我有一个存储过程，它动态生成MERGE个TSQL语句，以处理给定Dimension / Staging表转储之间的SCD Type 1更新。
这是T-SQL代码（由Alex Whittles原始代码修改）：
USE [OPPY_DWUSD]
GO
ALTER PROCEDURE [dbo].[GenerateMerge]
/***************************************************************
* *
* Script for use with blog post *
* "Automating T-SQL Merge to load Dimensions (SCD)" *
* http://www.purplefrogsystems.com/blog/2012/04/automating-t-sql-merge-to-load-dimensions-scd
* *
* Posted: 6th April 2012 *
* *
* By: Alex Whittles - Purple Frog Business Intelligence *
* www.PurpleFrogSystems.com *
* *
* All code samples are provided “AS IS” without warranty of *
* any kind, either express or implied, including but not *
* limited to the implied warranties of merchantability *
* and/or fitness for a particular purpose. *
* *
***************************************************************/
@Dimension varchar(50),
@Schema varchar(50),
@ETLTable varchar(50),
@ETLSchema varchar(50),
@Execute bit=0 --Should the resulting merge be returned or executed
AS
BEGIN
SET NOCOUNT ON;
--Create Carriage return variable to format the resulting query
DECLARE @crlf char(2)
SET @crlf = CHAR(13)
--Find out which Audit fields are used
DECLARE @UseIsInferred bit
DECLARE @UseFirstCreated bit
DECLARE @UseValidTo bit
DECLARE @UseIsRowCurrent bit
DECLARE @UseLastUpdated bit
SET @UseIsInferred = ISNULL((SELECT MAX(1)
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
WHERE s.name = @Schema
AND t.name = @Dimension
AND c.name = 'IsInferred'
),0)
SET @UseFirstCreated= ISNULL((SELECT MAX(1)
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
WHERE s.name = @Schema
AND t.name = @Dimension
AND c.name = 'FirstCreated'
),0)
SET @UseValidTo = ISNULL((SELECT MAX(1)
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
WHERE s.name = @Schema
AND t.name = @Dimension
AND c.name = 'ValidTo'
),0)
SET @UseIsRowCurrent = ISNULL((SELECT MAX(1)
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
WHERE s.name = @Schema
AND t.name = @Dimension
AND c.name = 'IsRowCurrent'
),0)
SET @UseLastUpdated = ISNULL((SELECT MAX(1)
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
WHERE s.name = @Schema
AND t.name = @Dimension
AND c.name = 'LastUpdated'
),0)
--Identify the business key column(s)
--Also define what the null replacement should be
DECLARE myCurBK Cursor FOR
SELECT c.name
, CASE WHEN ty.name IN ('text', 'ntext', 'varchar', 'char', 'nvarchar', 'nchar') THEN ''''''
WHEN ty.name IN ('tinyint', 'smallint', 'int', 'real', 'money', 'float', 'bit', 'decimal', 'numeric','smallmoney','bigint') THEN '0'
WHEN ty.name IN ('date', 'datetime') THEN '''19000101'''
ELSE 'NULL' END AS NullRep
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
INNER JOIN sys.types ty ON c.user_type_id = ty.user_type_id
INNER JOIN sys.extended_properties ep
ON t.object_id=ep.major_id
AND c.column_id=ep.minor_id
AND ep.class=1
AND ep.name='SCD'
WHERE s.name = @Schema
AND t.name = @Dimension
AND ep.value = 'BK'
ORDER BY c.column_id
--Identify all fields to be merged (Exclude Type 0)
DECLARE myCurType1 Cursor
FOR SELECT c.name
, CASE WHEN ty.name IN ('text', 'ntext', 'varchar', 'char', 'nvarchar', 'nchar') THEN ''''''
WHEN ty.name IN ('tinyint', 'smallint', 'int', 'real', 'money', 'float', 'bit', 'decimal', 'numeric','smallmoney','bigint') THEN '0'
WHEN ty.name IN ('date', 'datetime') THEN '''19000101'''
ELSE 'NULL' END AS NullRep
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
INNER JOIN sys.types ty ON c.user_type_id = ty.user_type_id
LEFT JOIN sys.extended_properties ep
ON t.object_id=ep.major_id
AND c.column_id=ep.minor_id
AND ep.class=1
AND ep.name='SCD'
WHERE s.name = @Schema
AND t.name = @Dimension
AND c.is_identity=0
AND ISNULL(ep.value,'1') NOT IN ('0', 'Audit', 'BK')
ORDER BY c.column_id ASC
--Identify all fields for insert
DECLARE myCurAll Cursor
FOR SELECT c.name
, CASE WHEN ty.name IN ('text', 'ntext', 'varchar', 'char', 'nvarchar', 'nchar') THEN ''''''
WHEN ty.name IN ('tinyint', 'smallint', 'int', 'real', 'money', 'float', 'bit', 'decimal', 'numeric','smallmoney','bigint') THEN '0'
WHEN ty.name IN ('date', 'datetime') THEN '''19000101'''
ELSE 'NULL' END AS NullRep
FROM sys.columns c
INNER JOIN sys.tables t on c.object_id = t.object_id
INNER JOIN sys.schemas s ON t.schema_id = s.schema_id
INNER JOIN sys.types ty ON c.user_type_id = ty.user_type_id
WHERE s.name = @Schema
AND t.name = @Dimension
AND c.name NOT IN ('LastUpdated', 'IsInferred', 'FirstCreated', 'ValidTo', 'IsRowCurrent')
AND c.is_identity=0
ORDER BY c.column_id ASC
DECLARE @Field varchar(255)
DECLARE @NullRep varchar(20)
DECLARE @SQL varchar(max)
DECLARE @SQL2 varchar(max)
DECLARE @SQL3 varchar(max)
--Now start building up the dynamic SQL
SET @SQL ='MERGE [' + @Schema + '].[' + @Dimension + '] AS Target'
SET @SQL = @SQL + @crlf + 'USING [' + @ETLSchema + '].[' + @ETLTable + '] AS Source'
OPEN myCurBK
FETCH NEXT FROM myCurBK INTO @Field, @NullRep
IF (@@FETCH_STATUS>=0)
BEGIN
SET @SQL = @SQL + @crlf + ' ON Target.' + '[' + @Field + ']' + ' = Source.' + '[' + @Field + ']'
FETCH NEXT FROM myCurBK INTO @Field, @NullRep
END
WHILE (@@FETCH_STATUS<>-1)
BEGIN
IF (@@FETCH_STATUS<>-2)
SET @SQL = @SQL + @crlf + ' AND Target.' + '[' + @Field + ']' + ' = Source.' + '[' + @Field + ']'
FETCH NEXT FROM myCurBK INTO @Field, @NullRep
END
CLOSE myCurBK
IF @UseIsRowCurrent>0 SET @SQL = @SQL + @crlf + ' AND Target.IsRowCurrent=1'
SET @SQL = @SQL + @crlf + ' WHEN MATCHED'
OPEN myCurType1
FETCH NEXT FROM myCurType1 INTO @Field, @NullRep
IF (@@FETCH_STATUS>=0)
BEGIN
SET @SQL = @SQL + @crlf + ' AND (ISNULL(Target.' + '[' + @Field + ']' + ',' + @NullRep + ') <> ISNULL(Source.' + '[' + @Field + ']' + ',' + @NullRep + ')'
FETCH NEXT FROM myCurType1 INTO @Field, @NullRep
END
WHILE (@@FETCH_STATUS<>-1)
BEGIN
IF (@@FETCH_STATUS<>-2)
SET @SQL = @SQL + @crlf + ' OR ISNULL(Target.' + '[' + @Field + ']' + ',' + @NullRep + ') <> ISNULL(Source.' + '[' + @Field + ']' + ',' + @NullRep + ')'
FETCH NEXT FROM myCurType1 INTO @Field, @NullRep
END
CLOSE myCurType1
SET @SQL = @SQL + @crlf + ' )'
SET @SQL2 = ' THEN UPDATE SET'
OPEN myCurType1
FETCH NEXT FROM myCurType1 INTO @Field, @NullRep
IF (@@FETCH_STATUS>=0)
BEGIN
SET @SQL2 = @SQL2 + @crlf + ' ' + '[' + @Field + ']' + ' = Source.' + '[' + @Field + ']'
FETCH NEXT FROM myCurType1 INTO @Field, @NullRep
END
WHILE (@@FETCH_STATUS<>-1)
BEGIN
IF (@@FETCH_STATUS<>-2)
SET @SQL2 = @SQL2 + @crlf + ' ,' + '[' + @Field + ']' + ' = Source.' + '[' + @Field + ']'
FETCH NEXT FROM myCurType1 INTO @Field, @NullRep
END
CLOSE myCurType1
IF @UseLastUpdated>0 SET @SQL2 = @SQL2 + @crlf + ' ,LastUpdated = GetDate()'
SET @SQL3 = ' WHEN NOT MATCHED THEN'
SET @SQL3 = @SQL3 + @crlf + ' INSERT ('
OPEN myCurAll
FETCH NEXT FROM myCurAll INTO @Field, @NullRep
IF (@@FETCH_STATUS>=0)
BEGIN
SET @SQL3 = @SQL3 + @crlf + ' ' + '[' + @Field + ']'
FETCH NEXT FROM myCurAll INTO @Field, @NullRep
END
WHILE (@@FETCH_STATUS<>-1)
BEGIN
IF (@@FETCH_STATUS<>-2)
SET @SQL3 = @SQL3 + @crlf + ' ,' + '[' + @Field + ']'
FETCH NEXT FROM myCurAll INTO @Field, @NullRep
END
CLOSE myCurAll
IF @UseIsInferred>0 SET @SQL3 = @SQL3 + @crlf + ' ,IsInferred'
IF @UseFirstCreated>0 SET @SQL3 = @SQL3 + @crlf + ' ,FirstCreated'
IF @UseValidTo>0 SET @SQL3 = @SQL3 + @crlf + ' ,ValidTo'
IF @UseIsRowCurrent>0 SET @SQL3 = @SQL3 + @crlf + ' ,IsRowCurrent'
IF @UseLastUpdated>0 SET @SQL3 = @SQL3 + @crlf + ' ,LastUpdated'
SET @SQL3 = @SQL3 + @crlf + ' ) VALUES ('
OPEN myCurAll
FETCH NEXT FROM myCurAll INTO @Field, @NullRep
IF (@@FETCH_STATUS>=0)
BEGIN
SET @SQL3 = @SQL3 + @crlf + ' Source.' + '[' + @Field + ']'
FETCH NEXT FROM myCurAll INTO @Field, @NullRep
END
WHILE (@@FETCH_STATUS<>-1)
BEGIN
IF (@@FETCH_STATUS<>-2)
SET @SQL3 = @SQL3 + @crlf + ' ,Source.' + '[' + @Field + ']'
FETCH NEXT FROM myCurAll INTO @Field, @NullRep
END
CLOSE myCurAll
IF @UseIsInferred>0 SET @SQL3 = @SQL3 + @crlf + ' ,0'
IF @UseFirstCreated>0 SET @SQL3 = @SQL3 + @crlf + ' ,GetDate()'
IF @UseValidTo>0 SET @SQL3 = @SQL3 + @crlf + ' ,NULL'
IF @UseIsRowCurrent>0 SET @SQL3 = @SQL3 + @crlf + ' ,1'
IF @UseLastUpdated>0 SET @SQL3 = @SQL3 + @crlf + ' ,GetDate()'
SET @SQL3 = @SQL3 + @crlf + ' );'
--clean up
DEALLOCATE myCurType1
DEALLOCATE myCurAll
DEALLOCATE myCurBK
IF @Execute = 1
BEGIN
EXEC(@SQL + @SQL2 + @SQL3)
END
ELSE
BEGIN
PRINT @SQL
PRINT @SQL2
PRINT @SQL3
END
END
在上面的存储过程中，当我传入变量时，它会产生以下代码：
MERGE [DIM].[COMPANY] AS Target
USING [DBO].[DWUSD_LIVE] AS Source
ON Target.[comp] = Source.[comp]
WHEN MATCHED
AND (ISNULL(Target.[comp name],'') <> ISNULL(Source.[comp name],'')
OR ISNULL(Target.[comp description],'') <> ISNULL(Source.[comp description],'')
)
THEN UPDATE SET
[comp name] = Source.[comp name]
,[comp description] = Source.[comp description]
,LastUpdated = GetDate()
WHEN NOT MATCHED THEN
INSERT (
[comp]
,[comp name]
,[comp description]
,FirstCreated
,LastUpdated
) VALUES (
Source.[comp]
,Source.[comp name]
,Source.[comp description]
,GetDate()
,GetDate()
);
我需要做的是在开头添加DISTINCT SELECT子查询......
MERGE [dim].[Company] AS Target
USING (
SELECT DISTINCT
[COMP NAME],
[COMP DESCRIPTION],
[COMP]
FROM [dbo].[DWUSD_LIVE]
) AS Source
因此，当它查看源（登台表）时，它会在维度中找到的相同列上执行SELECT DISTINCT，但不包括审核列（FirstCreated / LastUpdated）
我的登台表有重复记录，因此我只需要SELECT DISTINCT，否则我的维度中会有多条记录。
答案 0 :(得分：2)
看起来您将要用作这些变量中存储过程的源的表传递给：
@ETLTable varchar(50),
@ETLSchema varchar(50),
这些相同的变量用于定义源：
--Now start building up the dynamic SQL
SET @SQL ='MERGE [' + @Schema + '].[' + @Dimension + '] AS Target'
SET @SQL = @SQL + @crlf + 'USING [' + @ETLSchema + '].[' + @ETLTable + '] AS Source'
我相信，只要您从实际源表中获得相同的列，就可以使用要使用的独特查询，使用它创建视图并将该表传递给存储过程。
CREATE VIEW dbo.vwDWUSD_LIVE
AS
SELECT DISTINCT
[COMP NAME],
[COMP DESCRIPTION],
[COMP]
FROM [dbo].[DWUSD_LIVE]
GO
exec语句看起来像：
EXEC [dbo].[GenerateMerge] 'COMPANY', 'DIM', 'vwDWUSD_LIVE', 'DBO', 1
此外，您不必修改存储过程，看起来它可能被其他进程调用，如果您对其他表有相同的情况，您可以应用类似的解决方案。