Fixing groq

app.py

@@ -71,45 +71,97 @@ app_state = ApplicationState()
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     # Startup
-    global groq_client, qdrant_client, embedding_service
-    
     if not Config.GROQ_API_KEY:
         raise ValueError("GROQ_API_KEY environment variable is required")
     
-    groq_client = AsyncGroq(api_key=Config.GROQ_API_KEY)
-    qdrant_client = AsyncQdrantClient(
-        url=Config.QDRANT_URL,
-        api_key=Config.QDRANT_API_KEY
-    )
+    print("Initializing services...")
     
-    # Initialize embedding service
-    embedding_service = None
+    # Initialize OpenAI client with Groq endpoint
+    try:
+        app_state.openai_client = AsyncOpenAI(
+            api_key=Config.GROQ_API_KEY,
+            base_url=Config.GROQ_BASE_URL
+        )
+        print("✓ OpenAI client initialized with Groq endpoint")
+    except Exception as e:
+        print(f"✗ Error initializing OpenAI client: {e}")
+        raise e
+    
+    # Initialize Qdrant client
+    try:
+        app_state.qdrant_client = AsyncQdrantClient(
+            url=Config.QDRANT_URL,
+            api_key=Config.QDRANT_API_KEY
+        )
+        print("✓ Qdrant client initialized")
+    except Exception as e:
+        print(f"✗ Error initializing Qdrant client: {e}")
+        raise e
     
-    # Verify connections
+    # Initialize embedding service
     try:
-        collections = await qdrant_client.get_collections()
-        print(f"Connected to Qdrant. Available collections: {[c.name for c in collections.collections]}")
+        print("Loading embedding model...")
+        app_state.embedding_service = EmbeddingService()
+        print(f"✓ Embedding model loaded: {Config.EMBEDDING_MODEL}")
+        print(f"✓ Model device: {Config.DEVICE}")
+        print(f"✓ Vector dimension: {app_state.embedding_service.dimension}")
     except Exception as e:
-        print(f"Warning: Could not connect to Qdrant: {e}")
+        print(f"✗ Error initializing embedding service: {e}")
+        raise e  # Fail fast if embedding service can't be initialized
     
-    # Check embedding model
+    # Verify Qdrant connection and auto-create collection
     try:
-        print(f"Embedding model loaded: {Config.EMBEDDING_MODEL}")
-        print(f"Model device: {Config.DEVICE}")
-        print(f"Vector dimension: {embedding_service.dimension}")
+        collections = await app_state.qdrant_client.get_collections()
+        collection_names = [c.name for c in collections.collections]
+        print(f"✓ Connected to Qdrant. Available collections: {collection_names}")
+        
+        # Check if our collection exists, if not create it
+        if Config.COLLECTION_NAME not in collection_names:
+            print(f"📁 Collection '{Config.COLLECTION_NAME}' not found. Creating automatically...")
+            try:
+                from qdrant_client.models import VectorParams, Distance
+                
+                await app_state.qdrant_client.create_collection(
+                    collection_name=Config.COLLECTION_NAME,
+                    vectors_config=VectorParams(
+                        size=app_state.embedding_service.dimension,
+                        distance=Distance.COSINE
+                    )
+                )
+                print(f"✓ Collection '{Config.COLLECTION_NAME}' created successfully!")
+                print(f"✓ Vector dimension: {app_state.embedding_service.dimension}")
+                print(f"✓ Distance metric: COSINE")
+            except Exception as create_error:
+                print(f"✗ Failed to create collection: {create_error}")
+                print("⚠ You may need to create the collection manually")
+        else:
+            print(f"✓ Collection '{Config.COLLECTION_NAME}' already exists")
+            
     except Exception as e:
-        print(f"Warning: Could not load embedding model: {e}")
+        print(f"⚠ Warning: Could not connect to Qdrant: {e}")
+        print("⚠ Collection auto-creation skipped")
+    
+    print("🚀 All services initialized successfully!")
     
     yield
     
     # Shutdown
-    if qdrant_client:
-        await qdrant_client.close()
+    print("Shutting down services...")
+    if app_state.qdrant_client:
+        await app_state.qdrant_client.close()
+        print("✓ Qdrant client closed")
+    if app_state.openai_client:
+        await app_state.openai_client.close()
+        print("✓ OpenAI client closed")
+    if app_state.embedding_service and hasattr(app_state.embedding_service, 'executor'):
+        app_state.embedding_service.executor.shutdown(wait=True)
+        print("✓ Embedding service executor shutdown")
+    print("✓ Shutdown complete")
 
 # Initialize FastAPI app
 app = FastAPI(
     title="RAG API with Groq and Qdrant",
-    description="OpenAI-compatible API for RAG using Groq LLM and Qdrant vector database",
+    description="OpenAI-compatible API for RAG using Groq and Qdrant",
     version="1.0.0",
     lifespan=lifespan
 )
@@ -192,8 +244,6 @@ class EmbeddingService:
                 "error": str(e)
             }
 
-embedding_service = EmbeddingService()
-
 class RAGService:
     """Service for retrieval-augmented generation"""
     
@@ -436,6 +486,7 @@ async def stream_chat_completion(messages: List[Dict], request: ChatCompletionRe
         }
         yield f"data: {json.dumps(error_chunk)}\n\n"
 
+# Additional endpoints for managing the vector database
 @app.post("/v1/embeddings/add")
 async def add_document(content: str, metadata: Optional[Dict] = None):
     """Add a document to the vector database"""