Filtering Join in Oracle DB - sql

Question

Problem:
Each KEY in Table A should have one RF record and one SJ record however I have some duplicated SJ records.
Objective:
I wish to use the SJ records in Table B to identify which SJ record in Table A to keep.
Info:
Table A and Table B share a KEY and SEQ_NBR field.
Inputs:
Table A looks as follows
KEY ID_TYPE SEQ_NBR BUS_NAME
1234 RF 1 COMP_A
1234 SJ 2 COMP_B
1234 SJ 4 COMP_C
5678 RF 1 COMP_L
5678 SJ 2 COMP_M
5678 SJ 3 COMP_N
Table B looks as follows
KEY SEQ_NBR BUS_NAME
1234 2 COMP_B
5678 3 COMP_N
Desired Outcome:
My output would look as follows
KEY ID_TYPE SEQ_NBR BUS_NAME
1234 RF 1 COMP_A
1234 SJ 2 COMP_B
5678 RF 1 COMP_L
5678 SJ 3 COMP_N

Answer 1

Here is one way:
select key, id_type, seq_nbr, bus_name
from (
select a.*,
row_number() over (partition by a.key, a.id_type
order by b.key) as rn
from a left outer join b on a.key = b.key and a.seq_nbr = b.seq_nbr
)
where rn = 1
;
The left outer join adds columns from table b to those of table a. We need that for a single purpose: as we partition by key and id_type, we have partitions of either a single row or (two or more) rows. In the latter case, only one row has a non-null value in b.key. If we order by b.key, the row with non-null b.key will get row number = 1 (and we don't care about the rest).
Then the outer query simply keeps all the rows with row number = 1 and ignores the rest.
An alternative solution, using the union all of the two tables (slightly modified as needed) and basic aggregation using the last aggregate function:
select key, id_type,
min(seq_nbr) keep (dense_rank last order by source) as seq_nbr,
min(bus_name) keep (dense_rank last order by source) as bus_name
from (
select 'A' as source, a.* from a
union all
select 'B', key, 'SJ', seq_nbr, bus_name from b
)
group by key, id_type
;
You can test both to see which is more efficient on your data (if performance is important).

Answer 2

Here goes your code:
select * from tablea a
where exists
(select 1 from tableb b where b.key=a.key and b.seq_nbr=a.seq_nbr)
or not exists (select tablea.id_type from tablea inner join tableb on tablea.key=tableb.key and tablea.SEQ_NBR=tableb.SEQ_NBR and tablea.id_type=a.id_type)

Answer 3

If I understand correctly, you can count the number of duplicates. Then use left join and filter based on both the count and the match:
select a.*
from (select a.*,
count(*) over (partition by key, id_type) as cnt
from a
) a left join
b
on b.key = a.key and
b.seq_nbr = a.seq_nbr and
b.bus_name = a.bus_name
where cnt = 1 or b.key is not null;