如何使用SQL Server解析嵌套的JSON数组

Question

我目前能够使用SQL Server的"OPENJSON WITH(..."语法解析大部分JSON文件.但是,这个特定的文件包含我不知道如何处理的嵌套数组.

我读过的许多例子都将JSON引用为变量.在这种情况下,我正在调用一个文件:

select DEV_JSON.*
from OPENROWSET
(BULK 'C:\Users\Myuser\Documents\JSON_extract.json', SINGLE_CLOB) as my_datafile
 CROSS APPLY OPENJSON(BulkColumn) 
WITH
  (DOC_ID           varchar(100)  '$.doc._id',
   DOC_REV          varchar(45)   '$.doc._rev',
   DELY_APPL_NAME   varchar(20)   '$.doc.delivery.application',
   DELY_SENT_BY     varchar(25)   '$.doc.delivery.sender.id',
   DELY_SENT_TYPO   varchar(20)   '$.doc.delivery.sender.type',
   .....
   ....
   ...
   ..) as DEV_JSON

其中一个属性包含嵌套数组.下面我复制了我的JSON的前5个属性,以及嵌套的"recipients"数组.

如何构建我的SQL来解析这一部分？

"doc": {
    "_id": "049d4e4030afcdeefedaa90f640f91d4a2be93d7-bd_abcxyz@somemail.com",
    "_rev": "3-e119db13dae8d50ae0c4579ba9c87fc9",
    "delivery": {
        "application": "App_XYZ",
        "sender": {
            "id": "MABarrera@yahoo.com",
            "type": "user"
        },
        "recipients": [{
                "type": "email",
                "recipient": "\"Artzer, Daniel J\" <DJArtzer@emailaddr.com>",
                "sentTS": "2017-10-18T13:04:00.133Z"
            },
            {
                "type": "email",
                "recipient": "\"Higgins, Laura L\" <LLHiggins@emailaddr.com>",
                "sentTS": "2017-10-18T13:04:00.133Z"
            },
            {
                "type": "email",
                "recipient": "\"Friedman, Brian\" <BFriedman@emailaddr.com>",
                "sentTS": "2017-10-18T13:04:00.133Z"
            },
            {
                "type": "email",
                "recipient": "\"Garcia, Charlie M\" <CMGarcia@emailaddr.com>",
                "sentTS": "2017-10-18T13:04:00.133Z"
            }
        ]
    },

Answer 1

我遇到了同样的问题,最后我用多个CROSS APPLY子句来解决它.

这是我的JSON的一个例子:

DECLARE @PermsJSON NVARCHAR(MAX) =
N'[{
    "AppId": 1,
    "Perms":
    [{

        "Permission": ["AA", "BB"],
        "PermissionTypeID": 2
    },
    {
        "Permission": ["10"],
        "PermissionTypeID": 1
    }]
},
{
    "AppId": 2,
    "Perms":
    [{

        "Permission": ["IM", "NM"],
        "PermissionTypeID": 2
    },
    {
        "Permission": ["42"],
        "PermissionTypeID": 1
    }]
}]';

然后我可以使用以下查询解析它:

SELECT
    a.AppId
    ,[Permission] = c.Value
    ,b.PermissionTypeID
FROM
    OPENJSON(@PermsJSON)
    WITH
        (
            AppId INT N'$.AppId'
            ,Perms NVARCHAR(MAX) AS JSON
        ) AS a
CROSS APPLY
    OPENJSON(a.Perms)
    WITH
        (
            PermissionTypeID INT
            ,[Permission] NVARCHAR(MAX) AS JSON
        ) AS b
CROSS APPLY OPENJSON(b.Permission) AS c;

结果如下:

AppId   Permission  PermissionTypeID
1       AA          2
1       BB          2
1       10          1
2       IM          2
2       NM          2
2       42          1

Answer 2

之后MUCH搜索我终于发现了解决这个问题的。我只需将嵌套数组作为另一个 JSON 列包含在我的查询中，例如：

WITH
 (DOC_ID                varchar(100)  '$.doc._id',
  DOC_REV               varchar(45)   '$.doc._rev',
  DELY_APPL_NAME        varchar(20)   '$.doc.delivery.application',
  DELY_SENT_BY          varchar(25)   '$.doc.delivery.sender.id',
  DELY_SENT_TYPO        varchar(20)   '$.doc.delivery.sender.type',
  RECIPS                nvarchar(max) '$.doc.delivery.recipients' as JSON,
  PAYLOAD_START_TIME    varchar(30)   '$.doc.payload.startTS',
  ....
  ...
  ..
  ) as my_query

因此，我最终得到每个 JSON 文档的一条记录，（在本例中）包含一个包含 JSON 文本的 varchar 列。

接下来，我可以在此列上运行单独的查询来解析 JSON 并创建与父级关联的“子表”。