So here is the output of my current query:

What I'd like to do is only keep the rows which satisfy the following:

• Col2 is Null; or
• ID Col1 == Col2 for one of the rows with the same ID, e.g., the bottom two rows are not included because the second from the bottom Col1==Col2 and the bottommost is not included because the ID and Col1 match the second from the bottom in which Col1==Col2.

For this table, the only rows which should remain are the ones with IDs 646 and 1223.

I tried the following where clause, however it's improperly retaining the bottommost row: (Col2 is null and Col1 is not null) or (Col1 <> Col2). Any help appreciated!!

CodePudding user response：

You can do this:

    select *
    from my_table
    where (Col2 is null and Col1 is not null) 
     or (id not in (select id 
                    from my_table
                    where Col1 = Col2)
         )

db-fiddle here: https://www.db-fiddle.com/f/xmufn82r9W522YiKKh1Qka/0

CodePudding user response：

I would firstly mark the IDs and col1 of bad rows, then filter the table using these marked rows:

WITH marked AS
(
SELECT DISTINCT ID, Col1
FROM table
WHERE Col1 = Col2 OR Col1 IS NULL AND Col2 IS NULL
)
SELECT ID
FROM table t1
LEFT JOIN marked t2 ON t1.ID = t2.ID AND t1.Col1 = T2.ColId
WHERE t2.ID IS NULL 

CodePudding user response：

Your logic can be applied with NOT EXISTS:

SELECT t1.*
FROM tablename t1
WHERE t1.Col2 IS NULL
   OR NOT EXISTS (
     SELECT 1 
     FROM tablename t2
     WHERE t2.Col1 = t2.Col2 AND t2.ID = t1.ID
   );

See the demo.