restore deleted lines

utils/bm25.py

@@ -357,7 +357,7 @@ def build_and_save(
 
 # ── load ──────────────────────────────────────────────────────────────────────
 
-def load(index_path: str | Path = "data/processed/bm25_index.pkl") -> BM25Retriever:
+def load(index_path: str | Path = "data/processed/tokenisation/bm25_index.pkl") -> BM25Retriever:
     """
     Load a previously saved BM25Retriever from disk.
     Call this in app.py instead of rebuilding every time.
@@ -368,8 +368,20 @@ def load(index_path: str | Path = "data/processed/bm25_index.pkl") -> BM25Retrie
             f"BM25 index not found at '{index_path}'.\n"
             "Run build_and_save() from your notebook first."
         )
+    # Patch: pickle saved simple_tokenize under 'utils' top-level namespace,
+    # but it now lives in utils.bm25 — register it where pickle expects it
+    import sys
+    import types
+    from utils import bm25 as bm25_module
+
+    if "utils" not in sys.modules or not hasattr(sys.modules["utils"], "simple_tokenize"):
+        fake_utils = types.ModuleType("utils")
+        fake_utils.simple_tokenize = bm25_module.simple_tokenize
+        sys.modules["utils"] = fake_utils
+
     with open(index_path, "rb") as f:
         retriever = pickle.load(f)
+
     print(f"BM25 index loaded ← {index_path}")
     return retriever