Validation For Master file

app.py

@@ -22,6 +22,19 @@ def to_excel(df):
     processed_data = output.getvalue()
     return processed_data
 
+def validate_columns(df):
+    required_columns = {
+        "Employer Number", "Employer Name", "Bank Statement Reference", "Employer Province",
+        "Region", "Industry", "Contributing Status", "Date Joined Plan", "Termination Date", "Email Addresses"
+    }
+    missing_columns = required_columns - set(df.columns)
+    
+    if missing_columns:
+        st.error(f"Upload a valid Master Sheet. These columns are missing: {', '.join(missing_columns)}")
+        return None
+    else:
+        return df
+
 def process_files(excel_file, text_file):
     """
     Process the uploaded Excel/CSV and text files and return cleaned dataframes.
@@ -41,6 +54,11 @@ def process_files(excel_file, text_file):
         df_excel = pd.read_csv(excel_file)
     else:
         df_excel = pd.read_excel(excel_file)
+
+    # Validate Master Sheet
+    df_excel = validate_columns(df_excel)
+    if df_excel is None:
+        return None, None
     
     # Ensure the 'Employer Number' column values are formatted as zero-padded 6-digit strings
     df_excel['Employer Number'] = [str(number).zfill(6) for number in df_excel['Employer Number']]
@@ -71,6 +89,8 @@ with st.sidebar:
 # Check if both files are uploaded
 if uploaded_excel and uploaded_text:
     master_data, df = process_files(uploaded_excel, uploaded_text)  # Process the files
+    if master_data is None and df is None:
+        st.stop()
 
     res, fbdm, fben, fbbm25, nf = get_res_df(master_data, df, threshold)  # Generate the result DataFrame using the helper function