SQL Server-如果新插入行； 更新（如果存在）

Question

Given:

Master Table:

A  B  C
a1 b1 NULL
a3 b2 NULL

New_Row Table:

A  B  C
a1 b1 c1
a2 b2 c2

Desired Output:

A  B  C 
a1 b1 c1
a2 b2 c2
a3 b2 NULL

I need a query that compares each columns between the Master Table and New_Row Table, and either updates the Master Table rows if all other columns in that row is identical or inserts a new row if the other columns in that row are different.

In the above example, the first row from the New_Row Table matches with the first row in the Master Table with the exception of the NULL value (assume Master Table is missing this data) so c1 is populated in the desired output. The second row from the New_Row Table doesn't match with any of the rows in the Master Table (as there is no a3 b2 rows in the Master), so it gets inserted in the desired output as a new row.

How would I write this query?

Answer 1

SQL Server supports the MERGE statement to perform INSERT , UPDATE or DELETE actions in a single query.

You can find details on this statement here: https://docs.microsoft.com/en-us/sql/t-sql/statements/merge-transact-sql

MERGE [master_table] AS TARGET
USING [new_row_table] AS SOURCE
ON (TARGET.A = SOURCE.A) AND (TARGET.B = SOURCE.B)
WHEN MATCHED THEN
    UPDATE SET TARGET.C = SOURCE.C
WHEN NOT MATCHED BY TARGET THEN
    INSERT (A, B, C)
    VALUES (SOURCE.A, SOURCE.B, SOURCE.C);

One caveat - MERGE is only supported in SQL Server 2008 and above.

Answer 2

So without a unique key on either of your tables, this is a little dicier, but still doable. You can do this either with a MERGE statement, or an explicit transaction. Which one you use is largely up to you. Personally I'm not a huge fan of MERGE because I find the syntax clunky and it can have some weird behaviors, but that's up to you. Also, i'm not sure what if any unique constraints exist on the table(s) but if there are none, and duplicates are possible, you may want to avoid MERGE as it doesn't play nicely with updating non-unique data sets.

With either approach, you also need to be careful how you're handling NULL values. If you just do a straight comparison (eg tc = sc ) and one of the columns is null, that will never evaulate as true (since a null never equals anything; even another null ). If that's something you care about, you'd need to either replace the nulls with a placeholder value (say, isnull(tc, '') = isnull(sc, '') or ad an additional check for whether the column(s) are null (eg (tc is null or sc is null or tc = sc) Setup Data

if object_id('tempdb.dbo.#New') is not null drop table #Master
create table #Master
(
    a varchar(10),
    b varchar(10),
    c varchar(10)
)

if object_id('tempdb.dbo.#new') is not null drop table #new
create table #new
(
    a varchar(10),
    b varchar(10),
    c varchar(10)
)

insert into #master
values ('a1', 'b1', null), ('a2', 'b2', null)

insert into #new
values ('a1', 'b1', 'c1'), ('a2', 'b2', 'c2')

Method 1: MERGE

    merge into #master t -- target
using #new s --source
    on t.a = s.a
        and t.b = s.b
        and t.c = s.c
when not matched by target then insert
    (
        a,
        b,
        c
    )
    values
    (
        s.a,
        s.b,
        s.c
    )
when matched then update
    set c = s.c;

Method 2: Explicit Transaction

begin tran

    update t
    set c = s.c
    from #master t -- target
    inner join #new s -- source
        on t.a = s.a
            and t.b = s.b
            and t.c = s.c

    insert into #master
    (
        a,
        b,
        c
    )
    select
        s.a,
        s.b,
        s.c
    from #new s -- source
    left outer join #master t --target
        on s.a = t.a
            and s.b = t.b
            and s.c = t.c
    where t.a is null

commit tran