Reputation: 53
I have 2 sql data set:
Data set 1:
select * from (
Select 1 rn, 'b' c, 3 Val from dual UNION ALL
Select 1 rn, 'r' c, 3 Val from dual UNION ALL
Select 1 rn, 'w' c, 3 Val from dual UNION ALL
Select 2 rn, 'b' c, 2 Val from dual UNION ALL
Select 2 rn, 'r' c, 2 Val from dual UNION ALL
Select 3 rn, 'b' c, 1 Val from dual
) t1
Data set 2:
select * from (
Select 'b' c from dual UNION ALL
Select 'r' c from dual UNION ALL
Select 'w' c from dual ) t2
I need to get following data set:
rn c val
1 b 3
1 r 3
1 w 3
2 b 2
2 r 2
2 w null
3 b 1
3 r null
3 w null
I tried do this with the following sql, but it doesn't work.
select t1.*, t2.*
FROM (
Select 1 rn, 'b' c, 3 Val from dual UNION ALL
Select 1 rn, 'r' c, 3 Val from dual UNION ALL
Select 1 rn, 'w' c, 3 Val from dual UNION ALL
Select 2 rn, 'b' c, 2 Val from dual UNION ALL
Select 2 rn, 'r' c, 2 Val from dual UNION ALL
Select 3 rn, 'b' c, 1 Val from dual
) t1 right join (
Select 'b' c from dual UNION ALL
Select 'r' c from dual UNION ALL
Select 'w' c from dual ) t2 on t1.c=t2.c
order by 1,2
Please, help me fix my query. I use Oracle database.
Upvotes: 2
Views: 98
Reputation: 17924
This is the exact use case for a partitioned outer join. It should be significantly more efficient for large data sets than the CROSS JOIN
approach.
The syntax is like this:
select t1.rn,
t2.c,
t1.val
from t1 partition by (rn) right join t2 on t2.c = t1.c;
The PARTITION BY
keyword tells Oracle to repeat the outer join for each distinct value of t1.rn
.
Here is a full example:
with t1 as (
Select 1 rn, 'b' c, 3 Val from dual UNION ALL
Select 1 rn, 'r' c, 3 Val from dual UNION ALL
Select 1 rn, 'w' c, 3 Val from dual UNION ALL
Select 2 rn, 'b' c, 2 Val from dual UNION ALL
Select 2 rn, 'r' c, 2 Val from dual UNION ALL
Select 3 rn, 'b' c, 1 Val from dual
),
t2 as (
Select 'b' c from dual UNION ALL
Select 'r' c from dual UNION ALL
Select 'w' c from dual
)
select t1.rn,
t2.c,
t1.val
from t1 partition by (rn) right join t2 on t2.c = t1.c;
+----+---+-----+ | RN | C | VAL | +----+---+-----+ | 1 | b | 3 | | 1 | r | 3 | | 1 | w | 3 | | 2 | b | 2 | | 2 | r | 2 | | 2 | w | - | | 3 | b | 1 | | 3 | r | - | | 3 | w | - | +----+---+-----+
Upvotes: 4
Reputation: 142720
With a little help of cross-join between those two tables (so that you'd get all [rn, c]
combinations; that's my temp
CTE) and then outer-joining it to t1
, you'd get the result.
Sample data in lines #1 - 14; query you might be interested in begins at line #15 (just precede it with the WITH
keyword):
SQL> with t1 as (
2 Select 1 rn, 'b' c, 3 Val from dual UNION ALL
3 Select 1 rn, 'r' c, 3 Val from dual UNION ALL
4 Select 1 rn, 'w' c, 3 Val from dual UNION ALL
5 Select 2 rn, 'b' c, 2 Val from dual UNION ALL
6 Select 2 rn, 'r' c, 2 Val from dual UNION ALL
7 Select 3 rn, 'b' c, 1 Val from dual
8 ),
9 t2 as (
10 Select 'b' c from dual UNION ALL
11 Select 'r' c from dual UNION ALL
12 Select 'w' c from dual
13 ),
14 --
15 temp as
16 (select distinct a.rn, b.c
17 from t1 a cross join t2 b
18 )
19 select t.rn, t.c, a.val
20 from temp t left join t1 a on a.rn = t.rn and a.c = t.c
21 order by t.rn, t.c;
RN C VAL
---------- - ----------
1 b 3
1 r 3
1 w 3
2 b 2
2 r 2
2 w
3 b 1
3 r
3 w
9 rows selected.
SQL>
Upvotes: 2