Update app.py

app.py

@@ -184,35 +184,35 @@ def save_to_csv_flat(all_extracted_values, schedule1_values, client_name="Unknow
         new_row["Rent/ Royalty (Schedule E) - Schedule 1, Line 5"] = schedule1_values[1] if schedule1_values[1] != '' else '0'
         new_row["Other Income - Schedule 1, Line 8"] = schedule1_values[2] if schedule1_values[2] != '' else '0'
 
-    # Check for duplicates - improved method
-    is_duplicate = False
-    
-    # Convert new_row to DataFrame for comparison
+    # Convert new_row to DataFrame
     new_row_df = pd.DataFrame([new_row])
-    
-    # Compare all columns except possibly the client name
-    comparison_cols = [col for col in flat_columns if col != "Client Name"]
-    
+
+    # Convert all values to strings for consistent comparison
+    df_str = df.astype(str)
+    new_row_str = new_row_df.astype(str)
+
+    # Check for duplicates by comparing all columns
+    is_duplicate = False
     if not df.empty:
-        # Check if any existing row matches the new data
-        for _, existing_row in df.iterrows():
-            match = True
-            for col in comparison_cols:
-                if str(existing_row[col]) != str(new_row[col]):
-                    match = False
-                    break
-            if match:
-                is_duplicate = True
-                print("Duplicate found - not adding new row")
-                break
+        # Merge check (efficient method)
+        merged = df_str.merge(new_row_str)
+        is_duplicate = not merged.empty
+
+        # Alternative method (row-by-row comparison)
+        # for _, row in df_str.iterrows():
+        #     if row.equals(new_row_str.iloc[0]):
+        #         is_duplicate = True
+        #         break
 
     # Append if not duplicate
     if not is_duplicate:
         df = pd.concat([df, new_row_df], ignore_index=True)
         df.to_csv(csv_path, index=False)
-        print(f"New data saved to CSV: {csv_path}")
+        print(f"✅ New data saved for client: {client_name}")
     else:
-        print("Duplicate data detected - no changes made to CSV")
+        print(f"⚠️ Duplicate data detected for client: {client_name} - no changes made")
+
+    return df
 
 # Gradio UI
 iface = gr.Interface(