通过 OPENJSON 使用 JSON 并保存在表中 - SQL Server 2016

Question

我们最近将后端数据库从 SQL Server 2012 升级到了 SQL Server 2016，我意识到它支持 JSON 对象。我们的网络服务之一以以下格式返回数据，我正在尝试使用 OPENJSON 函数直接使用它。

{
"RESULT_1": {
    "columns": ["col1", "col2", "col3", "col4"],
    "data": [
        ["0", null, "12345", "other"],
        ["1", "a", "54321", "MA"],
        ["0", null, "76543", "RI"]
    ]
   }
}

我想确保从“列部分”读取列名，并确保在 SQL Server 2016 中读取和推送正确的数据。

DECLARE @Json_Array2 nvarchar(max) = '{
"RESULT_1": {
    "columns": ["col1", "col2", "col3", "col4"],
    "data": [
        ["0", null, "12345", "other"],
        ["1", "a", "54321", "MA"],
        ["0", null, "76543", "RI"]
    ]
} }';

SELECT [key], value
FROM OPENJSON(@Json_Array2,'$.RESULT_1.columns')

SELECT [key], value
FROM OPENJSON(@Json_Array2,'$.RESULT_1.data')

通过上述语句，我能够提取列和数据数组。是否可以将这些数据转储到具有相同列名的平面表（已经存在）中？

我可以通过以下方式查看某一特定行的所有值：

SELECT [key], value
FROM OPENJSON(@Json_Array2,'$.RESULT_1.data[0]')


key value
0   0
1   NULL
2   12345
3   other

感谢任何指针。

谢谢

Answer 1

如果我理解正确，您正在尝试从“数据”数组中提取所有元素

Col1    Col2    Col3    Col4
0      NULL    12345    other
1      a       54321    MA
0      NULL    76543    RI

假设您的 Json 结构不会改变，您可以使用以下查询来实现：

SET NOCOUNT ON 
IF OBJECT_ID ('tempdb..##ToPvt') IS NOT NULL DROP TABLE ##ToPvt
DECLARE @Cmd NVARCHAR(MAX)=''
DECLARE @Table TABLE (Col1 nvarchar(100), Col2 nvarchar(100), Col3 nvarchar(100) , Col4 nvarchar(100))
DECLARE @Json_Array2 nvarchar(max) = '{
"RESULT_1": {
    "columns": ["col1", "col2", "col3", "col4"],
    "data": [
        ["0", null, "12345", "other"],
        ["1", "a", "54321", "MA"],
        ["0", null, "76543", "RI"]
    ]
} }';

;WITH Cols ([Key], [Value]) as
(
SELECT [key], value
FROM OPENJSON(@Json_Array2,'$.RESULT_1.columns')
)

, [Rows] as
(
SELECT ROW_NUMBER () OVER (ORDER BY [Key]) Seq, [key], value
FROM OPENJSON(@Json_Array2,'$.RESULT_1.data')
) 
,ToPvt as 
(
SELECT c.[Key]+1 Cols, x.Value , 'col'+CONVERT(VARCHAR(10),ROW_NUMBER () OVER (PARTITION BY c.Value ORDER BY t.[Key])) Seq
 FROM [Rows] t
 INNER JOIN Cols C
 ON t.[key] = c.[key]
 CROSS APPLY OPENJSON(t.value) X
 )

 SELECT * 
 INTO ##ToPvt
 FROM ToPvt 

 ;WITH Final (Cmd) as
 (
 SELECT DISTINCT 'SELECT [col1], [col2], [col3],[col4] FROM ##ToPvt 
            PIVOT
                (
                MAX([Value]) FOR Seq IN ([col1], [col2], [col3],[col4])
                ) T
             WHERE Cols = '+CONVERT(VARCHAR(10),Cols)+'
                ;'
 FROM ##ToPvt
 )

 SELECT @Cmd += Cmd
 FROM Final 

 INSERT INTO @Table 
 EXEC sp_executesql @Cmd

 SELECT * 
 FROM @Table