SQL Server：基于另一个表的值从表中删除

Question

如果在t1上不存在相同的itemid，storeid，MSRTime的值，并且在t3上存在相同的itemid，storeid，MSRTime的值，并且状态为D.在下面的例子中，我应该能够删除t2的第二行，但不是第一行。

表1：T1

itemid |storeid|MSRTime 
x  y  z 

表2：T2

itemid |storeid|MSRTime 
    x  y  z 
    a  b  c 

表3：T3

itemid |storeid|MSRTime|status 
    x  y  z D 
    a  b  c D 

我试图做此使用加入，但我不能达到期望的结果。请帮忙。
谢谢。

Answer 1

你几乎可以编写查询完全按照你描述的那样：

declare @t1 table(itemid varchar(7),storeid varchar(9),MSRTime varchar(3)) 
insert into @t1(itemid,storeid,MSRTime) values 
('x','y','z') 
declare @t2 table(itemid varchar(7),storeid varchar(9),MSRTime varchar(3)) 
insert into @t2(itemid,storeid,MSRTime) values 
('x','y','z'), 
('a','b','c') 
declare @t3 table(itemid varchar(7),storeid varchar(9),MSRTime varchar(3),status varchar(4)) 
insert into @t3(itemid,storeid,MSRTime,status) values 
('x','y','z','D'), 
('a','b','c','D') 

delete from t2 
from @t2 t2 
    inner join 
    @t3 t3 
     on 
      t2.itemid = t3.itemid and 
      t2.storeid = t3.storeid and 
      t2.MSRTime = t3.MSRTime and 
      t3.status = 'D' 
where 
    not exists (
     select * 
     from @t1 t1 
     where t1.itemid = t2.itemid and 
      t1.storeid = t2.storeid and 
      t1.MSRTime = t2.MSRTime 
    ) 

select * from @t2 

结果：

itemid storeid MSRTime 
------- --------- ------- 
x  y   z 

Answer 2

应该是这样的

-- delete t2 
select * 
from table2 t2 
JOIN table3 t3 on t2.itemid = t3.itemid and 
        t2.storeid = t3.storeid and 
        t2.MSRTime = t3.MSRTime 
LEFT JOIN table1 t1 on t2.itemid = t1.itemid and 
         t2.storeid = t1.storeid and 
         t2.MSRTime = t1.MSRTime 
where t1.itemID IS NULL 

运行SELECT第一，如果它给你回右排，只是取消注释删除，你是好去

Answer 3

我已经创建了整个脚本供您参考。请为您的方案使用最后一个DELETE查询。这将做到这一点。

CREATE TABLE #T1 
(itemid VARCHAR(10) 
,storeid VARCHAR(10) 
,MSRTime VARCHAR(10)) 

INSERT INTO #T1 VALUES ('x','y','z') 

SELECT * FROM #T1 

GO 

CREATE TABLE #T2 
(itemid VARCHAR(10) 
,storeid VARCHAR(10) 
,MSRTime VARCHAR(10)) 

INSERT INTO #T2 VALUES ('x','y','z'),('a','b','c') 

SELECT * FROM #T2 

GO 

CREATE TABLE #T3 
(itemid VARCHAR(10) 
,storeid VARCHAR(10) 
,MSRTime VARCHAR(10) 
,status VARCHAR(10)) 

INSERT INTO #T3 VALUES ('x','y','z','D'),('a','b','c','D') 

SELECT * FROM #T3 

GO 

DELETE M 
FROM #T2 AS M INNER JOIN 
(SELECT itemid,storeid,MSRTime FROM 
(SELECT itemid,storeid,MSRTime FROM #T3 WHERE status='D') T1 
INTERSECT 
(SELECT itemid,storeid,MSRTime FROM 
      (SELECT * FROM #T2 
      EXCEPT 
      SELECT * FROM #T1) T2)) X 
ON X.itemid = M.itemid AND X.storeid = M.storeid AND X.MSRTime = M.MSRTime   

GO 

Answer 4

不知道这是否您的环境相匹配，但就方案而言可能是有益的限制，可以比较结果与加入，只有那些在状态有d的值。我也会尝试用Coalese制作一个复合键，这样你就不必在三个单独的连接上进行匹配。

例如 -

itemid |storeid|MSRTime|Key x y z xyz a b c abc