# Flink Opensource SQL如何解析复杂嵌套 JSON？
- **kafka message**
  ```
  {
    "id": 1234567890,
    "name": "swq",
    "date": "1997-04-25",
    "obj": {
      "time1": "12:12:12",
      "str": "test",
      "lg": 1122334455
    },
    "arr": [
      "ly",
      "zpk",
      "swq",
      "zjy"
    ],
    "rowinarr": [
      {
        "f1": "f11",
        "f2": 111
      },
      {
        "f1": "f12",
        "f2": 222
      }
    ],
    "time": "13:13:13",
    "timestamp": "1997-04-25 14:14:14",
    "map": {
      "flink": 123
    },
    "mapinmap": {
      "inner_map": {
        "key": 234
      }
    }
  }
  ```
  
- **flink opensource sql**
  ```
  create table kafkaSource(
    id            BIGINT,
    name          STRING,
    `date`        DATE,
    obj           ROW<time1 TIME,str STRING,lg BIGINT>,
    arr           ARRAY<STRING>,
    rowinarr      ARRAY<ROW<f1 STRING,f2 INT>>,
    `time`        TIME,
    `timestamp`   TIMESTAMP(3),
    `map`         MAP<STRING,BIGINT>,
    mapinmap      MAP<STRING,MAP<STRING,INT>>
  ) with (
    'connector' = 'kafka',
    'topic' = 'topic-swq-3',
    'properties.bootstrap.servers' = '10.128.0.138:9092,10.128.0.119:9092,10.128.0.212:9092',
    'properties.group.id' = 'swq-test',
    'scan.startup.mode' = 'latest-offset',
    'format' = 'json'
  );
  create table printSink (
    id            BIGINT,
    name          STRING,
    `date`        DATE,
    str           STRING,
    arr           ARRAY<STRING>,
    nameinarray   STRING,
    rowinarr      ARRAY<ROW<f1 STRING,f2 INT>>,
    f2            INT,
    `time`        TIME,
    `timestamp`   TIMESTAMP(3),
    `map`         MAP<STRING,BIGINT>,
    flink         BIGINT,
    mapinmap      MAP<STRING,MAP<STRING,INT>>,
    `key`         INT
  ) with ('connector' = 'print');
   
  insert into
    printSink
  select
    id,
    name,
    `date`,
    obj.str,
    arr,
    arr[4],
    rowinarr,
    rowinarr[1].f2,
    `time`,
    `timestamp`,
    `map`,
    `map`['flink'],
    mapinmap,
    mapinmap['inner_map']['key']
  from kafkaSource;
  ```
  
- **result**
  ```
  +I(1234567890,swq,1997-04-25,test,[ly, zpk, swq, zjy],zjy,[f11,111, f12,222],111,13:13:13,1997-04-25T14:14:14,{flink=123},123,{inner_map={key=234}},234)
  ```
  
![](https://support.huaweicloud.com/dli_faq/public_sys-resources/caution_3.0-zh-cn.png)
1. 各数据类型获取元素的方法： - map：map\['key'\]
   - array：array\[index\]
   - row：row.key
   
2. array 的起始下标从 1 开始，即 array\[1\] 是 array 的第一个元素。
3. array 的元素必须同类型，row 的元素可以不同类型。