feat: add versions of files without duplicates

app.py

@@ -36,7 +36,6 @@ if uploaded_file is not None and new_website_id:
 
     # Create DataFrame for website_event data
     df_website_event = df[website_event_columns]
-    df_website_event.to_csv('website_event.csv', index=False)
     st.download_button(label="Download Website Event CSV", data=df_website_event.to_csv(index=False), file_name='website_event.csv', mime='text/csv')
 
     # Define the columns for the session table
@@ -47,8 +46,18 @@ if uploaded_file is not None and new_website_id:
     ]
 
     # Create DataFrame for session data
-    df_session = df[session_columns].drop_duplicates('session_id')
-    df_session.to_csv('session.csv', index=False)
+    df_session = df[session_columns]
     st.download_button(label="Download Session CSV", data=df_session.to_csv(index=False), file_name='session.csv', mime='text/csv')
 
-    st.success("Successfully generated website_event.csv and session.csv")
+    # Versions of files without duplicates
+    df_website_event_nodup = df[website_event_columns].drop_duplicates('event_id')
+    df_session_nodup = df[session_columns].drop_duplicates('session_id')
+    st.text(f"Versions with duplicates removed ({len(df_website_event) - len(df_website_event_nodup)} events removed, {len(df_session) - len(df_session_nodup)} session entries removed):")
+
+    # Create DataFrame for website_event data without duplicates
+    st.download_button(label="Download Website Event CSV (No Duplicates)", data=df_website_event_nodup.to_csv(index=False), file_name='website_event_no_duplicates.csv', mime='text/csv')
+
+    # Create DataFrame for session data without duplicates
+    st.download_button(label="Download Session CSV (No Duplicates)", data=df_session_nodup.to_csv(index=False), file_name='session_no_duplicates.csv', mime='text/csv')
+    
+    st.success("Successfully generated csv files")