Merge pull request #5 from llm-agents-php/feature/embedding-generator

Adds embedding generator

composer.json

@@ -5,7 +5,7 @@
   "require": {
     "php": "^8.3",
     "openai-php/client": "^0.10.1",
-    "llm-agents/agents": "^1.4",
+    "llm-agents/agents": "^1.5",
     "guzzlehttp/guzzle": "^7.0"
   },
   "require-dev": {
@@ -24,7 +24,10 @@
     }
   },
   "config": {
-    "sort-packages": true
+    "sort-packages": true,
+    "allow-plugins": {
+      "php-http/discovery": false
+    }
   },
   "extra": {
     "laravel": {

src/Embeddings/EmbeddingGenerator.php

@@ -0,0 +1,34 @@
+<?php
+
+declare(strict_types=1);
+
+namespace LLM\Agents\OpenAI\Client\Embeddings;
+
+use LLM\Agents\Embeddings\Document;
+use LLM\Agents\Embeddings\Embedding;
+use LLM\Agents\Embeddings\EmbeddingGeneratorInterface;
+use OpenAI\Contracts\ClientContract;
+
+final readonly class EmbeddingGenerator implements EmbeddingGeneratorInterface
+{
+    public function __construct(
+        private ClientContract $client,
+        private OpenAIEmbeddingModel $model = OpenAIEmbeddingModel::TextEmbeddingAda002,
+    ) {}
+
+    public function generate(Document ...$documents): array
+    {
+        $documents = \array_values($documents);
+
+        $response = $this->client->embeddings()->create([
+            'model' => $this->model->value,
+            'input' => \array_map(static fn(Document $doc): string => $doc->content, $documents),
+        ]);
+
+        foreach ($response->embeddings as $i => $embedding) {
+            $documents[$i] = $documents[$i]->withEmbedding(new Embedding($embedding->embedding));
+        }
+
+        return $documents;
+    }
+}

src/Embeddings/OpenAIEmbeddingModel.php

@@ -0,0 +1,12 @@
+<?php
+
+declare(strict_types=1);
+
+namespace LLM\Agents\OpenAI\Client\Embeddings;
+
+enum OpenAIEmbeddingModel: string
+{
+    case TextEmbedding3Small = 'text-embedding-3-small';
+    case TextEmbedding3Large = 'text-embedding-3-large';
+    case TextEmbeddingAda002 = 'text-embedding-ada-002';
+}

src/Integration/Laravel/OpenAIClientServiceProvider.php

@@ -6,10 +6,14 @@
 
 use Illuminate\Contracts\Foundation\Application;
 use Illuminate\Support\ServiceProvider;
+use LLM\Agents\Embeddings\EmbeddingGeneratorInterface;
 use LLM\Agents\LLM\LLMInterface;
+use LLM\Agents\OpenAI\Client\Embeddings\EmbeddingGenerator;
+use LLM\Agents\OpenAI\Client\Embeddings\OpenAIEmbeddingModel;
 use LLM\Agents\OpenAI\Client\LLM;
 use LLM\Agents\OpenAI\Client\Parsers\ChatResponseParser;
 use LLM\Agents\OpenAI\Client\StreamResponseParser;
+use OpenAI\Contracts\ClientContract;
 use OpenAI\Responses\Chat\CreateStreamedResponse;
 
 final class OpenAIClientServiceProvider extends ServiceProvider
@@ -21,6 +25,24 @@ public function register(): void
             LLM::class,
         );
 
+        $this->app->singleton(
+            EmbeddingGeneratorInterface::class,
+            EmbeddingGenerator::class,
+        );
+
+        $this->app->singleton(
+            EmbeddingGenerator::class,
+            static function (
+                ClientContract $client,
+            ): EmbeddingGenerator {
+                return new EmbeddingGenerator(
+                    client: $client,
+                    // todo: use config
+                    model: OpenAIEmbeddingModel::TextEmbeddingAda002,
+                );
+            },
+        );
+
         $this->app->singleton(
             StreamResponseParser::class,
             static function (Application $app): StreamResponseParser {

src/Integration/Spiral/OpenAIClientBootloader.php

@@ -5,7 +5,10 @@
 namespace LLM\Agents\OpenAI\Client\Integration\Spiral;
 
 use GuzzleHttp\Client as HttpClient;
+use LLM\Agents\Embeddings\EmbeddingGeneratorInterface;
 use LLM\Agents\LLM\LLMInterface;
+use LLM\Agents\OpenAI\Client\Embeddings\EmbeddingGenerator;
+use LLM\Agents\OpenAI\Client\Embeddings\OpenAIEmbeddingModel;
 use LLM\Agents\OpenAI\Client\LLM;
 use LLM\Agents\OpenAI\Client\Parsers\ChatResponseParser;
 use LLM\Agents\OpenAI\Client\StreamResponseParser;
@@ -20,6 +23,19 @@ public function defineSingletons(): array
     {
         return [
             LLMInterface::class => LLM::class,
+            EmbeddingGeneratorInterface::class => EmbeddingGenerator::class,
+
+            EmbeddingGenerator::class => static function (
+                ClientContract $client,
+                EnvironmentInterface $env,
+            ): EmbeddingGenerator {
+                return new EmbeddingGenerator(
+                    client: $client,
+                    model: OpenAIEmbeddingModel::from(
+                        $env->get('OPENAI_EMBEDDING_MODEL', OpenAIEmbeddingModel::TextEmbedding3Small->value),
+                    ),
+                );
+            },
 
             ClientContract::class => static fn(
                 EnvironmentInterface $env,