GuillaumeRoques / elastically

🔍 JoliCode's Elastica wrapper to bootstrap Elasticsearch PHP integrations

Elastically, Elastica based framework

Opinionated Elastica based framework to bootstrap PHP and Elasticsearch implementations.

Main features:

• DTO are first class citizen, you send PHP object as documents, and get objects back on search results, like an ODM;
• All indexes are versioned / aliased automatically;
• Mappings is done in YAML;
• Analysis is separated from mappings to ease reuse;
• 100% compatibility with ruflin/elastica;
• Designed for Elasticsearch 7+;
• Symfony HttpClient compatible transport;
• Symfony support (optional):
  • See dedicated chapter;
  • Tested with Symfony 4.4 to 5;
  • Symfony Messenger Handler support (with or without spool);

Require PHP 7.4+ and Elasticsearch 7+.

Installation

composer require jolicode/elastically

Demo

If you are using Symfony, you can move to the Symfony chapter

Quick example of what the library do on top of Elastica:

// Your own DTO, or one generated by Jane (see below)
class Beer
{
    public string $foo;
    public string $bar;
}

use JoliCode\Elastically\Factory;
use Elastica\Document;

// Factory object with Elastica options + new Elastically options in the same array
$factory = new Factory([
    // Where to find the mappings
    Factory::CONFIG_MAPPINGS_DIRECTORY => __DIR__.'/mappings',
    // What object to find in each index
    Factory::CONFIG_INDEX_CLASS_MAPPING => [
        'beers' => Beer::class,
    ],
]);

// Class to perform request, same as the Elastica Client
$client = $factory->buildClient();

// Class to build Indexes
$indexBuilder = $factory->buildIndexBuilder();

// Create the Index in Elasticsearch
$index = $indexBuilder->createIndex('beers');

// Set the proper aliases
$indexBuilder->markAsLive($index, 'beers');

// Class to index DTO(s) in an Index
$indexer = $factory->getIndexer();

$dto = new Beer();
$dto->bar = 'American Pale Ale';
$dto->foo = 'Hops from Alsace, France';

// Add a document to the queue
$indexer->scheduleIndex('beers', new Document('123', $dto));
$indexer->flush();

// Set parameters on the Bulk
$indexer->setBulkRequestParams([
    'pipeline' => 'covfefe',
    'refresh' => 'wait_for'
]);

// Force index refresh if needed
$indexer->refresh('beers');

// Get the Document (new!)
$results = $client->getIndex('beers')->getDocument('123');

// Get the DTO (new!)
$results = $client->getIndex('beers')->getModel('123');

// Perform a search
$results = $client->getIndex('beers')->search('alsace');

// Get the Elastic Document
$results->getDocuments()[0];

// Get the Elastica compatible Result
$results->getResults()[0];

// Get the DTO 🎉 (new!)
$results->getResults()[0]->getModel();

// Create a new version of the Index "beers"
$index = $indexBuilder->createIndex('beers');

// Slow down the Refresh Interval of the new Index to speed up indexation
$indexBuilder->slowDownRefresh($index);
$indexBuilder->speedUpRefresh($index);

// Set proper aliases
$indexBuilder->markAsLive($index, 'beers');

// Clean the old indices (close the previous one and delete the older)
$indexBuilder->purgeOldIndices('beers');

// Mapping change? Just call migrate and enjoy a full reindex (use the Task API internally to avoid timeout)
$newIndex = $indexBuilder->migrate($index);
$indexBuilder->speedUpRefresh($newIndex);
$indexBuilder->markAsLive($newIndex, 'beers');

mappings/beers_mapping.yaml

# Anything you want, no validation
settings:
    number_of_replicas: 1
    number_of_shards: 1
    refresh_interval: 60s
mappings:
    dynamic: false
    properties:
        foo:
            type: text
            analyzer: english
            fields:
                keyword:
                    type: keyword

Configuration

This library add custom configurations on top of Elastica's:

Factory::CONFIG_MAPPINGS_DIRECTORY (required with default configuration)

The directory Elastically is going to look for YAML.

When creating a foobar index, a foobar_mapping.yaml file is expected.

If an analyzers.yaml file is present, all the indices will get it.

Factory::CONFIG_INDEX_CLASS_MAPPING (required)

An array of index name to class FQN.

[
  'indexName' => My\AwesomeDTO::class,
]

Factory::CONFIG_MAPPINGS_PROVIDER

An instance of MappingProviderInterface.

If this option is not defined, the factory will fallback to YamlProvider and will use Factory::CONFIG_MAPPINGS_DIRECTORY option.

There are two providers available in Elastically: YamlProvider and PhpProvider.

Factory::CONFIG_SERIALIZER (optional)

A SerializerInterface compatible object that will by used on indexation.

Default to Symfony Serializer with Object Normalizer.

A faster alternative is to use Jane to generate plain PHP Normalizer, see below. Also we recommend customization to handle things like Date.

Factory::CONFIG_DENORMALIZER (optional)

A DenormalizerInterface compatible object that will by used on search results to build your objects back.

If this option is not defined, the factory will fallback to Factory::CONFIG_SERIALIZER option.

Factory::CONFIG_SERIALIZER_CONTEXT_BUILDER (optional)

An instance of ContextBuilderInterface that build a serializer context from a class name.

If it is not defined, Elastically, will use a StaticContextBuilder with the configuration from Factory::CONFIG_SERIALIZER_CONTEXT_PER_CLASS.

Factory::CONFIG_SERIALIZER_CONTEXT_PER_CLASS (optional)

Allow to specify the Serializer context for normalization and denormalization.

[
    Beer::class => ['attributes' => ['title']],
];

Default to [].

Factory::CONFIG_BULK_SIZE (optional)

When running indexation of lots of documents, this setting allow you to fine-tune the number of document threshold.

Default to 100.

Factory::CONFIG_INDEX_PREFIX (optional)

Add a prefix to all indexes and aliases created via Elastically.

Default to null.

Usage in Symfony

Client as a service

Just declare the proper service in services.yaml:

services:
    JoliCode\Elastically\IndexNameMapper:
        arguments:
            $prefix: null # or a string to prefix index name
            $indexClassMapping:
                indexName: My\AwesomeDTO

    JoliCode\Elastically\Serializer\StaticContextBuilder:
        arguments:
            $mapping:
                My\AwesomeDTO: []

    JoliCode\Elastically\ResultSetBuilder:
        arguments:
            $indexNameMapper: '@JoliCode\Elastically\IndexNameMapper'
            $contextBuilder: '@JoliCode\Elastically\Serializer\StaticContextBuilder'
            $denormalizer: '@serializer'

    JoliCode\Elastically\Client:
        arguments:
            $config:
                host: '%env(ELASTICSEARCH_HOST)%'
            $logger: '@logger'
            $resultSetBuilder: '@JoliCode\Elastically\ResultSetBuilder'
            $indexNameMapper: '@JoliCode\Elastically\IndexNameMapper'

    JoliCode\Elastically\Indexer:
        arguments:
            $client: '@JoliCode\Elastically\Client'
            $serializer: '@serializer'
            $bulkMaxSize: 100
            $bulkRequestParams: []
            $contextBuilder: '@JoliCode\Elastically\Serializer\StaticContextBuilder'

    JoliCode\Elastically\Mapping\YamlProvider:
        arguments:
            $configurationDirectory: '%kernel.project_dir%/config/elasticsearch'

    JoliCode\Elastically\IndexBuilder:
        arguments:
            $mappingProvider: '@JoliCode\Elastically\Mapping\YamlProvider'
            $client: '@JoliCode\Elastically\Client'
            $indexNameMapper: '@JoliCode\Elastically\IndexNameMapper'

Using HttpClient as Transport

You can also use the Symfony HttpClient for all Elastica communications:

JoliCode\Elastically\Transport\HttpClientTransport: ~

JoliCode\Elastically\Client:
    arguments:
        $config:
            host: '%env(ELASTICSEARCH_HOST)%'
            transport: '@JoliCode\Elastically\Transport\HttpClientTransport'
            ...

Using Messenger for async indexing

Elastically ships with a default Message and Handler for Symfony Messenger.

Register the message in your configuration:

framework:
    messenger:
        transports:
            async: "%env(MESSENGER_TRANSPORT_DSN)%"

        routing:
            # async is whatever name you gave your transport above
            'JoliCode\Elastically\Messenger\IndexationRequest':  async

services:
    JoliCode\Elastically\Messenger\IndexationRequestHandler: ~

The IndexationRequestHandler service depends on an implementation of JoliCode\Elastically\Messenger\DocumentExchangerInterface, which isn't provided by this library. You must provide a service that implements this interface, so you can plug your database or any other source of truth.

Then from your code you have to call:

use JoliCode\Elastically\Messenger\IndexationRequest;
use JoliCode\Elastically\Messenger\IndexationRequestHandler;

$bus->dispatch(new IndexationRequest(Product::class, '1234567890'));

// Third argument is the operation, so for a delete:
// new IndexationRequest(Product::class, 'ref9999', IndexationRequestHandler::OP_DELETE);

And then consume the messages:

$ php bin/console messenger:consume async

Grouping IndexationRequest in a spool

Sending multiple IndexationRequest during the same Symfony Request is not always appropriate, it will trigger multiple Bulk operations. Elastically provides a Kernel listener to group all the IndexationRequest in a single MultipleIndexationRequest message.

To use this mechanism, we send the IndexationRequest in a memory transport to be consumed and grouped in a really async transport:

messenger:
    transports:
        async: "%env(MESSENGER_TRANSPORT_DSN)%"
        queuing: 'in-memory:///'

    routing:
        'JoliCode\Elastically\Messenger\MultipleIndexationRequest': async
        'JoliCode\Elastically\Messenger\IndexationRequest': queuing

You also need to register the subscriber:

services:
    JoliCode\Elastically\Messenger\IndexationRequestSpoolSubscriber:
        arguments:
            - '@messenger.transport.queuing' # should be the name of the memory transport
            - '@messenger.default_bus'
        tags:
            - { name: kernel.event_subscriber }

Using Jane to build PHP DTO and fast Normalizers

Install JanePHP json-schema tools to build your own DTO and Normalizers. All you have to do is setting the Jane-completed Serializer on the Factory:

$factory = new Factory([
    Factory::CONFIG_SERIALIZER => $serializer,
]);

Not compatible with Jane < 6.

To be done

• some "todo" in the code
• optional Doctrine connector
• better logger - maybe via a processor? extending _log is supposed to be deprecated :(
• extra commands to monitor, update mapping, reindex... Commonly implemented tasks
• optional Symfony integration (DIC)
  • web debug toolbar!
• scripts / commands for common tasks:
  • auto-reindex when the mapping change, handle the aliases and everything
  • micro monitoring for cluster / indexes
  • health-check method

Sponsors

Open Source time sponsored by JoliCode.