import os import time from dotenv import load_dotenv from langchain_chroma import Chroma from langchain_google_genai import GoogleGenerativeAIEmbeddings load_dotenv() # 1. Initialize with explicit logging print(" Initializing Embedding Model...") embeddings = GoogleGenerativeAIEmbeddings(model="models/gemini-embedding-001") CHROMA_PATH = "./chroma_db" if os.path.exists(CHROMA_PATH): print(f" Found {CHROMA_PATH}. Connecting...") vectorstore = Chroma( persist_directory=CHROMA_PATH, embedding_function=embeddings ) # 2. Test a very simple query to minimize token usage test_query = "protein" print(f" Sending query '{test_query}' to Google API for embedding...") start_time = time.time() try: # We use search_with_score to see the mathematical distance (the "Hard STEM" metric) results = vectorstore.similarity_search_with_score(test_query, k=1) elapsed = time.time() - start_time if results: doc, score = results[0] print(f"✅ Success! Latency: {elapsed:.2f}s | Distance Score: {score:.4f}") print(f"\n--- Top Result Preview ---\n{doc.page_content[:300]}...") else: print(" API returned successfully but found 0 matches in your DB.") except Exception as e: print(f" API Call Failed after {time.time() - start_time:.2f}s") print(f"Error Details: {e}") else: print(" Directory './chroma_db' does not exist.")