V1.9 hybrid search (#649)

* Add retrieveVector method to getSimilarDocuments * Update vector search test checking `_vector` field * Make document query request the vector * Rollback compose version * linter * Update src/Contracts/DocumentsQuery.php Co-authored-by: Tomas Norkūnas <norkunas.tom@gmail.com> --------- Co-authored-by: Tomas Norkūnas <norkunas.tom@gmail.com>
meilisearch · Jun 25, 2024 · 0b89ce9 · 0b89ce9
1 parent bbc1116
commit 0b89ce9
Show file tree

Hide file tree

Showing 5 changed files with 68 additions and 6 deletions.
diff --git a/src/Contracts/DocumentsQuery.php b/src/Contracts/DocumentsQuery.php
@@ -10,6 +10,7 @@ class DocumentsQuery
     private int $limit;
     private array $fields;
     private array $filter;
+    private ?bool $retrieveVectors = null;
 
     public function setOffset(int $offset): DocumentsQuery
     {
@@ -46,6 +47,16 @@ public function setFilter(array $filter): DocumentsQuery
         return $this;
     }
 
+    /**
+     * @param bool|null $retrieveVectors boolean value to show _vector details
+     */
+    public function setRetrieveVectors(?bool $retrieveVectors): DocumentsQuery
+    {
+        $this->retrieveVectors = $retrieveVectors;
+
+        return $this;
+    }
+
     /**
      * Checks if the $filter attribute has been set.
      *
@@ -84,6 +95,7 @@ public function toArray(): array
             'limit' => $this->limit ?? null,
             'filter' => $this->filter ?? null,
             'fields' => $this->fields(),
-        ], function ($item) { return null != $item || is_numeric($item); });
+            'retrieveVectors' => (null !== $this->retrieveVectors ? ($this->retrieveVectors ? 'true' : 'false') : null),
+        ], function ($item) { return null !== $item; });
     }
 }
diff --git a/src/Contracts/SimilarDocumentsQuery.php b/src/Contracts/SimilarDocumentsQuery.php
@@ -16,6 +16,7 @@ class SimilarDocumentsQuery
     private ?array $attributesToRetrieve = null;
     private ?bool $showRankingScore = null;
     private ?bool $showRankingScoreDetails = null;
+    private ?bool $retrieveVectors = null;
     private ?array $filter = null;
 
     /**
@@ -97,7 +98,17 @@ public function setShowRankingScoreDetails(?bool $showRankingScoreDetails): Simi
     }
 
     /**
-     * @return array{id: int|string, offset: non-negative-int, limit: positive-int, filter: array<int, array<int, string>|string>, embedder: non-empty-string, attributesToRetrieve: list<non-empty-string>, showRankingScore: bool, showRankingScoreDetails: bool} SimilarDocumentsQuery converted to an array with non null fields
+     * @param bool|null $retrieveVectors boolean value to show _vector details
+     */
+    public function setRetrieveVectors(?bool $retrieveVectors): SimilarDocumentsQuery
+    {
+        $this->retrieveVectors = $retrieveVectors;
+
+        return $this;
+    }
+
+    /**
+     * @return array{id: int|string, offset: non-negative-int, limit: positive-int, filter: array<int, array<int, string>|string>, embedder: non-empty-string, attributesToRetrieve: list<non-empty-string>, showRankingScore: bool, showRankingScoreDetails: bool, retrieveVectors: bool} SimilarDocumentsQuery converted to an array with non null fields
      */
     public function toArray(): array
     {
@@ -110,6 +121,7 @@ public function toArray(): array
             'attributesToRetrieve' => $this->attributesToRetrieve,
             'showRankingScore' => $this->showRankingScore,
             'showRankingScoreDetails' => $this->showRankingScoreDetails,
+            'retrieveVectors' => $this->retrieveVectors,
         ], function ($item) {
             return null !== $item;
         });

diff --git a/tests/Endpoints/DocumentsTest.php b/tests/Endpoints/DocumentsTest.php
@@ -11,6 +11,7 @@
 use Meilisearch\Exceptions\InvalidArgumentException;
 use Meilisearch\Exceptions\InvalidResponseBodyException;
 use Meilisearch\Exceptions\JsonEncodingException;
+use Meilisearch\Http\Client;
 use Psr\Http\Message\ResponseInterface;
 use Tests\TestCase;
 
@@ -624,6 +625,28 @@ public function testGetDocumentsWithoutFilterCorrectFieldsFormat(): void
         );
     }
 
+    public function testGetDocumentsWithVector(): void
+    {
+        $http = new Client($this->host, getenv('MEILISEARCH_API_KEY'));
+        $http->patch('/experimental-features', ['vectorStore' => true]);
+        $index = $this->createEmptyIndex($this->safeIndexName('movies'));
+
+        $promise = $index->updateEmbedders(['manual' => ['source' => 'userProvided', 'dimensions' => 3]]);
+        $this->assertIsValidPromise($promise);
+        $index->waitForTask($promise['taskUid']);
+        $promise = $index->updateDocuments(self::VECTOR_MOVIES);
+        $this->assertIsValidPromise($promise);
+        $index->waitForTask($promise['taskUid']);
+
+        $response = $index->getDocuments(new DocumentsQuery());
+        self::assertArrayNotHasKey('_vectors', $response->getResults()[0]);
+        $query = (new DocumentsQuery())->setRetrieveVectors(true);
+        $response = $index->getDocuments($query);
+        self::assertArrayHasKey('_vectors', $response->getResults()[0]);
+
+        self::assertCount(5, $response);
+    }
+
     public function testGetDocumentsMessageHintException(): void
     {
         $responseMock = $this->createMock(ResponseInterface::class);

diff --git a/tests/Endpoints/SearchTest.php b/tests/Endpoints/SearchTest.php
@@ -694,14 +694,22 @@ public function testVectorSearch(): void
         $http->patch('/experimental-features', ['vectorStore' => true]);
         $index = $this->createEmptyIndex($this->safeIndexName());
 
-        $promise = $index->updateEmbedders(['default' => ['source' => 'userProvided', 'dimensions' => 1]]);
+        $promise = $index->updateEmbedders(['manual' => ['source' => 'userProvided', 'dimensions' => 3]]);
         $this->assertIsValidPromise($promise);
         $index->waitForTask($promise['taskUid']);
+        $promise = $index->updateDocuments(self::VECTOR_MOVIES);
+        $this->assertIsValidPromise($promise);
+        $index->waitForTask($promise['taskUid']);
+
+        $response = $index->search('', ['vector' => [-0.5, 0.3, 0.85], 'hybrid' => ['semanticRatio' => 1.0]]);
+
+        self::assertSame(5, $response->getSemanticHitCount());
+        self::assertArrayNotHasKey('_vectors', $response->getHit(0));
 
-        $response = $index->search('', ['vector' => [1], 'hybrid' => ['semanticRatio' => 1.0]]);
+        $response = $index->search('', ['vector' => [-0.5, 0.3, 0.85], 'hybrid' => ['semanticRatio' => 1.0], 'retrieveVectors' => true]);
 
-        self::assertSame(0, $response->getSemanticHitCount());
-        self::assertEmpty($response->getHits());
+        self::assertSame(5, $response->getSemanticHitCount());
+        self::assertArrayHasKey('_vectors', $response->getHit(0));
     }
 
     public function testShowRankingScoreDetails(): void

diff --git a/tests/Endpoints/SimilarDocumentsTest.php b/tests/Endpoints/SimilarDocumentsTest.php
@@ -32,6 +32,13 @@ public function testBasicSearchWithSimilarDocuments(): void
         $documentId = $response->getHit(0)['id'];
         $response = $this->index->searchSimilarDocuments(new SimilarDocumentsQuery($documentId));
 
+        self::assertGreaterThanOrEqual(4, $response->getHitsCount());
+        self::assertArrayNotHasKey('_vectors', $response->getHit(0));
+        self::assertArrayHasKey('id', $response->getHit(0));
+        self::assertSame($documentId, $response->getId());
+
+        $similarQuery = new SimilarDocumentsQuery($documentId);
+        $response = $this->index->searchSimilarDocuments($similarQuery->setRetrieveVectors(true));
         self::assertGreaterThanOrEqual(4, $response->getHitsCount());
         self::assertArrayHasKey('_vectors', $response->getHit(0));
         self::assertArrayHasKey('id', $response->getHit(0));