动态 JSON 使用 T-SQL 解析

Question

我有一个高度非结构化的 JSON 字符串，作为调用 REST 的响应 API:

{
    "info": "Test Json Structure",
    "Owner": "Self-Owned",
    "managedObjects": [{
            "Name": "Device1",
            "Class": "A"
        }, {
            "Name": "Device2",
            "Class_145": "Ax01",
            "Class_11": "B",
            "Type_125478": {
                "Model": "1",
                "Manufacturer": "External"
            },
            "Type_SD": {
                "Model": "00",
                "Manufacturer": "Internal"
                }
        }, {
            "Name": "Device3",
            "Class_x": "Cx11",
            "Class_T": "C8Y",
            "Type": {
                "Model": "1x",
                "Manufacturer": "Internal"
            }
        }
    ]
}

如何使用 T-SQL 动态解析此对象，以便所有子元素代表 table 的列？更重要的是，如何处理 Type、Type_125478、Type_SD 对象，实际上它们具有相同的结构（Model, Manufacturer），但有些名称不同。还要记住，设备可能会发送一个新的标识符 (Type_XYZ)，它以前不存在，但幸运的是具有相同的结构 (Model, Manufacturer)。

Answer 1

您可以使用类似这样的方法将整批分解为表格结构并继续执行此操作（需要 SQL-Server 版本 v2016+）：

DECLARE @YourJSON NVARCHAR(MAX)=
N'{
    "info": "Test Json Structure",
    "Owner": "Self-Owned",
    "managedObjects": [{
            "Name": "Device1",
            "Class": "A"
        }, {
            "Name": "Device2",
            "Class_145": "Ax01",
            "Class_11": "B",
            "Type_125478": {
                "Model": "1",
                "Manufacturer": "External"
            },
            "Type_SD": {
                "Model": "00",
                "Manufacturer": "Internal"
                }
        }, {
            "Name": "Device3",
            "Class_x": "Cx11",
            "Class_T": "C8Y",
            "Type": {
                "Model": "1x",
                "Manufacturer": "Internal"
            }
        }
    ]
}';

--查询

SELECT A.info
      ,A.[Owner]
      ,C.[key] AS TagName
      ,CASE WHEN D.Model IS NULL THEN C.[value] END AS TagValue 
      ,D.Model
      ,D.Manufacturer
FROM OPENJSON(@YourJSON)
WITH(info NVARCHAR(MAX)
    ,[Owner] NVARCHAR(MAX)
    ,managedObjects NVARCHAR(MAX) AS JSON) A
OUTER APPLY OPENJSON(A.managedObjects) B
OUTER APPLY OPENJSON(B.[value]) C
OUTER APPLY OPENJSON(CASE WHEN ISJSON(C.[value])=1 THEN C.[value] END) 
WITH (Model NVARCHAR(MAX)
     ,Manufacturer NVARCHAR(MAX))D;

--结果

+---------------------+------------+-------------+----------+-------+--------------+
| info                | Owner      | TagName     | TagValue | Model | Manufacturer |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Name        | Device1  |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Class       | A        |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Name        | Device2  |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Class_145   | Ax01     |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Class_11    | B        |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Type_125478 |          | 1     | External     |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Type_SD     |          | 00    | Internal     |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Name        | Device3  |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Class_x     | Cx11     |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Class_T     | C8Y      |       |              |
+---------------------+------------+-------------+----------+-------+--------------+
| Test Json Structure | Self-Owned | Type        |          | 1x    | Internal     |
+---------------------+------------+-------------+----------+-------+--------------+

提示：您可以将 B.[key] 作为对象标识符添加到结果中。

简而言之

我们用第一个OPENJSON进入你的JSON。 WITH 子句允许将 JSON-props 作为列进行寻址。我们 return managedObejcts 作为 JSON 他们自己。
我们再使用一个 OPENJSON 来深入研究托管对象。
这将 return 一个对象数组。我们可以将 value 传递给另一个 OPENJSON.
每当 value 可以单独解释为 JSON 时，我们再使用一个 OPENJSON，这次再次使用 WITH 子句来获得内部道具作为列。

您可以将此结果插入 table（声明的、临时的、物理的...）并继续此 easy-to-read 集。

动态 JSON 使用 T-SQL 解析

Dynamic JSON Parsing using T-SQL

tsql

sql-server

json

sql-server-json

sql-server-2017