在mssql中的合并查询中从源中筛选值

Question

I have the tableA with Key1 , datetime columns when I use merge query to insert I have get the duplicate rows in source.
How can we filer the source inside the query by datetime has maximum

MERGE tableA AS t
USING (VALUES 
        ('datakeyA1', 'datetime value'), 
        ('datakeyB1', 'datetime value'),
        ('datakeyA1', 'max datetime value')
    ) AS s (Key1, datetime)
        ON s.Key1 = t.Key1
WHEN MATCHED THEN 
    UPDATE 
    SET    Val = s.datetime
WHEN NOT MATCHED THEN 
    INSERT (Key1, datetime)
    VALUES (s.Key1, s.datetime);

When run the above query I get the:

The MERGE statement attempted to UPDATE or DELETE the same row more than once.

Is there any way to query only these rows (' datakeyB1 ', ' datetime value '),(' datakeyA1 ', ' max datetime value ') from all the values without using the intermediate table?

Expected results only insert/update the datakeyB1 , datakeyA1 values. If the date is same in both the data get only one row.

MERGE tableA AS t
USING (VALUES
        ('datakeyB1', 'datetime value'),
        ('datakeyA1', 'max datetime value')
    ) AS s (Key1, datetime)
        ON s.Key1 = t.Key1
WHEN MATCHED THEN 
    UPDATE 
    SET    Val = s.datetime
WHEN NOT MATCHED THEN 
    INSERT (Key1, datetime)
    VALUES (s.Key1, s.datetime);

Answer 1

You could wrap the values clause with a select statement, using top 1 with ties and order by row_number... - like this:

MERGE tableA AS t
USING (
        SELECT TOP 1 WITH TIES *
        FROM
            (VALUES 
                ('datakeyA1', 'datetime value'), 
                ('datakeyB1', 'datetime value'),
                ('datakeyA1', 'max datetime value')
            ) AS s (Key1, datetime)
        ORDER BY ROW_NUMBER() OVER(PARTITION BY Key1 ORDER BY datetime DESC)
        ) s
        ON s.Key1 = t.Key1
WHEN MATCHED THEN 
    UPDATE 
    SET    Val = s.datetime
WHEN NOT MATCHED THEN 
    INSERT (Key1, datetime)
    VALUES (s.Key1, s.datetime);