feature #548 [Store] Add $options parameter to Vectorizer methods for platform configuration (OskarStark)

OskarStark · OskarStark · commit b2f0973000ba · 2025-09-12T13:26:14.000+02:00
This PR was squashed before being merged into the main branch. Discussion ---------- [Store] Add `$options` parameter to `Vectorizer` methods for platform configuration | Q | A | ------------- | --- | Bug fix? | no | New feature? | yes | Docs? | yes | Issues | Fix #495, fix #511 | License | MIT Commits ------- 60e4034 [Store] Add `$options` parameter to `Vectorizer` methods for platform configuration
diff --git a/src/store/src/Document/Vectorizer.php b/src/store/src/Document/Vectorizer.php
@@ -28,14 +28,14 @@ public function __construct(
     ) {
     }
 
-    public function vectorizeTextDocuments(array $documents): array
+    public function vectorizeTextDocuments(array $documents, array $options = []): array
     {
         $documentCount = \count($documents);
         $this->logger->info('Starting vectorization process', ['document_count' => $documentCount]);
 
         if ($this->model->supports(Capability::INPUT_MULTIPLE)) {
             $this->logger->debug('Using batch vectorization with model that supports multiple inputs');
-            $result = $this->platform->invoke($this->model, array_map(fn (TextDocument $document) => $document->content, $documents));
+            $result = $this->platform->invoke($this->model, array_map(fn (TextDocument $document) => $document->content, $documents), $options);
 
             $vectors = $result->asVectors();
             $this->logger->debug('Batch vectorization completed', ['vector_count' => \count($vectors)]);
@@ -44,7 +44,7 @@ public function vectorizeTextDocuments(array $documents): array
             $results = [];
             foreach ($documents as $i => $document) {
                 $this->logger->debug('Vectorizing document', ['document_index' => $i, 'document_id' => $document->id]);
-                $results[] = $this->platform->invoke($this->model, $document->content);
+                $results[] = $this->platform->invoke($this->model, $document->content, $options);
             }
 
             $vectors = [];
@@ -67,11 +67,11 @@ public function vectorizeTextDocuments(array $documents): array
         return $vectorDocuments;
     }
 
-    public function vectorize(string|\Stringable $string): Vector
+    public function vectorize(string|\Stringable $string, array $options = []): Vector
     {
         $this->logger->debug('Vectorizing string', ['string' => (string) $string]);
 
-        $result = $this->platform->invoke($this->model, (string) $string);
+        $result = $this->platform->invoke($this->model, (string) $string, $options);
         $vectors = $result->asVectors();
 
         if (!isset($vectors[0])) {
diff --git a/src/store/src/Document/VectorizerInterface.php b/src/store/src/Document/VectorizerInterface.php
@@ -22,14 +22,17 @@
 interface VectorizerInterface
 {
     /**
-     * @param TextDocument[] $documents
+     * @param TextDocument[]       $documents
+     * @param array<string, mixed> $options   Options to pass to the underlying platform
      *
      * @return VectorDocument[]
      */
-    public function vectorizeTextDocuments(array $documents): array;
+    public function vectorizeTextDocuments(array $documents, array $options = []): array;
 
     /**
      * Vectorizes a single string or Stringable object into a Vector.
+     *
+     * @param array<string, mixed> $options Options to pass to the underlying platform
      */
-    public function vectorize(string|\Stringable $string): Vector;
+    public function vectorize(string|\Stringable $string, array $options = []): Vector;
 }
diff --git a/src/store/tests/Document/VectorizerTest.php b/src/store/tests/Document/VectorizerTest.php
@@ -21,6 +21,7 @@
 use Symfony\AI\Platform\Model;
 use Symfony\AI\Platform\ModelClientInterface;
 use Symfony\AI\Platform\Platform;
+use Symfony\AI\Platform\PlatformInterface;
 use Symfony\AI\Platform\Result\RawHttpResult;
 use Symfony\AI\Platform\Result\RawResultInterface;
 use Symfony\AI\Platform\Result\ResultInterface;
@@ -379,4 +380,151 @@ public function testVectorizeStringThrowsExceptionWhenNoVectorReturned()
 
         $vectorizer->vectorize($text);
     }
+
+    public function testVectorizeTextDocumentsPassesOptionsToInvoke()
+    {
+        $documents = [
+            new TextDocument(Uuid::v4(), 'Test document', new Metadata(['source' => 'test'])),
+        ];
+
+        $vector = new Vector([0.1, 0.2, 0.3]);
+        $options = ['max_tokens' => 1000, 'temperature' => 0.5];
+
+        $platform = $this->createMock(PlatformInterface::class);
+        $platform->expects($this->once())
+            ->method('invoke')
+            ->with(
+                $this->isInstanceOf(Model::class),
+                $this->equalTo('Test document'),
+                $this->equalTo($options)
+            )
+            ->willReturn(new ResultPromise(fn () => new VectorResult($vector), $this->createMock(RawResultInterface::class)));
+
+        $model = new Embeddings();
+
+        $vectorizer = new Vectorizer($platform, $model);
+        $result = $vectorizer->vectorizeTextDocuments($documents, $options);
+
+        $this->assertCount(1, $result);
+        $this->assertEquals($vector, $result[0]->vector);
+    }
+
+    public function testVectorizeTextDocumentsWithEmptyOptions()
+    {
+        $documents = [
+            new TextDocument(Uuid::v4(), 'Test document'),
+        ];
+
+        $vector = new Vector([0.1, 0.2, 0.3]);
+
+        $platform = $this->createMock(PlatformInterface::class);
+        $platform->expects($this->once())
+            ->method('invoke')
+            ->with(
+                $this->isInstanceOf(Model::class),
+                $this->equalTo('Test document'),
+                $this->equalTo([])
+            )
+            ->willReturn(new ResultPromise(fn () => new VectorResult($vector), $this->createMock(RawResultInterface::class)));
+
+        $model = new Embeddings();
+
+        $vectorizer = new Vectorizer($platform, $model);
+        $result = $vectorizer->vectorizeTextDocuments($documents);
+
+        $this->assertCount(1, $result);
+        $this->assertEquals($vector, $result[0]->vector);
+    }
+
+    public function testVectorizeStringPassesOptionsToInvoke()
+    {
+        $text = 'Test string';
+        $vector = new Vector([0.1, 0.2, 0.3]);
+        $options = ['temperature' => 0.7, 'max_tokens' => 500];
+
+        $platform = $this->createMock(PlatformInterface::class);
+        $platform->expects($this->once())
+            ->method('invoke')
+            ->with(
+                $this->isInstanceOf(Model::class),
+                $this->equalTo($text),
+                $this->equalTo($options)
+            )
+            ->willReturn(new ResultPromise(fn () => new VectorResult($vector), $this->createMock(RawResultInterface::class)));
+
+        $model = new Embeddings();
+
+        $vectorizer = new Vectorizer($platform, $model);
+        $result = $vectorizer->vectorize($text, $options);
+
+        $this->assertEquals($vector, $result);
+    }
+
+    public function testVectorizeStringWithEmptyOptions()
+    {
+        $text = 'Test string';
+        $vector = new Vector([0.1, 0.2, 0.3]);
+
+        $platform = $this->createMock(PlatformInterface::class);
+        $platform->expects($this->once())
+            ->method('invoke')
+            ->with(
+                $this->isInstanceOf(Model::class),
+                $this->equalTo($text),
+                $this->equalTo([])
+            )
+            ->willReturn(new ResultPromise(fn () => new VectorResult($vector), $this->createMock(RawResultInterface::class)));
+
+        $model = new Embeddings();
+
+        $vectorizer = new Vectorizer($platform, $model);
+        $result = $vectorizer->vectorize($text);
+
+        $this->assertEquals($vector, $result);
+    }
+
+    public function testVectorizeTextDocumentsWithoutBatchSupportPassesOptions()
+    {
+        $model = $this->createMock(Model::class);
+        $model->expects($this->once())
+            ->method('supports')
+            ->with(Capability::INPUT_MULTIPLE)
+            ->willReturn(false);
+
+        $documents = [
+            new TextDocument(Uuid::v4(), 'Document 1'),
+            new TextDocument(Uuid::v4(), 'Document 2'),
+        ];
+
+        $vectors = [
+            new Vector([0.1, 0.2]),
+            new Vector([0.3, 0.4]),
+        ];
+
+        $options = ['max_tokens' => 2000];
+
+        $platform = $this->createMock(PlatformInterface::class);
+
+        $invokeCallCount = 0;
+        $platform->expects($this->exactly(2))
+            ->method('invoke')
+            ->willReturnCallback(function ($passedModel, $passedContent, $passedOptions) use ($options, $vectors, &$invokeCallCount) {
+                $this->assertInstanceOf(Model::class, $passedModel);
+                $this->assertEquals($options, $passedOptions);
+
+                $expectedContent = 0 === $invokeCallCount ? 'Document 1' : 'Document 2';
+                $this->assertEquals($expectedContent, $passedContent);
+
+                $vector = $vectors[$invokeCallCount++];
+
+                return new ResultPromise(fn () => new VectorResult($vector), $this->createMock(RawResultInterface::class));
+            });
+
+        $vectorizer = new Vectorizer($platform, $model);
+        $result = $vectorizer->vectorizeTextDocuments($documents, $options);
+
+        $this->assertCount(2, $result);
+        $this->assertEquals($vectors[0], $result[0]->vector);
+        $this->assertEquals($vectors[1], $result[1]->vector);
+    }
 }
