SQL Server: Delete from table based on value of another table - sql

Question

I want to delete from t2 if same value of itemid,storeid,MSRTime does not exist on t1 and Same value of itemid,storeid,MSRTime exist on t3 and status is D. In below example i should be able to delete second row on t2 but not 1st row.
Table 1: t1
itemid |storeid|MSRTime
x y z
Table 2: t2
itemid |storeid|MSRTime
x y z
a b c
Table 3: t3
itemid |storeid|MSRTime|status
x y z D
a b c D
I tried doing this using join but i could not reach the desired result. Please help.
Thank You.

Answer 1

You can write the query almost exactly as you've described it:
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
Result:
itemid storeid MSRTime
------- --------- -------
x y z

Answer 2

Should be something like this
-- 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
Run the select first, if it gives you back right row, just un-comment delete and you are good to go

Answer 3

I have created the whole script for your reference. Please use the last DELETE query for your scenario. That will do the trick.
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

Not sure if this matches your environment, but programmatically it may be beneficial to limit the results you are comparing with the joins to only those that have a value of D in status. I would also try making a compound key using Coalese so that you are not having to match on three separate joins.
For example -
itemid |storeid|MSRTime|Key
x y z xyz
a b c abc