Update download_assets.py

download_assets.py

@@ -12,7 +12,7 @@ def download_assets():
     # allow either env var name
     token     = os.getenv("HUGGINGFACE_HUB_TOKEN") or os.getenv("HUGGINGFACE_API_TOKEN")
 
-    # list of PDF filenames *at the repo root*
+    # list of PDF filenames at the repo root
     pdf_files = [
         "71763-gale-encyclopedia-of-medicine.-vol.-1.-2nd-ed.pdf",
         "Depression-NIM-2024.pdf",
@@ -28,28 +28,29 @@ def download_assets():
             path = hf_hub_download(
                 repo_id=repo_id,
                 repo_type=repo_type,
-                filename=fname,            # no "data/" prefix
-                local_dir="data",          # drop into ./data/
+                filename=fname,
+                local_dir="data",
                 token=token,
             )
             print(f"✅ Downloaded {fname} → {path}")
         except Exception as e:
             print(f"⚠️ Failed to download {fname}: {e}")
 
-    # FAISS index files also at the repo root
+    # Now fetch the FAISS index files from the nested path
     index_files = ["index.faiss", "index.pkl"]
     for fname in index_files:
+        remote_path = f"vectorstore/db_faiss/{fname}"
         try:
             path = hf_hub_download(
                 repo_id=repo_id,
                 repo_type=repo_type,
-                filename=fname,                   # root filename
-                local_dir="vectorstore/db_faiss", # into ./vectorstore/db_faiss/
+                filename=remote_path,  # this is where they actually live in the repo
+                local_dir=".",         # so that ./vectorstore/db_faiss/... is created
                 token=token,
             )
-            print(f"✅ Downloaded {fname} → {path}")
+            print(f"✅ Downloaded {remote_path} → {path}")
         except Exception as e:
-            print(f"⚠️ Failed to download {fname}: {e}")
+            print(f"⚠️ Failed to download {remote_path}: {e}")
 
 if __name__ == "__main__":
-    download_assets()
+    download_assets()