Fix: add repo_type=dataset and force_download=True for public dataset

chatbot_retriever.py

@@ -100,13 +100,19 @@ def ensure_data_dir():
     for f in files:
         local_path = os.path.join(data_dir, f.replace("/", "_"))
         if not os.path.exists(local_path):
-            print(f"📥 Downloading {f} from Hugging Face...")
-            downloaded = hf_hub_download(repo_id=DATASET_REPO, filename=f)
+            print(f"📥 Downloading {f} from Hugging Face (public dataset)...")
+            downloaded = hf_hub_download(
+                repo_id=DATASET_REPO,
+                filename=f,
+                repo_type="dataset",       # ✅ tells HF it's a dataset
+                force_download=True        # ✅ bypass any bad cached 401s
+            )
             os.rename(downloaded, local_path)
         local_paths.append(local_path)
     return local_paths
 
 
+
 def detect_subject(fname: str) -> Optional[str]:
     # light heuristic to guess subject code from filename
     t = (fname or "").lower()
@@ -191,7 +197,7 @@ def load_all_docs(base_dir: str = DATA_DIR) -> List:
 def build_or_load_indexes(force_reindex: bool = False):
     if os.getenv("FORCE_REINDEX", "0").lower() in ("1", "true", "yes"):
         force_reindex = True
-        
+
     ensure_data_dir()
     docs = load_all_docs(DATA_DIR)
     if not docs: