Update download_assets.py

download_assets.py

@@ -3,16 +3,15 @@ import os
 
 def download_assets():
     """Download necessary assets from Hugging Face Hub."""
-    # ensure local dirs exist
+    # Prepare your local folders
     os.makedirs("data", exist_ok=True)
     os.makedirs("vectorstore/db_faiss", exist_ok=True)
 
     repo_id   = "MoizK/mindmedic-assets"
     repo_type = "dataset"
-    # 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
+    # -- PDFs (these live at the repo root) --
     pdf_files = [
         "71763-gale-encyclopedia-of-medicine.-vol.-1.-2nd-ed.pdf",
         "Depression-NIM-2024.pdf",
@@ -28,29 +27,29 @@ def download_assets():
             path = hf_hub_download(
                 repo_id=repo_id,
                 repo_type=repo_type,
-                filename=fname,
-                local_dir="data",
-                token=token,
+                filename=fname,     # root-level file
+                local_dir="data",   # save under ./data/
+                use_auth_token=token,
             )
-            print(f"✅ Downloaded {fname} → {path}")
+            print(f"✅ Downloaded PDF {fname} → {path}")
         except Exception as e:
-            print(f"⚠️ Failed to download {fname}: {e}")
+            print(f"⚠️ Failed to download PDF {fname}: {e}")
 
-    # Now fetch the FAISS index files from the nested path
+    # -- FAISS index files (nested under vectorstore/db_faiss in the repo) --
     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=remote_path,  # this is where they actually live in the repo
-                local_dir=".",         # so that ./vectorstore/db_faiss/... is created
-                token=token,
+                subfolder="vectorstore/db_faiss",  # look in this folder in the repo
+                filename=fname,                   # just the filename here
+                local_dir="vectorstore/db_faiss", # save under ./vectorstore/db_faiss/
+                use_auth_token=token,
             )
-            print(f"✅ Downloaded {remote_path} → {path}")
+            print(f"✅ Downloaded index file {fname} → {path}")
         except Exception as e:
-            print(f"⚠️ Failed to download {remote_path}: {e}")
+            print(f"⚠️ Failed to download index file {fname}: {e}")
 
 if __name__ == "__main__":
     download_assets()