Skip to content

Support embedded models usage for complex ActiveRecord like MongoDB or ElasticSearch

License

Notifications You must be signed in to change notification settings

utrain-tm/embedded

 
 

Repository files navigation

Embedded (Nested) Models Extension for Yii 2


This extension provides support for embedded (nested) models usage in Yii2. In particular it allows working with sub-documents in MongoDB and ElasticSearch as well as processing complex JSON attributes at relational databases.

For license information check the LICENSE-file.

Latest Stable Version Total Downloads Build Status

Installation

The preferred way to install this extension is through composer.

Either run

php composer.phar require --prefer-dist yii2tech/embedded

or add

"yii2tech/embedded": "*"

to the require section of your composer.json.

Usage

This extension grants the ability to work with complex model attributes, represented as arrays, as nested models, represented as objects. To use this feature the target class should implement [[\yii2tech\embedded\ContainerInterface]] interface. This can be easily achieved using [[\yii2tech\embedded\ContainerTrait]].

For each embedded entity a mapping declaration should be provided. In order to do so you need to declare method, which name is prefixed with 'embedded', which should return the [[Mapping]] instance. You may use [[hasEmbedded()]] and [[hasEmbeddedList()]] for this.

Per each of source field or property a new virtual property will be declared, which name will be composed by removing 'embedded' prefix from the declaration method name.

Note: watch for the naming collisions: if you have a source property named 'profile' the mapping declaration for it should have different name, like 'profileModel'.

Example:

use yii\base\Model;
use yii2tech\embedded\ContainerInterface;
use yii2tech\embedded\ContainerTrait;

class User extends Model implements ContainerInterface
{
    use ContainerTrait;

    public $profileData = [];
    public $commentsData = [];

    public function embedProfile()
    {
        return $this->mapEmbedded('profileData', Profile::className());
    }

    public function embedComments()
    {
        return $this->mapEmbeddedList('commentsData', Comment::className());
    }
}

$user = new User();
$user->profile->firstName = 'John';
$user->profile->lastName = 'Doe';

$comment = new Comment();
$user->comments[] = $comment;

Each embedded mapping may have additional options specified. Please refer to [[\yii2tech\embedded\Mapping]] for more details.

Processing embedded objects

Embedded feature is similar to regular ActiveRecord relation feature. Their declaration and processing are similar and have similar specifics and limitations. All embedded objects are lazy loaded. This means they will not be created until first demand. This saves memory but may produce unexpected results at some point. By default, once embedded object is instantiated its source attribute will be unset in order to save memory usage. You can control this behavior via [[\yii2tech\embedded\Mapping::$unsetSource]].

Embedded objects allow simplification of nested data processing, but usually they know nothing about their source data meaning and global processing. For example: nested object is not aware if its source data comes from database and it does not know how this data should be saved. Such functionality usually is handled by container object. Thus at some point you will need to convert data from embedded objects back to its raw format, which allows its native processing like saving. This can be done using method refreshFromEmbedded():

use yii\base\Model;
use yii2tech\embedded\ContainerInterface;
use yii2tech\embedded\ContainerTrait;

class User extends Model implements ContainerInterface
{
    use ContainerTrait;

    public $profileData = [
        'firstName' => 'Unknown',
        'lastName' => 'Unknown',
    ];

    public function embedProfile()
    {
        return $this->mapEmbedded('profileData', Profile::className());
    }
}

$user = new User();
var_dump($user->profileData); // outputs array: ['firstName' => 'Unknown', 'lastName' => 'Unknown']

$user->profile->firstName = 'John';
$user->profile->lastName = 'Doe';

var_dump($user->profileData); // outputs empty array

$user->refreshFromEmbedded();
var_dump($user->profileData); // outputs array: ['firstName' => 'John', 'lastName' => 'Doe']

While embedding list of objects (using [[\yii2tech\embedded\ContainerTrait::mapEmbeddedList()]]) the produced virtual field will be not an array, but an object, which satisfies [[\ArrayAccess]] interface. Thus all manipulations with such property (even if it may look like using array) will affect container object. For example:

use yii\base\Model;
use yii2tech\embedded\ContainerInterface;
use yii2tech\embedded\ContainerTrait;

class User extends Model implements ContainerInterface
{
    use ContainerTrait;

    public $commentsData = [];

    public function embedComments()
    {
        return $this->mapEmbeddedList('commentsData', Comment::className());
    }
}

$user = new User();
// ...

$comments = $user->comments; // not a copy of array - copy of object reference!
foreach ($comments as $key => $comment) {
    if (...) {
        unset($comments[$key]); // unsets `$user->comments[$key]`!
    }
}

$comments = clone $user->comments; // creates a copy of list, but not a copy of contained objects!
$comments[0]->title = 'new value'; // actually sets `$user->comments[0]->title`!

Validating embedded models

Each embedded model should declare its own validation rules and, in general, should be validated separately. However, you may simplify complex model validation using [[\yii2tech\embedded\Validator]]. For example:

use yii\base\Model;
use yii2tech\embedded\ContainerInterface;
use yii2tech\embedded\ContainerTrait;

class User extends Model implements ContainerInterface
{
    use ContainerTrait;

    public $contactData;

    public function embedContact()
    {
        return $this->mapEmbedded('contactData', Contact::className());
    }

    public function rules()
    {
        return [
            ['contact', 'yii2tech\embedded\Validator'],
            // ...
        ]
    }
}

class Contact extends Model
{
    public $email;

    public function rules()
    {
        return [
            ['email', 'required'],
            ['email', 'email'],
        ]
    }
}

$user = new User();
if ($user->load(Yii::$app->request->post()) && $user->contact->load(Yii::$app->request->post())) {
    if ($user->validate()) { // automatically validates 'contact' as well
        // ...
    }
}

Note: pay attention that [[\yii2tech\embedded\Validator]] must be set for the embedded model name - not for its source attribute. Do not mix them up!

You can enable [[\yii2tech\embedded\Validator::$initializedOnly]], allowing to skip validation for the embedded model, if it has not been initialized, e.g. requested at least once. This will save the performance in case source model can be used in different scenarios, some of which may not require embedded model manipulations. However, in this case embedded source attribute value will not be validated. You should ensure it validated in other way or it is 'unsafe' for population via [[\yii\base\Model::load()]] method.

Saving embedded models

Keep in mind that embedded models are stored separately from the source model attributes. You will need to use [[\yii2tech\embedded\ContainerInterface::refreshFromEmbedded()]] method in order to populate source model attributes with the data from embedded models.

Also note, that attempt to get 'dirty' value for embedded source attribute will also fail until you use refreshFromEmbedded() even, if embedded model has changed:

$user = User::findOne(1); // declares embedded model 'contactModel' from attribute 'contactData'

if ($user->contactModel->load(Yii::$app->request->post())) {
    var_dump($user->isAttributeChanged('contactData')); // outputs `false`

    $user->refreshFromEmbedded();
    var_dump($user->isAttributeChanged('contactData')); // outputs `true`
}

In case you are applying 'embedded' functionality to an ActiveRecord class, the best place for the data synchronization is [[\yii\db\BaseActiveRecord::beforeSave()]] method. For example, application of this extension to the [[\yii\mongodb\ActiveRecord]] class may look like following:

use yii2tech\embedded\ContainerInterface;
use yii2tech\embedded\ContainerTrait;

class ActiveRecord extends \yii\mongodb\ActiveRecord implements ContainerInterface
{
    use ContainerTrait;

    public function beforeSave($insert)
    {
        if (!parent::beforeSave($insert)) {
            return false;
        }
        $this->refreshFromEmbedded(); // populate this model attributes from embedded models' ones, ensuring they are marked as 'changed' before saving
        return true;
    }
}

Predefined model classes

This extension is generic and may be applied to any model with complex attributes. However, to simplify integration with common solutions several base classes are provided by this extension:

  • [[\yii2tech\embedded\mongodb\ActiveRecord]] - MongoDB ActiveRecord with embedded feature built-in
  • [[\yii2tech\embedded\mongodb\ActiveRecordFile]] - MongoDB GridFS ActiveRecord with embedded feature built-in
  • [[\yii2tech\embedded\elasticsearch\ActiveRecord]] - ElasticSearch ActiveRecord with embedded feature built-in

Provided ActiveRecord classes already implement [[\yii2tech\embedded\ContainerInterface]] and invoke refreshFromEmbedded() on beforeSave() stage. For example, if you are using MongoDB and wish to work with sub-documents, you may simply switch extending from regular [[\yii\mongodb\ActiveRecord]] to [[\yii2tech\embedded\mongodb\ActiveRecord]]:

class User extends \yii2tech\embedded\mongodb\ActiveRecord
{
    public static function collectionName()
    {
        return 'customer';
    }

    public function attributes()
    {
        return ['_id', 'name', 'email', 'addressData', 'status'];
    }

    public function embedAddress()
    {
        return $this->mapEmbedded('addressData', UserAddress::className());
    }
}

About

Support embedded models usage for complex ActiveRecord like MongoDB or ElasticSearch

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • PHP 100.0%