I’m trying to insert several Pandas DataFrames to corresponding PostgreSQL tables, each with a different Primary Key. I’ve attempted to implement the following method according to the documentation:

def insert_on_conflict_nothing(table, conn, keys, data_iter):
    # tableIndex is the primary key in "conflict_table"
    data = [dict(zip(keys, row)) for row in data_iter]
    pkey = [pk_column.name for pk_column in table.primary_key.columns.values()]
    stmt = insert(table.table).values(data).on_conflict_do_nothing(index_elements=pkey)
    result = conn.execute(stmt)
    return result.rowcount

and then pass it on in a function:

def insert_workouts(workouts_df):
    engine = create_engine('postgresql+psycopg2://' + db_user + ':' + db_pass + '@' + db_host + '/' + db_name)
    workouts_table_columns = pd.read_sql('SELECT * FROM fitness.workouts WHERE 1=2', engine).columns.to_list()
    workouts_df.columns = workouts_table_columns
    workouts_df.to_sql(
        name='workouts',
        schema='fitness',
        con=engine,
        if_exists='append',
        index=False,
        method=insert_on_conflict_nothing
    )

This, however, results in the following error: AttributeError: 'SQLTable' object has no attribute 'primary_key'.

Can anyone help me figure out a way to check the respective table’s primary key (in my case, it will always be one column, if that matters), so that I don’t have to hard-code it in multiple methods?

Thanks!