README

Working with Elasticsearch in an Eloquent-like fashion.

Installation

You can install the package via composer:

composer require ensi/laravel-elastic-query

Publish config file like this:

php artisan vendor:publish --provider="Ensi\LaravelElasticQuery\ElasticQueryServiceProvider"

Set ELASTICSEARCH_HOSTS in your .env file. , can be used as a delimeter.

Version Compatibility

Basic usage

Let's create and index class. It's someting like Eloquent model.

use Ensi\LaravelElasticQuery\ElasticIndex;

class ProductsIndex extends ElasticIndex
{
    protected string $name = 'test_products';
    protected string $tiebreaker = 'product_id';
}

You should set a unique in document attribute name in $tiebreaker. It is used as an additional sort in search_after

Now we can get some documents

$searchQuery = ProductsIndex::query();

$hits = $searchQuery
             ->where('rating', '>=', 5)
             ->whereDoesntHave('offers', fn(BoolQuery $query) => $query->where('seller_id', 10)->where('active', false))
             ->sortBy('rating', 'desc')
             ->sortByNested('offers', fn(SortableQuery $query) => $query->where('active', true)->sortBy('price', mode: 'min'))
             ->take(25)
             ->get();

Filtering

$searchQuery->where('field', 'value');
$searchQuery->where('field', '>', 'value'); // supported operators: `=` `!=` `>` `<` `>=` `<=`
$searchQuery->whereNot('field', 'value'); // equals `where('field', '!=', 'value')`

$searchQuery->whereIn('field', ['value1', 'value2']);
$searchQuery->whereNotIn('field', ['value1', 'value2']);

$searchQuery->whereNull('field');
$searchQuery->whereNotNull('field');

$searchQuery->whereHas('nested_field', fn(BoolQuery $subQuery) => $subQuery->where('field_in_nested', 'value'));
$searchQuery->whereDoesntHave(
    'nested_field',
    function (BoolQuery $subQuery) {
        $subQuery->whereHas('nested_field', fn(BoolQuery $subQuery2) => $subQuery2->whereNot('field', 'value'));
    }
);

nested_field must have nested type. Subqueries cannot use fields of main document only subdocument.

Full text search

$searchQuery->whereMatch('field_one', 'query string');
$searchQuery->whereMultiMatch(['field_one^3', 'field_two'], 'query string', MatchType::MOST_FIELDS);
$searchQuery->whereMultiMatch([], 'query string');  // search by all text fields

field_one and field_two must be of text type. If no type is given, the MatchType::BEST_FIELDS is used.

Sorting

$searchQuery->sortBy('field', SortOrder::DESC, SortMode::MAX, MissingValuesMode::FIRST); // field is from main document
$searchQuery->sortByNested(
    'nested_field',
    fn(SortableQuery $subQuery) => $subQuery->where('field_in_nested', 'value')->sortBy('field')
);

Second attribute is a direction. It supports asc and desc values. Defaults to asc.
Third attribute - sorting type. List of supporting types: min, max, avg, sum, median. Defaults to min.

There are also dedicated sort methods for each sort type.

$searchQuery->minSortBy('field', 'asc');
$searchQuery->maxSortBy('field', 'asc');
$searchQuery->avgSortBy('field', 'asc');
$searchQuery->sumSortBy('field', 'asc');
$searchQuery->medianSortBy('field', 'asc');

Pagination

Offset Pagination

$page = $searchQuery->paginate(15, 45);

Offset pagination returns total documents count as total and current position as size/offset.

Cursor pagination

$page = $searchQuery->cursorPaginate(10);
$pageNext = $searchQuery->cursorPaginate(10, $page->next);

current, next, previous is returned in this case instead of total, size and offset. You can check Laravel docs for more info about cursor pagination.

Aggregation

Aggregaction queries can be created like this

$aggQuery = ProductsIndex::aggregate();

/** @var \Illuminate\Support\Collection $aggs */
$aggs = $aggQuery
            ->where('active', true)
            ->terms('codes', 'code')
            ->count('product_count', 'product_id')
            ->nested(
                'offers',
                fn(AggregationsBuilder $builder) => $builder->where('seller_id', 10)->minmax('price', 'price')
            );
            

Type of $aggs->price is MinMax. Type of $aggs->codes is BucketCollection. Aggregate names must be unique for whole query.

Aggregate types

Get all variants of attribute values:

$aggQuery->terms('agg_name', 'field', 25);

Get min and max attribute values. E.g for date:

$aggQuery->minmax('agg_name', 'field');

Get count unique attribute values:

$aggQuery->count('agg_name', 'field');

Aggregation plays nice with nested documents.

$aggQuery->nested('nested_field', function (AggregationsBuilder $builder) {
    $builder->terms('name', 'field_in_nested');
});

There is also a special virtual composite aggregate on the root level. You can set special conditions using it.

$aggQuery->composite(function (AggregationsBuilder $builder) {
    $builder->where('field', 'value')
        ->whereHas('nested_field', fn(BoolQuery $query) => $query->where('field_in_nested', 'value2'))
        ->terms('field1', 'agg_name1')
        ->minmax('field2', 'agg_name2');
});

Suggesting

Suggest queries can be created like this

$sugQuery = ProductsIndex::suggest();

/** @var \Illuminate\Support\Collection $suggests */
$suggests = $sugQuery->phrase('suggestName', 'name.trigram')
    ->text('glves')
    ->size(1)
    ->shardSize(3)
    ->get();
            

Global suggest text

User can set global text like this

$sugQuery = ProductsIndex::suggest()->text('glves');

$sugQuery->phrase('suggestName1', 'name.trigram')->size(1)->shardSize(3);
    
$sugQuery->phrase('suggestName2', 'name.trigram');
    
/** @var \Illuminate\Support\Collection $suggests */
$suggests = $sugQuery->get();
            

Suggester types

Term suggester:

$aggQuery->term('suggestName', 'name.trigram')->text('glves')->...->get();

Phrase Suggester:

$aggQuery->phrase('suggestName', 'name.trigram')->text('glves')->...->get();

Additional methods

$index = new ProductsIndex();

$index->isCreated(); // Check if index are created 
$index->create(); // Create index with structure from settings() method
$index->bulk(); // Send bulk request
$index->get(); // Send get request
$index->documentDelete(); // Send documentDelete request
$index->deleteByQuery(); // Send deleteByQuery request
$index->termvectors(); // Send termvectors request

$index->catIndices();
$index->indicesDelete();
$index->indicesRefresh();
$index->indicesReloadSearchAnalyzers();

Query Log

Just like Eloquent ElasticQuery has its own query log, but you need to enable it manually Each message contains indexName, query and timestamp

ElasticQuery::enableQueryLog();

/** @var \Illuminate\Support\Collection|Ensi\LaravelElasticQuery\Debug\QueryLogRecord[] $records */
$records = ElasticQuery::getQueryLog();

ElasticQuery::disableQueryLog();

Environment Variables

Below see the environment variables that you can configure with the default values, Hosts should be comma seperated string of hosts with protocol prefix and port suffix, e.g. http://localhost:9200,http://localhost:9201

 ELASTICSEARCH_HOSTS=https://localhost:9200'
 ELASTICSEARCH_RETRIES=2
 ELASTICSEARCH_USERNAME=admin
 ELASTICSEARCH_PASSWORD=admin
 ELASTICSEARCH_SSL_VERIFICATION=true,

Async Usage

All methods can return a Promise.
To enable this, you will need to add http_async_client to your config and then execute ElasticSearch::getClient()->setAsync(true).
To disable: ElasticQuery::getClient()->setAsync(false).

For example:

laravel-elastic-query.php:

return [
    'connection' => [
    
        // ..

        'http_async_client' => [HttpClientOptionsBuilder::class, 'getAsyncClient'],
    ],
];

HttpClientOptionsBuilder:

use Http\Adapter\Guzzle7\Client as GuzzleAdapter;
use Http\Client\HttpAsyncClient;

class HttpClientOptionsBuilder
{
    public static function getAsyncClient(): HttpAsyncClient
    {
        return GuzzleAdapter::createWithConfig([]);
    }
}

Action:

use Ensi\LaravelElasticQuery\ElasticQuery;

ElasticQuery::getClient()->setAsync(true);

// With async
$promises = [
    'key1' => FirstIndex::query()->get(),
    'key2' => FirstIndex::suggest()->paginate(/* ... */),
];

$results = [];
foreach ($promises as $key => $promise) {
    $results[$key] = $promise->wait();
}

$firstResponse = $results['key1'];

ElasticQuery::getClient()->setAsync(false);

// Without async
$firstResponse = FirstIndex::query()->get()

Elasticsearch 7 and 8 support.

Due to the incompatibility of clients for Elasticsearch 7 and 8, separate releases will be created for these versions. Development for each version is carried out in the corresponding branch.

To make changes to version 7, you need to create a task branch based on v7 and make a pull request to it. For version 8 it is similar, but based on the v8 branch.

Contributing

Please see CONTRIBUTING for details.

Testing

1. composer install
2. start Elasticsearch in your preferred way
3. if you need change ELASTICSEARCH_HOSTS, copy phpunit.xml.dist to phpunit.xml and fill value
4. composer test

Security Vulnerabilities

Please review our security policy on how to report security vulnerabilities.

License

The MIT License (MIT). Please see License File for more information.