chore: clean up after bundle 0.9 release

Author: chr-hertel

assets/styles/app.css

@@ -72,10 +72,18 @@ body {
 
             .rag & {
                 background: #dc8b6e;
+
+                a {
+                    color: #f4e973;
+                }
             }
 
             .youtube & {
                 background: #df3535;
+
+                a {
+                    color: #3e2926;
+                }
             }
 
             .wikipedia & {

composer.json

@@ -10,7 +10,7 @@
         "codewithkyrian/chromadb-php": "^0.3.0",
         "league/commonmark": "^2.6",
         "php-llm/llm-chain": "^0.9.3",
-        "php-llm/llm-chain-bundle": "dev-feat-prepare-0.8",
+        "php-llm/llm-chain-bundle": "^0.9",
         "phpdocumentor/reflection-docblock": "^5.5",
         "phpstan/phpdoc-parser": "^1.33",
         "runtime/frankenphp-symfony": "^0.2.0",
@@ -93,8 +93,6 @@
             "App\\Tests\\": "tests/"
         }
     },
-    "minimum-stability": "stable",
-    "prefer-stable": true,
     "scripts": {
         "post-install-cmd": [
             "@auto-scripts"

composer.lock

@@ -1,9 +1,23 @@
 llm_chain:
     platform:
+        # anthropic:
+        #     api_key: '%env(ANTHROPIC_API_KEY)%'
+        # azure:
+        #     gpt_deployment:
+        #         base_url: '%env(AZURE_GPT_BASE_URL)%'
+        #         deployment: '%env(AZURE_GPT_DEPLOYMENT)%'
+        #         api_key: '%env(AZURE_GPT_API_KEY)%'
+        #         api_version: '%env(AZURE_GPT_VERSION)%'
+        #     embeddings_deployment:
+        #         base_url: '%env(AZURE_EMBEDDINGS_BASE_URL)%'
+        #         deployment: '%env(AZURE_EMBEDDINGS_DEPLOYMENT)%'
+        #         api_key: '%env(AZURE_EMBEDDINGS_API_KEY)%'
+        #         api_version: '%env(AZURE_EMBEDDINGS_VERSION)%'
         openai:
             api_key: '%env(OPENAI_API_KEY)%'
     chain:
         rag:
+            # platform: 'llm_chain.platform.anthropic'
             model:
                 name: 'GPT'
                 version: 'gpt-4o-mini'
@@ -18,23 +32,34 @@ llm_chain:
             model:
                 name: 'GPT'
                 version: 'gpt-4o-mini'
+                options:
+                    temperature: 0.5
             tools:
                 - 'PhpLlm\LlmChain\Chain\ToolBox\Tool\Wikipedia'
     store:
         chroma_db:
             symfonycon:
-                host: '%env(CHROMADB_HOST)%'
                 collection: 'symfony_blog'
+            # web_summer_camp:
+            #     host: '%env(CHROMADB_HOST)%'
+            #     collection: 'wsc_program'
+    embedder:
+        default:
+            # platform: 'llm_chain.platform.anthropic'
+            # store: 'llm_chain.store.chroma_db.symfonycon'
+            model:
+                name: 'Embeddings'
+                version: 'text-embedding-ada-002'
 
 services:
     _defaults:
         autowire: true
         autoconfigure: true
 
+    # PhpLlm\LlmChain\Chain\ToolBox\Tool\Clock: ~
+    # PhpLlm\LlmChain\Chain\ToolBox\Tool\OpenMeteo: ~
+    # PhpLlm\LlmChain\Chain\ToolBox\Tool\SerpApi:
+    #     $apiKey: '%env(SERP_API_KEY)%'
     PhpLlm\LlmChain\Chain\ToolBox\Tool\Wikipedia: ~
     PhpLlm\LlmChain\Chain\ToolBox\Tool\SimilaritySearch: ~
 
-    # TODO: move to configuration
-    PhpLlm\LlmChain\Bridge\OpenAI\Embeddings: ~
-    PhpLlm\LlmChain\Model\EmbeddingsModel: '@PhpLlm\LlmChain\Bridge\OpenAI\Embeddings'
-

config/packages/llm_chain.yaml

@@ -4,57 +4,25 @@
 
 namespace App\Blog;
 
-use Codewithkyrian\ChromaDB\Client;
-use PhpLlm\LlmChain\Bridge\OpenAI\Embeddings;
-use PhpLlm\LlmChain\Document\Vector;
-use PhpLlm\LlmChain\Model\Response\AsyncResponse;
-use PhpLlm\LlmChain\Model\Response\VectorResponse;
-use PhpLlm\LlmChain\PlatformInterface;
+use PhpLlm\LlmChain\Document\Metadata;
+use PhpLlm\LlmChain\Document\TextDocument;
+use PhpLlm\LlmChain\Embedder as LlmChainEmbedder;
 
 final readonly class Embedder
 {
     public function __construct(
         private Loader $loader,
-        private PlatformInterface $platform,
-        private Client $chromaClient,
+        private LlmChainEmbedder $embedder,
     ) {
     }
 
     public function embedBlog(): void
     {
-        $posts = $this->loader->load();
-        $vectors = $this->createEmbeddings($posts);
-        $this->pushToChromaDB($posts, $vectors);
-    }
-
-    /**
-     * @param Post[] $posts
-     *
-     * @return Vector[]
-     */
-    private function createEmbeddings(array $posts): array
-    {
-        $texts = array_map(fn (Post $post) => $post->toString(), $posts);
-        $response = $this->platform->request(new Embeddings(), $texts);
-
-        assert($response instanceof AsyncResponse);
-        $response = $response->unwrap();
-        assert($response instanceof VectorResponse);
-
-        return $response->getContent();
-    }
-
-    /**
-     * @param Post[]   $posts
-     * @param Vector[] $vectors
-     */
-    private function pushToChromaDB(array $posts, array $vectors): void
-    {
-        $collection = $this->chromaClient->getOrCreateCollection('symfony_blog');
-
-        $ids = array_map(fn (Post $post) => $post->id, $posts);
-        $vectors = array_map(fn (Vector $vector) => $vector->getData(), $vectors);
+        $documents = [];
+        foreach ($this->loader->load() as $post) {
+            $documents[] = new TextDocument($post->id, $post->toString(), new Metadata($post->toArray()));
+        }
 
-        $collection->upsert($ids, $vectors, $posts);
+        $this->embedder->embed($documents);
     }
 }