使用临时表删除其他表中的重复行及其依赖项

create table dbo.hasduplicates
(
    id int identity,
    --assume colA, colB is the entity/unique combo
    colA varchar(10),
    colB int,
    someOtherColumn varchar(40)
);


insert into dbo.hasduplicates(colA, colB, someOtherColumn)
values
('A', 1, 'A1 - 1'),
('A', 1, 'A1 - 2'),
('A', 1, 'A1 - 3'),
--
('A', 2, 'A2 - 1'),
('A', 2, 'A2 - 2'),
--
('B', 1, 'B1 - 1'),
('B', 1, 'B1 - 2'),
('B', 1, 'B1 - 3');


select *
from dbo.hasduplicates;


--temp table holding the to-be-deleted ids (of the duplicates)
create table #ToBedeleted(IdToDelete int);

with dup
as
(
    select *, row_number() over (partition by colA, colB /*<--cols of your entity go here*/ order by id) as RowNum
    from dbo.hasduplicates
)
insert into #ToBedeleted(IdToDelete)
select Id
from dup
where RowNum >= 2;

--contains the ids for deletion
select * from #ToBedeleted;

--cleanup the referencing tables
/*
DELETE FROM dbo.Table1 WHERE Table1Id IN (SELECT IdToDelete FROM #ToBedeleted);
DELETE FROM dbo.Table2 WHERE Table2Id IN (SELECT IdToDelete FROM #ToBedeleted);
.............
DELETE FROM dbo.Table6 WHERE Table6Id IN (SELECT IdToDelete FROM #ToBedeleted);
--finally cleanup your products table
DELETE FROM dbo.hasduplicates WHERE Id IN (SELECT IdToDelete FROM #ToBedeleted);
*/

--/*
drop table #ToBedeleted;
drop table dbo.hasduplicates;
--*/