ElasticSearch

The ElasticSearch plugin provides an ORM-like abstraction on top of elasticsearch. The plugin provides features that make testing, indexing documents and searching your indexes easier.

Installation

To install the ElasticSearch plugin, you can use composer. From your application’s ROOT directory (where composer.json file is located) run the following:

php composer.phar require cakephp/elastic-search "@stable"

You will need to add the following line to your application’s config/bootstrap.php file:

Plugin::load('Cake/ElasticSearch', ['bootstrap' => true]);

Additionally, you will need to configure the ‘elastic’ datasource connection in your config/app.php file. An example configuration would be:

// in config/app.php
'Datasources' => [
    // other datasources
    'elastic' => [
        'className' => 'Cake\ElasticSearch\Datasource\Connection',
        'driver' => 'Cake\ElasticSearch\Datasource\Connection',
        'host' => '127.0.0.1',
        'port' => 9200,
        'index' => 'my_apps_index',
    ],
]

Overview

The ElasticSearch plugin makes it easier to interact with an elasticsearch index and provides an interface similar to the Database Access & ORM. To get started you should create a Type object. Type objects are the “Repository” or table-like class in elasticsearch:

// in src/Model/Type/ArticlesType.php
namespace App\Model\Type;

use Cake\ElasticSearch\Type;

class ArticlesType extends Type
{
}

You can then use your type class in your controllers:

public function beforeFilter(Event $event)
{
    parent::beforeFilter($event);
    // Load the Type using the 'Elastic' provider.
    $this->loadModel('Articles', 'Elastic');
}

public function add()
{
    $article = $this->Articles->newEntity();
    if ($this->request->is('post')) {
        $article = $this->Articles->patchEntity($article, $this->request->getData());
        if ($this->Articles->save($article)) {
            $this->Flash->success('It saved');
        }
    }
    $this->set(compact('article'));
}

We would also need to create a basic view for our indexed articles:

// in src/Template/Articles/add.ctp
<?= $this->Form->create($article) ?>
<?= $this->Form->control('title') ?>
<?= $this->Form->control('body') ?>
<?= $this->Form->button('Save') ?>
<?= $this->Form->end() ?>

You should now be able to submit the form and have a new document added to elasticsearch.

Document Objects

Like the ORM, the Elasticsearch ODM uses Entities-like classes. The base class you should inherit from is Cake\ElasticSearch\Document. Document classes are found in the Model\Document namespace in your application or plugin:

namespace App\Model\Document;

class Article extends Document
{
}

Outside of constructor logic that makes Documents work with data from elasticsearch, the interface and functionality provided by Document are the same as those in Entities

Searching Indexed Documents

After you’ve indexed some documents you will want to search through them. The ElasticSearch plugin provides a query builder that allows you to build search queries:

$query = $this->Articles->find()
    ->where([
        'title' => 'special',
        'or' => [
            'tags in' => ['cake', 'php'],
            'tags not in' => ['c#', 'java']
        ]
    ]);

foreach ($query as $article) {
    echo $article->title;
}

You can use the FilterBuilder to add filtering conditions:

$query->where(function ($builder) {
    return $builder->and(
        $builder->gt('views', 99),
        $builder->term('author.name', 'sally')
    );
});

The FilterBuilder source has the complete list of methods with examples for many commonly used methods.

Validating Data & Using Application Rules

Like the ORM, the ElasticSearch plugin lets you validate data when marshalling documents. Validating request data, and applying application rules works the same as it does with the relational ORM. See the Validating Data Before Building Entities and Applying Application Rules sections for more information.

Saving New Documents

When you’re ready to index some data into elasticsearch, you’ll first need to convert your data into a Document that can be indexed:

$article = $this->Articles->newEntity($data);
if ($this->Articles->save($article)) {
    // Document was indexed
}

When marshalling a document, you can specify which embedded documents you wish to marshall using the associated key:

$article = $this->Articles->newEntity($data, ['associated' => ['Comments']]);

Saving a document will trigger the following events:

  • Model.beforeSave - Fired before the document is saved. You can prevent the save operation from happening by stopping this event.
  • Model.buildRules - Fired when the rules checker is built for the first time.
  • Model.afterSave - Fired after the document is saved.

Note

There are no events for embedded documents, as the parent document and all of its embedded documents are saved as one operation.

Updating Existing Documents

When you need to re-index data, you can patch existing entities and re-save them:

$query = $this->Articles->find()->where(['user.name' => 'jill']);
foreach ($query as $doc) {
    $doc->set($newProperties);
    $this->Articles->save($doc);
}

Deleting Documents

After retrieving a document you can delete it:

$doc = $this->Articles->get($id);
$this->Articles->delete($doc);

You can also delete documents matching specific conditions:

$this->Articles->deleteAll(['user.name' => 'bob']);

Embedding Documents

By defining embedded documents, you can attach entity classes to specific property paths in your documents. This allows you to provide custom behavior to the documents within a parent document. For example, you may want the comments embedded in an article to have specific application specific methods. You can use embedOne and embedMany to define embedded documents:

// in src/Model/Type/ArticlesType.php
namespace App\Model\Type;

use Cake\ElasticSearch\Type;

class ArticlesType extends Type
{
    public function initialize()
    {
        $this->embedOne('User');
        $this->embedMany('Comments', [
            'entityClass' => 'MyComment'
        ]);
    }
}

The above would create two embedded documents on the Article document. The User embed will convert the user property to instances of App\Model\Document\User. To get the Comments embed to use a class name that does not match the property name, we can use the entityClass option to configure a custom class name.

Once we’ve setup our embedded documents, the results of find() and get() will return objects with the correct embedded document classes:

$article = $this->Articles->get($id);
// Instance of App\Model\Document\User
$article->user;

// Array of App\Model\Document\Comment instances
$article->comments;

Getting Type Instances

Like the ORM, the ElasticSearch plugin provides a factory/registry for getting Type instances:

use Cake\ElasticSearch\TypeRegistry;

$articles = TypeRegistry::get('Articles');

Flushing the Registry

During test cases you may want to flush the registry. Doing so is often useful when you are using mock objects, or modifying a type’s dependencies:

TypeRegistry::flush();

Test Fixtures

The ElasticSearch plugin provides seamless test suite integration. Just like database fixtures, you can create test fixtures for elasticsearch. We could define a test fixture for our Articles type with the following:

namespace App\Test\Fixture;

use Cake\ElasticSearch\TestSuite\TestFixture;

/**
 * Articles fixture
 */
class ArticlesFixture extends TestFixture
{
    /**
     * The table/type for this fixture.
     *
     * @var string
     */
    public $table = 'articles';

    /**
     * The mapping data.
     *
     * @var array
     */
    public $schema = [
        'id' => ['type' => 'integer'],
        'user' => [
            'type' => 'nested',
            'properties' => [
                'username' => ['type' => 'string'],
            ]
        ]
        'title' => ['type' => 'string'],
        'body' => ['type' => 'string'],
    ];

    public $records = [
        [
            'user' => [
                'username' => 'billy'
            ],
            'title' => 'First Post',
            'body' => 'Some content'
        ]
    ];
}

The schema property uses the native elasticsearch mapping format. You can safely omit the type name and top level properties key. Once your fixtures are created you can use them in your test cases by including them in your test’s fixtures properties:

public $fixtures = ['app.articles'];