README

Mysticquent is an Elasticsearch ODM and mapper for Laravel. It renders the developer experience more enjoyable while using Elasticsearch, by providing a fluent syntax for mapping, querying, and storing eloquent models. This package inspired from sleimanx2/plastic and searchkick

This package is still under active development and may change.

Installing Mysticquent

composer require bidzm/mysticquent

Then we need to add the plastic service provider to config/app.php under the providers key:

Mysticquent\MysticquentServiceProvider::class

Finally we need to run:

php artisan vendor:publish

This will create a config file at config/mysticquent.php.

Usage

Defining Searchable Models
Storing Model Content
Searching
Aggregation
Suggestions
Mappings
Access The Client

Defining Searchable Models

To get started, enable searching capabilities in your model by adding the Mysticquent\Searchable trait:

use Mysticquent\Searchable;

class User extends Model
{
    use Searchable;
}

Defining what data to store.

By default, Mysticquent will store all visible properties of your model.

In addition, Mysticquent provides you to manually specify which data should be stored in Elasticsearch.

Providing a buildDocument method

public function buildDocument()
{
    return [
        'id' => $this->id,
        'email' => $this->email
    ];
}

Custom elastic type name

By the default Mysticquent will use class name of model as document type. You can customize it by adding a $documentType property to your model:

public $documentType = 'custom_type';

Custom elastic index name

By the default Mysticquent will use the model table name as index. You can customize in which index your model data will be stored by setting the $documentIndex property to your model:

public $documentIndex = 'custom_index';

Mysticquent automatically syncs model data with elastic when you save or delete your model from our SQL DB, however this feature can be disabled by adding public $syncDocument = false to your model.

Its important to note that manual document update should be performed in multiple scenarios:

1 - When you perform a bulk update or delete, no Eloquent event is triggered, therefore the document data won't be synced.

2 - Mysticquent doesn't listen to related models events (yet), so when you update a related model's content you should consider updating the parent document.

Saving a document

$book = User::first()->document()->save();

Partial updating a document

$book = User::first()->document()->update();

Deleting a document

$book = User::first()->document()->delete();

Saving documents in bulk

Mysticquent::document()->bulkSave(User::find(1)->books);

Deleting documents in bulk

$users = User::where('age','>',25)->get();

Mysticquent::document()->bulkDelete($users);

Searching Model Content

Plastic provides a fluent syntax to query Elasticsearch which leads to compact readable code. Lets dig into it:

$result = Book::search()->match('title','pulp')->get();

// Returns a collection of Book Models
$books = $result->hits();

// Returns the total number of matched documents
$result->totalHits();

// Returns the highest query score
$result->maxScore();

//Returns the time needed to execute the query
$result->took();

To get the raw DSL query that will be executed you can call toDSL():

$dsl = Book::search()->match('title','pulp')->toDSL();

Pagination

$books = Book::search()
    ->multiMatch(['title', 'description'], 'ham on rye', ['fuzziness' => 'AUTO'])
    ->sortBy('date')
    ->paginate();

You can still access the result object after pagination using the result method:

$books->result();

Bool Query

User::search()
    ->must()
        ->term('name','kimchy')
    ->mustNot()
        ->range('age',['from'=>10,'to'=>20]);
    ->should()
        ->match('bio','developer')
        ->match('bio','elastic')
    ->filter()
        ->term('tag','tech')
    ->get();

Nested Query

$contain = 'foo';

Post::search()
    ->multiMatch(['title', 'body'], $contain)
    ->nested('tags', function (SearchBuilder $builder) use ($contain) {
        $builder->match('tags.name', $contain);
    })->get();

Check out this documentation of supported search queries within Plastic and how to apply unsupported queries.

Change index on the fly

To switch to a different index for a single query, simply use the index method.

$result = Book::search()->index('special-books')->match('title','pulp')->get();

Aggregation

$result = User::search()
    ->match('bio', 'elastic')
    ->aggregate(function (AggregationBuilder $builder) {
        $builder->average('average_age', 'age');
    })->get();

$aggregations = $result->aggregations();

Check out this documentation of supported aggregations within plastic and how to apply unsupported aggregations.

Suggestions

Plastic::suggest()->completion('tag_suggest', 'photo')->get();

The suggestions query builder can also be accessed directly from the model as well:

//this be handy if you have a custom index for your model
Tag::suggest()->term('tag_term','admin')->get();

Model Mapping

Mappings are an important aspect of Elasticsearch. You can compare them to indexing in SQL databases. Mapping your models yields better and more efficient search results, and allows us to use some special query functions like nested fields and suggestions.

Generate a Model Mapping

php artisan make:mapping "App\User"

The new mapping will be placed in your database/mappings directory.

Mapping Structure

A mapping class contains a single method map. The map method is used to map the given model fields.

Within the map method you may use the Plastic Map builder to expressively create field maps. For example, let's look at a sample mapping that creates a Tag model map:

use Sleimanx2\Plastic\Map\Blueprint;
use Sleimanx2\Plastic\Mappings\Mapping;

class AppTag extends Mapping
{
    /**
     * Full name of the model that should be mapped
     *
     * @var string
     */
    protected $model = App\Tag::class;

    /**
     * Run the mapping.
     *
     * @return void
     */
    public function map()
    {
        Map::create($this->getModelType(), function (Blueprint $map) {
            $map->string('name')->store('true')->index('analyzed');

            // instead of the fluent syntax we can use the second method argument to fill the attributes
            $map->completion('suggestion', ['analyzer' => 'simple', 'search_analyzer' => 'simple']);
        },$this->getModelIndex());
    }
}

To learn about all of the methods available on the Map builder, check out this documentation.

Run Mappings

Running the created mappings can be done using the Artisan console command:

php artisan mapping:run

Updating Mappings

If your update consists only of adding a new field mapping you can always update our model map with your new field and run:

php artisan mapping:rerun

The mapping for existing fields cannot be updated or deleted, so you'll need to use one of following techniques to update existing fields.

1 - Create a new index

You can always create a new Elasticsearch index and re-run the mappings. After running the mappings you can use the bulkSave method to sync your SQL data with Elasticsearch.

2 - Using aliases

Its recommended to create your Elasticsearch index with an alias to ease the process of updating your model mappings with zero downtime. To learn more check out:

https://www.elastic.co/blog/changing-mapping-with-zero-downtime

Access The Client

You can access the Elasticsearch client to manage your indices and aliases as follows:

$client = Plastic::getClient();

//index delete
$client->indices()->delete(['index'=> Plastic::getDefaultIndex()]);
//index create
$client->indices()->create(['index' => Plastic::getDefaultIndex()]);

More about the official elastic client : https://github.com/elastic/elasticsearch-php

Contributing

Thank you for contributing, The contribution guide can be found Here.

License

Plastic is open-sourced software licensed under the MIT license.

To Do

Search Query Builder

implement Boosting query
implement ConstantScore query
implement DisMaxQuery query
implement MoreLikeThis query (with raw eloquent models)
implement GeoShape query

Aggregation Query Builder

implement Nested aggregation
implement ExtendedStats aggregation
implement TopHits aggregation

Mapping

Find a seamless way to update field mappings with zero downtime with aliases

General

Better query builder documentation

bidzm / mysticquent

Maintainers

Details