOT NULL,
comment_id VARCHAR(255) NOT NULL,
PRIMARY KEY(user_id, comment_id)
) ENGINE = InnoDB;
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Establishing Associations
Establishing an association between two entities is
straight-forward. Here are some examples for the unidirectional
relations of the User
:
The interaction code would then look like in the following snippet
($em
here is an instance of the EntityManager):
1 <?php
$user = $em->find('User', $userId);
// unidirectional many to many
$comment = $em->find('Comment', $readCommentId);
$user->getReadComments()->add($comment);
$em->flush();
// unidirectional many to one
$myFirstComment = new Comment();
$user->setFirstComment($myFirstComment);
$em->persist($myFirstComment);
$em->flush();
2
3
4
5
6
7
8
9
10
11
12
13
14
15
In the case of bi-directional associations you have to update the fields on both sides:
1 <?php
class User
{
// ..
public function getAuthoredComments() {
return $this->commentsAuthored;
}
public function getFavoriteComments() {
return $this->favorites;
}
}
class Comment
{
// ...
public function getUserFavorites() {
return $this->userFavorites;
}
public function setAuthor(User $author = null) {
$this->author = $author;
}
}
// Many-to-Many
$user->getFavorites()->add($favoriteComment);
$favoriteComment->getUserFavorites()->add($user);
$em->flush();
// Many-To-One / One-To-Many Bidirectional
$newComment = new Comment();
$user->getAuthoredComments()->add($newComment);
$newComment->setAuthor($user);
$em->persist($newComment);
$em->flush();
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Notice how always both sides of the bidirectional association are updated. The previous unidirectional associations were simpler to handle.
Removing Associations
Removing an association between two entities is similarly straight-forward. There are two strategies to do so, by key and by element. Here are some examples:
1 <?php
// Remove by Elements
$user->getComments()->removeElement($comment);
$comment->setAuthor(null);
$user->getFavorites()->removeElement($comment);
$comment->getUserFavorites()->removeElement($user);
// Remove by Key
$user->getComments()->remove($ithComment);
$comment->setAuthor(null);
2
3
4
5
6
7
8
9
10
11
You need to call $em->flush()
to make persist these changes in
the database permanently.
Notice how both sides of the bidirectional association are always
updated. Unidirectional associations are consequently simpler to
handle. Also note that if you use type-hinting in your methods, i.e.
setAddress(Address $address)
, PHP will only allow null
values if null
is set as default value. Otherwise
setAddress(null) will fail for removing the association. If you
insist on type-hinting a typical way to deal with this is to
provide a special method, like removeAddress()
. This can also
provide better encapsulation as it hides the internal meaning of
not having an address.
When working with collections, keep in mind that a Collection is
essentially an ordered map (just like a PHP array). That is why the
remove
operation accepts an index/key. removeElement
is a
separate method that has O(n) complexity using array_search
,
where n is the size of the map.
Since Doctrine always only looks at the owning side of a bidirectional association for updates, it is not necessary for write operations that an inverse collection of a bidirectional one-to-many or many-to-many association is updated. This knowledge can often be used to improve performance by avoiding the loading of the inverse collection. |
You can also clear the contents of a whole collection using the
Collections::clear()
method. You should be aware that using
this method can lead to a straight and optimized database delete or
update call during the flush operation that is not aware of
entities that have been re-added to the collection.
Say you clear a collection of tags by calling
$post->getTags()->clear();
and then call
$post->getTags()->add($tag)
. This will not recognize the tag having
already been added previously and will consequently issue two separate database
calls.
Association Management Methods
It is generally a good idea to encapsulate proper association management inside the entity classes. This makes it easier to use the class correctly and can encapsulate details about how the association is maintained.
The following code shows updates to the previous User and Comment example that encapsulate much of the association management code:
1 <?php
class User
{
//...
public function markCommentRead(Comment $comment) {
// Collections implement ArrayAccess
$this->commentsRead[] = $comment;
}
public function addComment(Comment $comment) {
if (count($this->commentsAuthored) == 0) {
$this->setFirstComment($comment);
}
$this->comments[] = $comment;
$comment->setAuthor($this);
}
private function setFirstComment(Comment $c) {
$this->firstComment = $c;
}
public function addFavorite(Comment $comment) {
$this->favorites->add($comment);
$comment->addUserFavorite($this);
}
public function removeFavorite(Comment $comment) {
$this->favorites->removeElement($comment);
$comment->removeUserFavorite($this);
}
}
class Comment
{
// ..
public function addUserFavorite(User $user) {
$this->userFavorites[] = $user;
}
public function removeUserFavorite(User $user) {
$this->userFavorites->removeElement($user);
}
}
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
You will notice that addUserFavorite
and removeUserFavorite
do not call addFavorite
and removeFavorite
, thus the
bidirectional association is strictly-speaking still incomplete.
However if you would naively add the addFavorite
in
addUserFavorite
, you end up with an infinite loop, so more work
is needed. As you can see, proper bidirectional association
management in plain OOP is a non-trivial task and encapsulating all
the details inside the classes can be challenging.
If you want to make sure that your collections are perfectly
encapsulated you should not return them from a
|
This will however always initialize the collection, with all the performance penalties given the size. In some scenarios of large collections it might even be a good idea to completely hide the read access behind methods on the EntityRepository.
There is no single, best way for association management. It greatly depends on the requirements of your concrete domain model as well as your preferences.
Synchronizing Bidirectional Collections
In the case of Many-To-Many associations you as the developer have the
responsibility of keeping the collections on the owning and inverse side
in sync when you apply changes to them. Doctrine can only
guarantee a consistent state for the hydration, not for your client
code.
Using the User-Comment entities from above, a very simple example can show the possible caveats you can encounter:
There are two approaches to handle this problem in your code:
- Ignore updating the inverse side of bidirectional collections, BUT never read from them in requests that changed their state. In the next Request Doctrine hydrates the consistent collection state again.
- Always keep the bidirectional collections in sync through association management methods. Reads of the Collections directly after changes are consistent then.
Transitive persistence / Cascade Operations
Persisting, removing, detaching, refreshing and merging individual entities can become pretty cumbersome, especially when a highly interweaved object graph is involved. Therefore Doctrine 2 provides a mechanism for transitive persistence through cascading of these operations. Each association to another entity or a collection of entities can be configured to automatically cascade certain operations. By default, no operations are cascaded.
The following cascade options exist:
- persist : Cascades persist operations to the associated entities.
- remove : Cascades remove operations to the associated entities.
- merge : Cascades merge operations to the associated entities.
- detach : Cascades detach operations to the associated entities.
- refresh : Cascades refresh operations to the associated entities.
- all : Cascades persist, remove, merge, refresh and detach operations to associated entities.
Cascade operations are performed in memory. That means collections and related entities are fetched into memory, even if they are still marked as lazy when the cascade operation is about to be performed. However this approach allows entity lifecycle events to be performed for each of these operations. However, pulling objects graph into memory on cascade can cause considerable performance overhead, especially when cascading collections are large. Makes sure to weigh the benefits and downsides of each cascade operation that you define. To rely on the database level cascade operations for the delete operation instead, you can configure each join column with the onDelete option. See the respective mapping driver chapters for more information. |
The following example is an extension to the User-Comment example of this chapter. Suppose in our application a user is created whenever he writes his first comment. In this case we would use the following code:
Even if you persist a new User that contains our new Comment this
code would fail if you removed the call to
EntityManager#persist($myFirstComment)
. Doctrine 2 does not
cascade the persist operation to all nested entities that are new
as well.
More complicated is the deletion of all of a user's comments when he is removed from the system:
Without the loop over all the authored comments Doctrine would use an UPDATE statement only to set the foreign key to NULL and only the User would be deleted from the database during the flush()-Operation.
To have Doctrine handle both cases automatically we can change the
User#commentsAuthored
property to cascade both the persist
and the remove operation.
Even though automatic cascading is convenient it should be used with care. Do not blindly apply cascade=all to all associations as it will unnecessarily degrade the performance of your application. For each cascade operation that gets activated Doctrine also applies that operation to the association, be it single or collection valued.
Persistence by Reachability: Cascade Persist
There are additional semantics that apply to the Cascade Persist operation. During each flush() operation Doctrine detects if there are new entities in any collection and three possible cases can happen:
- New entities in a collection marked as cascade persist will be directly persisted by Doctrine.
- New entities in a collection not marked as cascade persist will produce an Exception and rollback the flush() operation.
- Collections without new entities are skipped.
This concept is called Persistence by Reachability: New entities that are found on already managed entities are automatically persisted as long as the association is defined as cascade persist.
Orphan Removal
There is another concept of cascading that is relevant only when removing entities
from collections. If an Entity of type A
contains references to privately
owned Entities B
then if the reference from A
to B
is removed the
entity B
should also be removed, because it is not used anymore.
OrphanRemoval works with one-to-one, one-to-many and many-to-many associations.
When using the |
As a better example consider an Addressbook application where you have Contacts, Addresses and StandingData:
1 <?php
namespace Addressbook;
use Doctrine\Common\Collections\ArrayCollection;
/**
* @Entity
*/
class Contact
{
/** @Id @Column(type="integer") @GeneratedValue */
private $id;
/** @OneToOne(targetEntity="StandingData", orphanRemoval=true) */
private $standingData;
/** @OneToMany(targetEntity="Address", mappedBy="contact", orphanRemoval=true) */
private $addresses;
public function __construct()
{
$this->addresses = new ArrayCollection();
}
public function newStandingData(StandingData $sd)
{
$this->standingData = $sd;
}
public function removeAddress($pos)
{
unset($this->addresses[$pos]);
}
}
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Now two examples of what happens when you remove the references:
In this case you have not only changed the Contact
entity itself but
you have also removed the references for standing data and as well as one
address reference. When flush is called not only are the references removed
but both the old standing data and the one address entity are also deleted
from the database.
Filtering Collections
Collections have a filtering API that allows to slice parts of data from a collection. If the collection has not been loaded from the database yet, the filtering API can work on the SQL level to make optimized access to large collections.
1 <?php
use Doctrine\Common\Collections\Criteria;
$group = $entityManager->find('Group', $groupId);
$userCollection = $group->getUsers();
$criteria = Criteria::create()
->where(Criteria::expr()->eq("birthday", "1982-02-17"))
->orderBy(array("username" => Criteria::ASC))
->setFirstResult(0)
->setMaxResults(20)
;
$birthdayUsers = $userCollection->matching($criteria);
2
3
4
5
6
7
8
9
10
11
12
13
14
15
You can move the access of slices of collections into dedicated methods of
an entity. For example |
The Criteria has a limited matching language that works both on the SQL and on the PHP collection level. This means you can use collection matching interchangeably, independent of in-memory or sql-backed collections.
1 <?php
use Doctrine\Common\Collections;
class Criteria
{
/**
* @return Criteria
*/
static public function create();
/**
* @param Expression $where
* @return Criteria
*/
public function where(Expression $where);
/**
* @param Expression $where
* @return Criteria
*/
public function andWhere(Expression $where);
/**
* @param Expression $where
* @return Criteria
*/
public function orWhere(Expression $where);
/**
* @param array $orderings
* @return Criteria
*/
public function orderBy(array $orderings);
/**
* @param int $firstResult
* @return Criteria
*/
public function setFirstResult($firstResult);
/**
* @param int $maxResults
* @return Criteria
*/
public function setMaxResults($maxResults);
public function getOrderings();
public function getWhereExpression();
public function getFirstResult();
public function getMaxResults();
}
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
You can build expressions through the ExpressionBuilder. It has the following methods:
andX($arg1, $arg2, ...)
orX($arg1, $arg2, ...)
eq($field, $value)
gt($field, $value)
lt($field, $value)
lte($field, $value)
gte($field, $value)
neq($field, $value)
isNull($field)
in($field, array $values)
notIn($field, array $values)
contains($field, $value)
There is a limitation on the compatibility of Criteria comparisons. You have to use scalar values only as the value in a comparison or the behaviour between different backends is not the same. |