etl.py 文件源码

def remove_duplicate_connections(self):
        query = """
            DELETE FROM connection
            WHERE exists(SELECT 1
              FROM connection t2
                 WHERE t2.arrivaltime = connection.arrivaltime AND
                       t2.arrivalstop = connection.arrivalstop AND
                       t2.departuretime = connection.departuretime AND
                       t2.departurestop = connection.departurestop AND
                       t2.ctid > connection.ctid);
        """

        with connect(self.connection_string) as con:
            con.set_isolation_level(ISOLATION_LEVEL_AUTOCOMMIT)
            with con.cursor() as cur:
                cur.execute(query)
                cur.close()