5

我需要在 SQL Server 2012 过程中执行以下伪逻辑,基于一个表变量和一个这样声明的表:

DECLARE @tmp TABLE
    (
    ID int IDENTITY(1,1),
    UserID int NOT NULL,
    SgsID int NOT NULL
    )

CREATE TABLE #Table1
    (
    ID int IDENTITY(1,1),
    UserID int NOT NULL,
    SgsID int NOT NULL
    )
  1. 对于表变量中的每一行数据@tmp
  2. 删除组合匹配的Table1UserID/SgsIDUserID/SgsIDTable1
  3. 删除那些已被删除的UserID/SgsID组合@tmpTable1

我一直在研究不同的方法,例如使用OUTPUT INTOand INTERSECT,但无法编写跨两个表删除的查询(实际上我认为这甚至不可能)。

我已经通过使用以下代码实现了上述步骤,但是,我想知道是否有任何 T-SQL 专业人士能够提出更简洁/有效的方法?

在线版本见SQLFiddle

CREATE TABLE #Table1
    (
    ID int IDENTITY(1,1),
    UserID int NOT NULL,
    SgsID int NOT NULL
    )

INSERT INTO #Table1 (UserID, SgsID) VALUES (5, 99)
INSERT INTO #Table1 (UserID, SgsID) VALUES (10, 89)
INSERT INTO #Table1 (UserID, SgsID) VALUES (150, 79)
INSERT INTO #Table1 (UserID, SgsID) VALUES (200, 69)
INSERT INTO #Table1 (UserID, SgsID) VALUES (250, 59)
SELECT * FROM #Table1 

DECLARE @tmp TABLE
    (
    ID int IDENTITY(1,1),
    UserID int NOT NULL,
    SgsID int NOT NULL
    )

INSERT INTO @tmp (UserID, SgsID) VALUES (150, 79)
INSERT INTO @tmp (UserID, SgsID) VALUES (200, 69)
INSERT INTO @tmp (UserID, SgsID) VALUES (250, 59)
INSERT INTO @tmp (UserID, SgsID) VALUES (999, 49)
SELECT * FROM @tmp

DECLARE @tbl_commonRows TABLE (UserID int, SgsID int)
INSERT INTO @tbl_commonRows 
    (
    UserID,
    SgsID
    ) 
SELECT 
    UserID,
    SgsID 
FROM
    #Table1
INTERSECT 
SELECT
    UserID,
    SgsID
FROM
    @tmp 

DELETE FROM 
    #Table1 
WHERE 
    (ID IN (
        SELECT 
            ID 
        FROM 
            #Table1 t1 INNER JOIN
            @tbl_commonRows c ON c.UserID = t1.UserID AND c.SgsID = t1.SgsID))

DELETE FROM
    @tmp
WHERE
    (ID IN (
        SELECT
            ID
        FROM
            @tmp t2 INNER JOIN
            @tbl_commonrows c ON c.UserID = t2.UserID AND c.SgsID = t2.SgsID))

SELECT * FROM #Table1 
SELECT * FROM @tmp
DROP TABLE #Table1
4

3 回答 3

4

这是解决方案:

DECLARE @tmp_ids TABLE (
    id1 INT,
    id2 INT
)

INSERT INTO @tmp_ids (id1, id2)
SELECT 
    t1.id,
    t2.id
FROM Table1 t1
INNER JOIN tmp t2
    on (t1.UserID = t2.UserID AND t1.SgsID = t2.SgsID)

DELETE FROM Table1
WHERE id IN (SELECT id1 FROM @tmp_ids)

DELETE FROM tmp
WHERE id IN (SELECT id2 FROM @tmp_ids)

请记住 - 我创建了物理表 tmp 和 Table1

于 2013-03-06T12:10:12.567 回答
2

您可以利用 OUTPUT 命令可以使用超过 INSERTED 和 DELETED 列进行删除(但不是插入,遗憾的是)这一事实:

    DECLARE @output TABLE (id int) 

    DELETE FROM tbl
    OUTPUT tmp.ID INTO @output(id)
    FROM #Table1 tbl
    JOIN @tmp tmp 
        ON tbl.UserID = tmp.UserID 
        AND tbl.SgsID = tmp.SgsID

    DELETE FROM tmp
    FROM @tmp tmp
    JOIN @Output outp ON tmp.id = outp.id   
于 2013-03-06T20:11:49.083 回答
1

您是否考虑过为此使用 MERGE?可能是另一种选择,语法很好且易于理解。

合并 (Transact-SQL)

于 2013-03-06T14:37:33.040 回答