Update rag.py

rag.py

@@ -2,7 +2,7 @@ import os
 from typing import List, Dict, Tuple
 import numpy as np
 from openai import OpenAI
-from services.faq_store import FAQ_ENTRIES, FAQ_VECS
+from faq_store import FAQ_ENTRIES, FAQ_VECS
 
 RAG_CONFIDENCE_THRESHOLD = 0.6
 MAX_FAQ_MATCHES = 3
@@ -23,7 +23,6 @@ FALLBACK_MESSAGE = (
 
 client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
 
-# In-memory embedding cache for repeated queries
 _query_cache: dict[str, np.ndarray] = {}
 
 
@@ -32,7 +31,6 @@ def cosine(a: np.ndarray, b: np.ndarray) -> float:
 
 
 async def get_answer(question: str, history: List[Dict]) -> Tuple[str, List[Dict]]:
-    # Embed query (with in-memory cache)
     if question in _query_cache:
         vec = _query_cache[question]
     else:
@@ -40,21 +38,17 @@ async def get_answer(question: str, history: List[Dict]) -> Tuple[str, List[Dict
         vec = np.array(resp.data[0].embedding, dtype=np.float32)
         _query_cache[question] = vec
 
-    # Cosine similarity against all FAQ vectors
     scores = [(fid, cosine(vec, fvec)) for fid, fvec in FAQ_VECS]
     scores.sort(key=lambda x: x[1], reverse=True)
 
-    # Fallback if no FAQ meets the confidence threshold
     if not scores or scores[0][1] < RAG_CONFIDENCE_THRESHOLD:
         return FALLBACK_MESSAGE, []
 
-    # Gather top matches
     matches = []
     for fid, score in scores[:MAX_FAQ_MATCHES]:
         faq = FAQ_ENTRIES[fid]
         matches.append({"id": fid, "answer": faq["answer"], "source": faq["source"], "score": score})
 
-    # Build message list for GPT
     messages: List[Dict] = [{"role": "system", "content": SYSTEM_PROMPT}]
     for msg in history:
         messages.append({"role": msg["role"], "content": msg["content"]})