将 MERGE 语句的更新列审计为不同表中的行
Posted
技术标签:
【中文标题】将 MERGE 语句的更新列审计为不同表中的行【英文标题】：Audit the Update columns of a MERGE statement as rows in a different table 【发布时间】：2020-09-28 13:48:19 【问题描述】：我正在努力实现以下目标
合并 TARGET 和 SOURCE 表。 使用输出子句审核更改。我可以使用 CASE 语句执行以下操作：
审计插入和删除很容易。我将插入和删除转换为以下格式：
| Action | Record | ChangedFrom | ChangedTo |
|----------|--------|-------------|-----------|
| Inserted | 1 | NULL | 1 |
| Deleted | 2 | 2 | NULL |
我无法获得：
| Action | Record | ChangedFrom | ChangedTo |
|---------|--------|--------------------|-------------------|
| UPDATED | 1 | ColA_Content_Before| ColA_Content_After|
Merge 语句将更改后的值输出为每行的列数。 我需要将这些列对更改为行，以便记录/行的每个更改的列都成为审计表中的一个条目。 同样在这里，为了示例，我使用了 3 列，实际上，当我不知道列数是多少时，我需要它来工作。
目标表
| Id | ColA | ColB | ColC |
|----|------|--------|--------|
| 1 | Orig | Orig_B | Orig_C |
源表
| Id | ColA | ColB | ColC |
|----|------------|------------|--------|
| 1 | Modified_A | Modified_B | Orig_C |
合并结果
| Id | ColA | ColB | ColC | Id | ColA | ColB | ColC |
|---:|------|--------|--------|----|-----------|------------|------|
| 1 | Orig | Orig_B | Orig_C | 1 | Modifed_A | Modified_C | |
我需要这些列作为行
| Action | Record | ChangedFrom | ChangedTo |
|--------:|--------|-------------|------------|
| Updated | 1 | Orig | Modified_A |
| Updated | 1 | Orig_B | MOdified_B |
我尝试过的方法：
在 proc 中使用 MERGE 来执行批处理语句 在这里，我尝试使用游标遍历行，然后在 case 语句中插入 cols。 使用动态 sql 并透视表。但是没有成功。
任何指针？
要复制的代码
CREATE TABLE [dbo].[Test_Source]
(
Id int NOT NULL,
ColA VARCHAR(50) NOT NULL,
ColB VARCHAR(50) NOT NULL,
ColC VARCHAR(50) NOT NULL,
ColD VARCHAR(50) NOT NULL,
[UDate] [DATETIME] NOT NULL
) ON [PRIMARY];
GO
CREATE TABLE [dbo].[Test_Target]
(
Id int NOT NULL,
ColA VARCHAR(50) NOT NULL,
ColB VARCHAR(50) NOT NULL,
ColC VARCHAR(50) NOT NULL,
ColD VARCHAR(50) NOT NULL,
[UDate] [DATETIME] NOT NULL
) ON [PRIMARY];
GO
---Insert some test values
INSERT INTO [dbo].[Test_Source]
(
Id,ColA,ColB,ColC,ColD,UDate
)
VALUES
(
1,'ColA','ColB','ColC','ColD',GETDATE()
);
INSERT INTO [dbo].[Test_Target]
(
Id,ColA,ColB,ColC,ColD,UDate
)
VALUES
(
1,'ColA_After','ColB_After','ColC_After','ColD_After',GETDATE()
);
CREATE PROCEDURE MERGE_TEST
As
BEGIN
DECLARE @MERGERESULTS TABLE
(
[Action] nvarchar(50),
Id_After int ,
ColA_After nvarchar(50),
ColB_After nvarchar(50),
ColC_After nvarchar(50),
ColD_After nvarchar(50),
UDate_After nvarchar(50),
Id_Before int,
ColA_Before nvarchar(50),
ColB_Before nvarchar(50),
ColC_Before nvarchar(50),
ColD_Before nvarchar(50),
UDate_Before nvarchar(50)
);
DECLARE @AUDITRESULTS TABLE
(
[Action] nvarchar(50),
Id_After int ,
ChangedFrom nvarchar(50),
ChangedTo nvarchar(50)
);
DECLARE
@cols AS NVARCHAR(MAX),
@query AS NVARCHAR(MAX)
MERGE Test_Target as T
USING Test_Source as S
ON T.Id = S.Id
WHEN MATCHED AND
T.ColA <> S.ColA OR
T.ColB <> S.ColB OR
T.ColC <> S.ColC OR
T.ColD <> S.ColD OR
T.UDate <> S.UDate
THEN UPDATE SET
T.ColA = S.ColA ,
T.ColB = S.ColB ,
T.ColC = S.ColC ,
T.ColD = S.ColD ,
T.UDate = S.UDate
WHEN NOT MATCHED BY TARGET
THEN INSERT
(Id,ColA,ColB,ColC,ColD,UDate)
VALUES
(S.Id,S.ColA,S.ColB,S.ColC,S.ColD,S.UDate)
WHEN NOT MATCHED BY SOURCE
THEN DELETE
OUTPUT $action,Inserted.*,Deleted.* into @MERGERESULTS ;
select * from @MERGERESULTS;
select * into #mytemp from @MERGERESULTS
-- Dynamic sql to pivot table
select @cols = STUFF((SELECT distinct ',' + QUOTENAME(ColA_After)
from @MERGERESULTS
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set @query = 'Select * from #mytemp
pivot
(
max(Id_After)
for ColA_After in (' + @cols + ')
) p '
select @Cols as 'Columns';
execute(@query)
-- Dynamic sql to pivot Table
-- Trying cursor and row count (incomplete)
--declare @Id int
--set rowcount 0
--select
--t.Action ,
--CASE WHEN (t.ColA_After <> t.ColA_Before)
-- THEN t.ColA_After
--END,
--CASE WHEN (t.ColA_After <> t.ColA_Before)
-- THEN t.ColA_Before
--END,
--CASE WHEN (t.ColB_After <> t.ColB_Before)
-- THEN t.ColB_After
--END,
--CASE WHEN (t.ColB_After <> t.ColB_Before)
-- THEN t.ColB_Before
--END,
--CASE WHEN (t.ColC_After <> t.ColC_Before)
-- THEN t.ColC_After
--END,
--CASE WHEN (t.ColC_After <> t.ColC_Before)
-- THEN t.ColC_Before
--END,
--CASE WHEN (t.ColD_After <> t.ColD_Before)
-- THEN t.ColD_After
--END,
--CASE WHEN (t.ColD_After <> t.ColD_Before)
-- THEN t.ColD_Before
--END
--from #mytemp t
--inner join #mytemp t1
--on t.Id_After = t1.Id_Before
--where
--t.ColA_After <> t1.ColA_Before OR
--t.ColB_After <> t1.ColB_Before OR
--t.ColC_After <> t1.ColC_Before OR
--t.ColD_After <> t1.ColD_Before
--set rowcount 1
--select @Id = Id_After from #mytemp
--while @@rowcount <> 0
--begin
-- set rowcount 0
-- (select * from #mytemp where Id_After = @Id)
-- delete #mytemp where au_id = @au_id
-- set rowcount 1
-- select @au_id = au_id from #mytemp
--end
--set rowcount 0
END
EXEC MERGE_TEST
这似乎是一个切线
【问题讨论】：
【参考方案1】：我通过使用 unpivot 实现了上述目标
不确定这是否是最好的方法，但可以完成工作。
我的解决方案：
使用过程并选择 MERGE 结果到 TABLE 变量中，列后缀为 _Before 和 _After。 使用 where 条件审核插入和更新。 使用 UNPIVOT 审核更新。 pivot col 值可用于获取更新后的列名。 UNPIVOT 需要所有 cols 具有相同的数据类型，因此可以对它们进行 CAST/CONVERT，我刚刚在选择时将所有列存储为 nvarchars()。创建审核脚本
SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO
CREATE TABLE [dbo].[Audit](
[Id] [int] IDENTITY(1,1) NOT NULL,
[ChangedFrom] [nvarchar](100) NULL,
[ChangedTo] [nvarchar](100) NULL,
[Action] [nvarchar](100) NULL,
[ChangedDate] [datetime2](7) NULL,
[Column] [nvarchar](100) NULL
CONSTRAINT [PK_Audit] PRIMARY KEY CLUSTERED
(
[Id] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON, OPTIMIZE_FOR_SEQUENTIAL_KEY = OFF) ON [PRIMARY]
) ON [PRIMARY]
GO
创建过程脚本
CREATE PROCEDURE MERGE_AUDIT
As
BEGIN
--Temporary TABLE variable to store the MERGE results
DECLARE @MERGERESULTS TABLE
(
[Action] nvarchar(50),
Id_After nvarchar(50),
ColA_After nvarchar(50),
ColB_After nvarchar(50),
ColC_After nvarchar(50),
ColD_After nvarchar(50),
UDate_After nvarchar(50),
Id_Before nvarchar(50),
ColA_Before nvarchar(50),
ColB_Before nvarchar(50),
ColC_Before nvarchar(50),
ColD_Before nvarchar(50),
UDate_Before nvarchar(50)
);
MERGE Test_Target as T
USING Test_Source as S
ON T.Id = S.Id
WHEN MATCHED AND
T.ColA <> S.ColA OR
T.ColB <> S.ColB OR
T.ColC <> S.ColC OR
T.ColD <> S.ColD OR
T.UDate <> S.UDate
THEN UPDATE SET
T.ColA = S.ColA ,
T.ColB = S.ColB ,
T.ColC = S.ColC ,
T.ColD = S.ColD ,
T.UDate = S.UDate
WHEN NOT MATCHED BY TARGET
THEN INSERT
(Id,ColA,ColB,ColC,ColD,UDate)
VALUES
(S.Id,S.ColA,S.ColB,S.ColC,S.ColD,S.UDate)
WHEN NOT MATCHED BY SOURCE
THEN DELETE
OUTPUT $action,Inserted.*,Deleted.* into @MERGERESULTS ;
-- Audit the Inserts
insert into [Audit]
(
ChangedFrom,
ChangedTo,
Action,
ChangedDate
)
select
Id_Before,
Id_After,
[Action],
GETDATE()
from @MERGERESULTS where [Action] = 'INSERT' ;
--Audit the DELETES
insert into [Audit]
(
ChangedFrom,
ChangedTo,
Action,
ChangedDate
)
select
Id_Before,
Id_After,
[Action],
GETDATE()
from @MERGERESULTS where [Action] = 'DELETE' ;
--unpivot to audit the UPDATES,
--USING REPLACE to remove the AFTER suffix when getting the column name
-- When storing the MERGE results in the TABLE variable having all columns as NVARCHAR is needed as to UNPIVOT the datataypes are needed to be same
insert into [Audit]
(
ChangedFrom,
ChangedTo,
Action,
ChangedDate,
[Column]
)
select * from (select
ChangedFrom,
ChangedTo,
[Action],
GETDATE() as ChangedDate,
REPLACE(p1col, '_After', '') as [Column]
from
(select
*
from @MERGERESULTS
UNPIVOT
(
ChangedFrom
FOR pcol IN (ColA_Before,ColB_Before,ColC_Before,ColD_Before)
) AS P
UNPIVOT
(
ChangedTo
For p1col IN (ColA_After,ColB_After,ColC_After,ColD_After )
) p1) as a
where
a.pcol = 'ColA_Before' and a.p1col = 'ColA_After' OR
a.pcol = 'ColB_Before' and a.p1col = 'ColB_After' OR
a.pcol = 'ColC_Before' and a.p1col = 'ColC_After' OR
a.pcol = 'ColD_Before' and a.p1col = 'ColD_After') as o -- this where condition is needed to get only relevant results, else we will get all the cols as it is a cross join
where o.ChangedFrom <> o.ChangedTo;
END
执行结果
EXEC MERGE_AUDIT
【讨论】：
以上是关于将 MERGE 语句的更新列审计为不同表中的行的主要内容，如果未能解决你的问题，请参考以下文章