sendynl / opensearch-query-builder
Build and execute an OpenSearch search query using a fluent PHP API
Requires
- php: ^8.2
Requires (Dev)
- friendsofphp/php-cs-fixer: ^3.0
- nyholm/psr7: ^1.8
- opensearch-project/opensearch-php: ^2.4.3
- php-http/mock-client: ^1.5
- phpstan/phpstan: ^2.1
- phpunit/phpunit: ^9.5
- spatie/ray: ^1.10
- symfony/http-client: ^7.3
README
This package is a lightweight query builder for OpenSearch. It is forked from the spatie/elasticsearch-query-builder library and modified to support OpenSearch. We're always open for PRs if you need anything specific!
use Sendy\OpenSearchQueryBuilder\Aggregations\MaxAggregation; use Sendy\OpenSearchQueryBuilder\Builder; use Sendy\OpenSearchQueryBuilder\Queries\MatchQuery; $transport = (new \OpenSearch\TransportFactory())->create(); $client = new \OpenSearch\Client($transport); $companies = (new Builder($client)) ->index('companies') ->addQuery(MatchQuery::create('name', 'sendy', fuzziness: 3)) ->addAggregation(MaxAggregation::create('score')) ->search();
Installation
You can install the package via composer:
composer require sendynl/opensearch-query-builder
Basic usage
The only class you really need to interact with is the Sendy\OpenSearchQueryBuilder\Builder
class. It requires an \OpenSearch\Client
passed in the constructor. Take a look at the OpenSearch SDK docs to learn more about connecting to your OpenSearch cluster.
The Builder
class contains some methods to add queries, aggregations, sorts, fields and some extras for pagination. You can read more about these methods below. Once you've fully built-up the query you can use $builder->search()
to execute the query or $builder->getPayload()
to get the raw payload for OpenSearch.
use Sendy\OpenSearchQueryBuilder\Queries\RangeQuery; use Sendy\OpenSearchQueryBuilder\Builder; $transport = (new \OpenSearch\TransportFactory())->create(); $client = new \OpenSearch\Client($transport); $builder = new Builder($client); $builder->addQuery(RangeQuery::create('age')->gte(18)); $results = $builder->search(); // raw response from OpenSearch
Multi-Search Queries
Multi-Search queries are also available using the MultiBuilder
class.
Adding queries
The $builder->addQuery()
method can be used to add any of the available Query
types to the builder. The available query types can be found below or in the src/Queries
directory of this repo. Every Query
has a static create()
method to pass its most important parameters.
The following query types are available:
ExistsQuery
https://docs.opensearch.org/docs/latest/query-dsl/term/exists/
\Sendy\OpenSearchQueryBuilder\Queries\ExistsQuery::create('terms_and_conditions');
GeoshapeQuery
https://docs.opensearch.org/docs/latest/query-dsl/geo-and-xy/geoshape/
\Sendy\OpenSearchQueryBuilder\Queries\GeoshapeQuery::create( 'location', \Sendy\OpenSearchQueryBuilder\Queries\GeoshapeQuery::TYPE_POLYGON, [[1.0, 2.0]], \Sendy\OpenSearchQueryBuilder\Queries\GeoShapeQuery::RELATION_INTERSECTS, );
MatchQuery
https://docs.opensearch.org/docs/latest/query-dsl/full-text/match/
\Sendy\OpenSearchQueryBuilder\Queries\MatchQuery::create('name', 'john doe', fuzziness: 2, boost: 5.0);
MatchPhraseQuery
https://docs.opensearch.org/docs/latest/query-dsl/full-text/match-phrase/
\Sendy\OpenSearchQueryBuilder\Queries\MatchPhraseQuery::create('name', 'john doe', slop: 2,zeroTermsQuery: "none",analyzer: "my_analyzer");
MultiMatchQuery
https://docs.opensearch.org/docs/latest/query-dsl/full-text/multi-match/
\Sendy\OpenSearchQueryBuilder\Queries\MultiMatchQuery::create('john', ['email', 'email'], fuzziness: 'auto');
NestedQuery
https://docs.opensearch.org/docs/latest/query-dsl/joining/nested/
\Sendy\OpenSearchQueryBuilder\Queries\NestedQuery::create( 'user', new \Sendy\OpenSearchQueryBuilder\Queries\MatchQuery('name', 'john') );
NestedQuery
InnerHits
https://docs.opensearch.org/docs/latest/search-plugins/searching-data/inner-hits/
$nestedQuery = \Sendy\OpenSearchQueryBuilder\Queries\NestedQuery::create( 'comments', \Sendy\OpenSearchQueryBuilder\Queries\TermsQuery::create('comments.published', true) ); $nestedQuery->innerHits( \Sendy\OpenSearchQueryBuilder\Queries\NestedQuery\InnerHits::create('top_three_liked_comments') ->size(3) ->addSort( \Sendy\OpenSearchQueryBuilder\Sorts\Sort::create( 'comments.likes', \Sendy\OpenSearchQueryBuilder\Sorts\Sort::DESC ) ) ->fields(['comments.content', 'comments.author', 'comments.likes']) );
RangeQuery
https://docs.opensearch.org/docs/latest/query-dsl/term/range/
\Sendy\OpenSearchQueryBuilder\Queries\RangeQuery::create('age') ->gte(18) ->lte(1337);
TermQuery
https://docs.opensearch.org/docs/latest/query-dsl/term/term/
\Sendy\OpenSearchQueryBuilder\Queries\TermQuery::create('user.id', 'flx');
TermsQuery
https://docs.opensearch.org/docs/latest/query-dsl/term/terms/
\Sendy\OpenSearchQueryBuilder\Queries\TermsQuery::create('user.id', ['flx', 'fly'], boost: 5.0);
WildcardQuery
https://docs.opensearch.org/docs/latest/query-dsl/term/wildcard/
\Sendy\OpenSearchQueryBuilder\Queries\WildcardQuery::create('user.id', '*doe');
PercolateQuery
https://docs.opensearch.org/docs/latest/field-types/supported-field-types/percolator/
\Sendy\OpenSearchQueryBuilder\Queries\PercolateQuery::create('query', ['title' => 'foo', 'body' => 'bar']);
BoolQuery
https://docs.opensearch.org/docs/latest/query-dsl/compound/bool/
\Sendy\OpenSearchQueryBuilder\Queries\BoolQuery::create() ->add($matchQuery, 'must_not') ->add($existsQuery, 'must_not');
collapse
The collapse
feature allows grouping search results by a specific field while retrieving top documents from each group using inner_hits
. This is useful for avoiding duplicate entities in search results while still accessing grouped data.
https://docs.opensearch.org/docs/latest/search-plugins/collapse-search/
use Sendy\OpenSearchQueryBuilder\Sorts\Sort; use Sendy\OpenSearchQueryBuilder\Builder; // Initialize ExtendedBuilder with an OpenSearch client $builder = new Builder($client); // Apply collapse to group by 'user_id' $builder->collapse( 'user_id', // Field to collapse on [ 'name' => 'top_three_liked_posts', 'size' => 3, // Retrieve top 3 posts per user 'sort' => [ Sort::create('post.likes', Sort::DESC), // Sort posts by likes (descending) ], 'fields' => ['post.title', 'post.content', 'post.likes'], // Select specific fields ], 10, // Max concurrent group searches ); // Execute the search $response = $builder->search();
Chaining multiple queries
Multiple addQuery()
calls can be chained on one Builder
. Under the hood they'll be added to a BoolQuery
with occurrence type must
. By passing a second argument to the addQuery()
method you can select a different occurrence type:
$builder ->addQuery( MatchQuery::create('name', 'billie'), 'must_not' // available types: must, must_not, should, filter ) ->addQuery( MatchQuery::create('team', 'eillish') );
More information on the boolean query and its occurrence types can be found in the OpenSearch docs.
Adding aggregations
The $builder->addAggregation()
method can be used to add any of the available Aggregation
s to the builder. The available aggregation types can be found below or in the src/Aggregations
directory of this repo. Every Aggregation
has a static create()
method to pass its most important parameters and sometimes some extra methods.
use Sendy\OpenSearchQueryBuilder\Aggregations\TermsAggregation; use Sendy\OpenSearchQueryBuilder\Builder; $transport = (new \OpenSearch\TransportFactory())->create(); $client = new \OpenSearch\Client($transport); $results = (new Builder($client)) ->addAggregation(TermsAggregation::create('genres', 'genre')) ->search(); $genres = $results['aggregations']['genres']['buckets'];
The following query types are available:
CardinalityAggregation
\Sendy\OpenSearchQueryBuilder\Aggregations\CardinalityAggregation::create('team_agg', 'team_name');
FilterAggregation
https://docs.opensearch.org/docs/latest/aggregations/bucket/filter/
\Sendy\OpenSearchQueryBuilder\Aggregations\FilterAggregation::create( 'tshirts', \Sendy\OpenSearchQueryBuilder\Queries\TermQuery::create('type', 'tshirt'), \Sendy\OpenSearchQueryBuilder\Aggregations\MaxAggregation::create('max_price', 'price') );
MaxAggregation
https://docs.opensearch.org/docs/latest/aggregations/metric/maximum/
\Sendy\OpenSearchQueryBuilder\Aggregations\MaxAggregation::create('max_price', 'price');
MinAggregation
https://docs.opensearch.org/docs/latest/aggregations/metric/minimum/
\Sendy\OpenSearchQueryBuilder\Aggregations\MinAggregation::create('min_price', 'price');
SumAggregation
https://docs.opensearch.org/docs/latest/aggregations/metric/sum/
\Sendy\OpenSearchQueryBuilder\Aggregations\SumAggregation::create('sum_price', 'price');
NestedAggregation
https://docs.opensearch.org/docs/latest/aggregations/bucket/nested/
\Sendy\OpenSearchQueryBuilder\Aggregations\NestedAggregation::create( 'resellers', 'resellers', \Sendy\OpenSearchQueryBuilder\Aggregations\MinAggregation::create('min_price', 'resellers.price'), \Sendy\OpenSearchQueryBuilder\Aggregations\MaxAggregation::create('max_price', 'resellers.price'), );
ReverseNestedAggregation
https://docs.opensearch.org/docs/latest/aggregations/bucket/reverse-nested/
\Sendy\OpenSearchQueryBuilder\Aggregations\ReverseNestedAggregation::create( 'name', ...$aggregations );
TermsAggregation
https://docs.opensearch.org/docs/latest/aggregations/bucket/terms/
\Sendy\OpenSearchQueryBuilder\Aggregations\TermsAggregation::create( 'genres', 'genre' ) ->size(10) ->order(['_count' => 'asc']) ->missing('N/A') ->aggregation(/* $subAggregation */);
TopHitsAggregation
https://docs.opensearch.org/docs/latest/aggregations/metric/top-hits/
\Sendy\OpenSearchQueryBuilder\Aggregations\TopHitsAggregation::create( 'top_sales_hits', size: 10, );
DateHistogramAggregation
https://docs.opensearch.org/docs/latest/aggregations/bucket/date-histogram/
\Sendynl\OpenSearchQueryBuilder\Aggregations\DateHistogramAggregation::create( 'name', '@timestamp' '1h', ) ->aggregation(/* $subAggregation */);
Adding sorts
The Builder
(and some aggregations) has a addSort()
method that takes a Sort
instance to sort the results. You can read more about how sorting works in the OpenSearch docs.
use Sendy\OpenSearchQueryBuilder\Sorts\Sort; $builder ->addSort(Sort::create('age', Sort::DESC)) ->addSort( Sort::create('score', Sort::ASC) ->unmappedType('long') ->missing(0) );
Nested sort
https://docs.opensearch.org/docs/latest/search-plugins/searching-data/sort/#sorting-nested-objects
use Sendy\OpenSearchQueryBuilder\Sorts\NestedSort; $builder ->addSort( NestedSort::create('books', 'books.rating', NestedSort::ASC) );
Nested sort with filter
use Sendy\OpenSearchQueryBuilder\Sorts\NestedSort; use Sendy\OpenSearchQueryBuilder\Queries\BoolQuery; use Sendy\OpenSearchQueryBuilder\Queries\TermQuery; $builder ->addSort( NestedSort::create( 'books', 'books.rating', NestedSort::ASC )->filter(BoolQuery::create()->add(TermQuery::create('books.category', 'comedy')) );
Retrieve specific fields
The fields()
method can be used to request specific fields from the resulting documents without returning the entire _source
entry. You can read more about the specifics of the fields parameter in the OpenSearch docs.
$builder->fields('user.id', 'http.*.status');
Highlighting
The highlight()
method can be used to add a highlight section to your query along the rules in the OpenSearch docs.
$highlightSettings = [ 'pre_tags' => ['<em>'], 'post_tags' => ['</em>'], 'fields' => [ '*' => (object) [] ] ]; $builder->highlight($highlightSettings);
Post filter
The addPostFilterQuery()
method can be used to add a post_filter BoolQuery to your query along the rules in the OpenSearch docs.
use Sendy\OpenSearchQueryBuilder\Queries\TermsQuery; $builder->addPostFilterQuery(TermsQuery::create('user.id', ['flx', 'fly']));
Pagination
Finally the Builder
also features a size()
and from()
method for the corresponding OpenSearch search parameters. These can be used to build a paginated search. Take a look the following example to get a rough idea:
use Sendy\OpenSearchQueryBuilder\Builder; $pageSize = 100; $pageNumber = $_GET['page'] ?? 1; $transport = (new \OpenSearch\TransportFactory())->create(); $client = new \OpenSearch\Client($transport); $pageResults = (new Builder($client)) ->size($pageSize) ->from(($pageNumber - 1) * $pageSize) ->search();
Multi-Search Query Builder
OpenSearch provides a "multi-search" API that allows for multiple query bodies to be included in a single request.
Use the MultiBuilder
class and add builders to add builders to your query request. The response will include a responses
array of the query results, in the same order the requests are added. Use the $multiBuilder->search()
to execute the queries, or $multiBuilder->getPayload()
for the raw request payload.
use Sendy\OpenSearchQueryBuilder\MultiBuilder; use Sendy\OpenSearchQueryBuilder\Builder; $transport = (new \OpenSearch\TransportFactory())->create(); $client = new \OpenSearch\Client($transport); $multiBuilder = (new MultiBuilder($client)); $multiBuilder->addBuilder( (new Builder($client))->index('custom_index')->size(10) ); // you can pass the index name to the addBuilder method second param $multiBuilder->addBuilder( (new Builder($client))->size(10) 'different_index' ); $multiResults = $multiBuilder->search();
Returns the following response JSON shape:
{
"took": 2,
"responses": [
{... first query result ...},
{... second query result ...},
]
}
Testing
composer test
Changelog
Please see CHANGELOG for more information on what has changed recently.
Contributing
Please see CONTRIBUTING for details.
Security Vulnerabilities
If you've found a bug regarding security please mail security@sendy.nl instead of using the issue tracker.
Credits
License
The MIT License (MIT). Please see License File for more information.