clean final: HF Inference API embeddings (no disk)

rag.py

@@ -1,4 +1,3 @@
-# rag.py  –- bullet-proof: online fetch with fallback on any error
 from __future__ import annotations
 import os, re, json, requests
 from functools import lru_cache
@@ -6,7 +5,7 @@ from typing import List, Tuple
 
 from langchain.text_splitter import RecursiveCharacterTextSplitter
 from langchain_community.vectorstores import FAISS
-from langchain_huggingface import HuggingFaceEndpoint
+from langchain_huggingface import HuggingFaceEndpoint, HuggingFaceInferenceAPIEmbeddings  # ✅ FIXED: added missing import
 from langchain_core.prompts import PromptTemplate
 from langchain.chains import RetrievalQA
 from supabase import create_client
@@ -96,8 +95,6 @@ def get_texts() -> List[str]:
 def get_vectorstore() -> FAISS:
     texts = get_texts()
 
-    # HF Inference Providers – zero disk, zero cache
-    from langchain_huggingface import HuggingFaceInferenceAPIEmbeddings
     embeddings = HuggingFaceInferenceAPIEmbeddings(
         api_key=HF_TOKEN,
         model_name="sentence-transformers/all-MiniLM-L6-v2"