Update app.py

fixed issues with database file name duplication

app.py

@@ -63,44 +63,99 @@ if 'db' not in st.session_state:
 with st.sidebar:
     st.header("🗄️ Knowledge Base")
     
-    # 1. Upload Section
-    uploaded_files = st.file_uploader("Upload Policy Documents", accept_multiple_files=True, type=['pdf', 'docx', 'txt', 'csv', 'xlsx'])
+    # 1. Initialize the Uploader Key
+    if "uploader_key" not in st.session_state:
+        st.session_state.uploader_key = 0
+
+    # 2. Upload Section
+    uploaded_files = st.file_uploader(
+        "Upload Policy Documents", 
+        accept_multiple_files=True, 
+        type=['pdf', 'docx', 'txt', 'csv', 'xlsx'],
+        key=f"uploader_{st.session_state.uploader_key}" 
+    )
     
     if uploaded_files and st.button("Ingest Documents"):
         progress_bar = st.progress(0)
         status = st.empty()
         
+        # Get current files to check for duplicates
+        existing_files = st.session_state.db.get_all_filenames()
+        
         for i, f in enumerate(uploaded_files):
             status.text(f"Processing: {f.name}...")
             
-            # A. Parse File (handled by src/parsers.py)
+            # --- DUPLICATION CHECK ---
+            if f.name in existing_files:
+                st.toast(f"♻️ Updating existing file: {f.name}")
+                # Remove the old version first (Cleans SQL and FAISS)
+                st.session_state.db.delete_document(f.name)
+            # -------------------------
+
+            # A. Parse File
             text, filename, method = process_file(f)
             
             if "Error" in method:
                 st.error(f"Failed {filename}: {method}")
                 continue
                 
-            # B. Chunk & ID (handled by src/parsers.py)
+            # B. Chunk & ID
             chunks, doc_id = chunk_text(text, filename)
             
-            # C. Save to SQLite (handled by src/database.py)
-            # We explicitly store the full text for reliable RAG later
+            # C. Save to SQLite
             st.session_state.db.add_document(doc_id, filename, text)
             
-            # D. Add to Vector Index (handled by src/search.py)
-            # We only vector search the chunks, but they link back to doc_id
+            # D. Add to Vector Index
             st.session_state.search_engine.add_features(chunks)
             
             progress_bar.progress((i + 1) / len(uploaded_files))
             
         status.text("Syncing to Cloud...")
         SyncManager.push_data()
-        st.success(f"Successfully ingested {len(uploaded_files)} documents!")
-        time.sleep(2)
+        
+        st.success(f"Successfully processed {len(uploaded_files)} documents!")
+        time.sleep(1)
+        st.session_state.uploader_key += 1 
         st.rerun()
 
     st.divider()
     
+    # 3. Document Library (Better Visibility)
+    st.subheader("Manage Files")
+    
+    # We fetch the latest list
+    all_files = st.session_state.db.get_all_filenames()
+    
+    if all_files:
+        st.caption(f"📚 **Library: {len(all_files)} Documents**")
+        
+        # VISUAL LIST: A clean, scrollable list of what you have
+        with st.expander("View File List", expanded=False):
+            for f in all_files:
+                st.text(f"• {f}")
+
+        # DELETION MENU
+        file_to_del = st.selectbox("Select file to delete:", [""] + all_files)
+        if file_to_del and st.button("🗑️ Delete Selected"):
+            deleted_id = st.session_state.db.delete_document(file_to_del)
+            st.toast(f"Removed {file_to_del}")
+            SyncManager.push_data()
+            time.sleep(1)
+            st.rerun()
+            
+        # DANGER ZONE: Nuke Everything
+        if st.button("⚠️ Reset Database (Delete All)", type="primary"):
+            for f in all_files:
+                st.session_state.db.delete_document(f)
+            # We also wipe the index explicitly to be safe
+            st.session_state.search_engine.reset_index()
+            SyncManager.push_data()
+            st.success("Database wiped clean.")
+            time.sleep(1)
+            st.rerun()
+    else:
+        st.info("Library is empty.")
+    
     # 2. Management Section
     st.subheader("Manage Files")
     all_files = st.session_state.db.get_all_filenames()
@@ -108,10 +163,7 @@ with st.sidebar:
         st.caption(f"Total Documents: {len(all_files)}")
         file_to_del = st.selectbox("Delete File:", [""] + all_files)
         if file_to_del and st.button("🗑️ Remove Document"):
-            # Delete from SQL
             deleted_id = st.session_state.db.delete_document(file_to_del)
-            # Note: FAISS deletion is hard, usually we just rebuild index.
-            # For now, we accept the "Ghost" vectors in FAISS until a full rebuild.
             st.toast(f"Removed {file_to_del} from Database.")
             SyncManager.push_data()
             time.sleep(1)
@@ -142,7 +194,7 @@ if query:
             st.caption(f"Analyzing primary source: {top_match['source']}")
             
             if st.button("✨ Generate Assessment"):
-                with st.spinner("Consulting Granite Model..."):
+                with st.spinner("Sending Data to Selected Model..."):
                     # Call our separated LLM client
                     response = ask_granite(query, full_doc_text)