hive瑙ｆ瀽json鏁扮粍(hive瑙ｆ瀽json鏁版嵁)

一、 JSON对象，数组

JSON是在开发中经常会遇到的一种轻量级的数据交换格式比xml效率要高，XML需要很多的标签，这占据了网络流量。

JSON可以有两种格式：

一种是对象格式

{
  "current/confirm/iedCount": 249608,
  "/confirm/iedCount": 405399,
  "suspectedCount": 16432,
  "curedCount": 146023,
  "deadCount": 9768
}

另一种是数组对象格式

{
  "current/confirm/iedCount": 249608,
  "/confirm/iedCount": 405399,
  "suspectedCount": 16432,
  "curedCount": 146023,
  "deadCount": 9768
}

二、 JSON数据的处理方式

在数仓中我们可以通过sql来处理json数据，主要通过get_json_object和json_tuple来处理。下面就举下这两个函数的使用案例。

get_json_object()

SELECt  get_json_object('{"current/confirm/iedCount": 249608,"/confirm/iedCount": 405399,"suspectedCount": 16432,"curedCount": 146023, "deadCount": 9768}'
            ,"$.current/confirm/iedCount")
--  结果为 249608

SELECT  get_json_object('[
{"imgUrl": "https://img1.dxycdn.com/2020/0324/278/3403801351376518263-135.png","title": "新增疑似/新增确诊"}
,{"imgUrl": "https://img1.dxycdn.com/2020/0324/812/3403847865872006755-135.png","title": "境外输入"}
,{"imgUrl": "https://img1.dxycdn.com/2020/0324/547/3403847848692336947-135.png","title": "现存确诊"}
]'
            ,"$.[1,2].title")
--  结果为 "境外输入","现存确诊"

json_tuple()

SELECT json_tuple('{"age":18, "name": "lili", "gender": "female"}', "age", "name", "gender");
--  结果为 18, "lili", "female"

--  也可以配合hive的侧视图使用
SELECT   *
FROM table1
LATERAL VIEW json_tuple('{"age":18, "name": "lili", "gender": "female"}', "age", "name", "gender") state_json 
                        AS age,name,gender