Update app.py

app.py

@@ -1,23 +1,33 @@
 import gradio as gr
-import zipfile
 import sqlite3
 import os
+import requests
 
-def get_connection_from_zip(zip_path: str, db_filename: str) -> sqlite3.Connection:
-    # Extract DB file from ZIP to disk as "temp.db"
-    with zipfile.ZipFile(zip_path) as z:
-        with z.open(db_filename) as f:
-            with open("temp.db", "wb") as out_file:
-                out_file.write(f.read())
+HF_DB_URL = "https://huggingface.co/datasets/babilonczyk/swiss_prot/resolve/main/proteins.db"
+LOCAL_DB_PATH = "proteins.db"
 
-    # Open SQLite connection (read-only)
-    return sqlite3.connect("temp.db", check_same_thread=False)
+# Download the DB only if not already cached
+def download_db(url: str, local_path: str):
+    if not os.path.exists(local_path):
+        print(f"⬇️ Downloading database from {url}...")
+        r = requests.get(url)
+        r.raise_for_status()
+        with open(local_path, "wb") as f:
+            f.write(r.content)
+        print("✅ Download complete.")
+    else:
+        print("🟢 Database already exists locally.")
 
-# Set up DB
-conn = get_connection_from_zip("proteins.db.zip", "proteins.db")
+# Connect to SQLite
+def get_connection(db_path: str) -> sqlite3.Connection:
+    return sqlite3.connect(db_path, check_same_thread=False)
+
+# Setup
+download_db(HF_DB_URL, LOCAL_DB_PATH)
+conn = get_connection(LOCAL_DB_PATH)
 cursor = conn.cursor()
 
-# Query logic
+# Query function
 def search_by_sequence(seq: str) -> str:
     seq = seq.strip().upper()
     cursor.execute("SELECT * FROM proteins WHERE sequence = ?", (seq,))