fix: truncate question input to 500 characters for embedding generation

services/embedding.py

@@ -14,7 +14,7 @@ def generate_embeddings(chunks):
 
 def embed_question(question):
 	model = get_model()
-	bge_query = "Represent this sentence for searching relevant passages: " + question
+	bge_query = "Represent this sentence for searching relevant passages: " + question[:500]  # Truncate to 500 chars to fit model limits
 	embedding = model.encode([bge_query])
 	return embedding
 

services/rate_limiter.py

@@ -6,17 +6,14 @@ from slowapi.errors import RateLimitExceeded
 def get_real_ip(request: Request) -> str:
     # check for the real physical IP forwarded by Cloudflare
     cf_ip = request.headers.get("CF-Connecting-IP")
-    print(f"[DEBUG Limiter] Caught CF-Connecting-IP: {cf_ip}", flush=True)
     if cf_ip:
         return cf_ip
     
     # fallback to standard forwarded header or raw host
     forwarded = request.headers.get("X-Forwarded-For")
-    print(f"[DEBUG Limiter] Caught X-Forwarded-For: {forwarded}", flush=True)
     if forwarded:
         return forwarded.split(",")[0].strip()
 
-    print(f"[DEBUG Limiter] Caught Fallback IP: {request.client.host}", flush=True)        
     return request.client.host
 
 # Initialize the limiter with your custom IP fetcher