Home

Awesome

Elasticsearch Datasource for CakePHP

Build Status Latest Stable Version Total Downloads Code Coverage Software License

Use Elastic Search as an alternative ORM backend in CakePHP 5.0+.

You can find the documentation for the plugin in the Cake Book.

Installing Elasticsearch via composer

You can install Elasticsearch into your project using composer. For existing applications you can add the following to your composer.json file:

"require": {
    "cakephp/elastic-search": "^4.0"
}

And run php composer.phar update

Versions Table

Cake\ElasticSearchCakePHPElasticSearch
1.x3.0 - 3.52.x - 5.x
2.x3.6+6.x
>3, <3.4.04.0+6.x
>=3.4.04.0+7.x
4.x5.0+7.x

You are seeing the 3.x version.

Connecting the Plugin to your Application

After installing, you should tell your application to load the plugin:

use Cake\ElasticSearch\Plugin as ElasticSearchPlugin;

class Application extends BaseApplication
{
    public function bootstrap()
    {
        $this->addPlugin(ElasticSearchPlugin::class);

        // If you want to disable to automatically configure the Elastic model provider
        // and FormHelper do the following:
        // $this->addPlugin(ElasticSearchPlugin::class, [ 'bootstrap' => false ]);
    }
}

Defining a connection

Before you can do any work with Elasticsearch models, you'll need to define a connection:

// 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
        ],
    ]

As an alternative you could use a link format if you like to use enviroment variables for example.

// in config/app.php
    'Datasources' => [
        // other datasources
        'elastic' => [
            'url' => env('ELASTIC_URL', null)
        ]
    ]

    // and make sure the folowing env variable is available:
    // ELASTIC_URL="Cake\ElasticSearch\Datasource\Connection://127.0.0.1:9200?driver=Cake\ElasticSearch\Datasource\Connection"

You can enable request logging by setting the log config option to true. By default the debug Log profile will be used. You can also define an elasticsearch log profile in Cake\Log\Log to customize where Elasticsearch query logs will go. Query logging is done at a 'debug' level.

Getting a Index object

Index objects are the equivalent of ORM\Table instances in elastic search. You can use the IndexRegistry factory to get instances, much like TableRegistry:

use Cake\ElasticSearch\IndexRegistry;

$comments = IndexRegistry::get('Comments');

If you have loaded the plugin with bootstrap enabled you could load indexes using the model factory in your controllers

class SomeController extends AppController
{
    public function initialize()
    {
        $this->loadModel('Comments', 'Elastic');
    }

    public function index()
    {
        $comments = $this->Comments->find();
    }

    ...

Each Index object needs a correspondent Elasticsearch index, just like most of ORM\Table needs a database table.

In the above example, if you have defined a class as CommentsIndex and the IndexRegistry can find it, the $comments will receive a initialized object with inner configurations of connection and index. But if you don't have that class, a default one will be initialized and the index name on Elasticsearch mapped to the class.

The Index class

You must create your own Index class to define the name of internal index for Elasticsearch, as well as to define the mapping type and define any entity properties you need like virtual properties. As you have to use only one mapping type for each index, you can use the same name for both (the default behavior when type is undefined is use singular version of index name). Index types were removed in ElasticSearch 7.

use Cake\ElasticSearch\Index;

class CommentsIndex extends Index
{
    /**
     * The name of index in Elasticsearch
     *
     * @return  string
     */
    public function getName()
    {
        return 'comments';
    }
}

Running tests

We recommend using the included docker-compose.yml for doing local development. The Dockerfile contains the development environment, and an Elasticsearch container will be downloaded and started on port 9200.

# Start elasticsearch
docker-compose up -d

# Open an terminal in the development environment
docker-compose run console bash

Once inside the container you can install dependencies and run tests.

./composer.phar install
vendor/bin/phpunit

Warning: Please, be very carefully when running tests as the Fixture will create and drop Elasticsearch indexes for its internal structure. Don't run tests in production or development machines where you have important data into your Elasticsearch instance.

Assuming you have PHPUnit installed system wide using one of the methods stated here, you can run the tests for CakePHP by doing the following:

  1. Copy phpunit.xml.dist to phpunit.xml
  2. Run phpunit