flinkSql将嵌套Json映射成source表

flinkSql在遇到复杂的json时,如何映射成表呢? 这里推荐一种方便的方式:
嵌套的json格式如下:

{
    "data":{
        "data":{
            "mac_value":0,
            "ad_name":2056,
            "voice":75,
            "dataTimeStamp":1598522106830
        },
        "type":3,
        "deviceId":"001C92F7DCd85A"
    },
    "timestamp":1598522106835,
    "type":1
}

期望映射成一个source表

CREATE TABLE t_1144_1 (
   -- 最主要的是这行,定义类型为ROW
   `data` ROW(deviceId string,`data` ROW(mac_value string,ad_name string,voice string,dataTimeStamp string)),
   -- 通过xx.xx取数据,映射成自己想要的表
    deviceId as cast(`data`.deviceId as varchar) ,
    mac_value as cast( `data`.`data`.mac_value as integer),
    ad_name as cast( `data`.`data`.ad_name as integer), 
    voice as cast( `data`.`data`.voice as integer), 
    eventTime  as cast( `data`.`data`.`dataTimeStamp` as varchar), 
     -- 时间窗口以及水位线
    windowEventTime AS TO_TIMESTAMP(FROM_UNIXTIME((cast(cast( `data`.`data`.`dataTimeStamp` as varchar) as bigint)+43200000)/1000)), 
    WATERMARK FOR windowEventTime AS windowEventTime - INTERVAL '2' SECOND 
) WITH (
     xxxx
)
 后续的sql,就可以基于这个表来进行操作了.

  通过使用ROW的方式, 可以让复杂的json转变为可操作的schema,使用时, 可以让xx.xx.xx来使用.