Working with Elasticsearch in an Eloquent-like fashion.
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.
Laravel Elastic Query | Laravel | PHP | Elasticsearch |
---|---|---|---|
^0.1.0 | ^8.0 | ^8.0 | 7.* |
^0.2.0 | ^8.0 | ^8.0 | 7.* |
^0.3.0 | ^8.0 | ^8.0 | 7.* |
^0.3.2 | ^8.0 || ^9.0 | ^8.0 | 7.* |
^7.x (see details) | ^9.0 || ^10.0 || ^11.0 | ^8.1 | 7.* |
^8.0.0 | ^8.0 || ^9.0 | ^8.0 | 8.* |
^8.0.13 | ^8.0 || ^9.0 || ^10.0 | ^8.0 | 8.* |
^8.0.23 | ^8.0 || ^9.0 || ^10.0 || ^11.0 | ^8.0 | 8.* |
^8.1.0 | ^9.0 || ^10.0 || ^11.0 | ^8.1 | 8.* |
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();
$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.
$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.
$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');
$page = $searchQuery->paginate(15, 45);
Offset pagination returns total documents count as total
and current position as size/offset
.
$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.
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.
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');
});
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();
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();
Term suggester:
$aggQuery->term('suggestName', 'name.trigram')->text('glves')->...->get();
Phrase Suggester:
$aggQuery->phrase('suggestName', 'name.trigram')->text('glves')->...->get();
$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();
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();
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,
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()
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.
Please see CONTRIBUTING for details.
- composer install
- start Elasticsearch in your preferred way
- if you need change
ELASTICSEARCH_HOSTS
, copyphpunit.xml.dist
tophpunit.xml
and fill value - composer test
Please review our security policy on how to report security vulnerabilities.
The MIT License (MIT). Please see License File for more information.