I have a table where I want every parent to be mapped to every child within each group.
Input:
group_id parent child
1 A E
1 B
2 C F
2 D
2 E G
3 X
3 Y
Output:
group_id parent child
1 A E
1 B E
2 C F
2 D F
2 E F
2 C G
2 D G
2 E G
So within each group, I want every value in parent
to map to every value in child
. If there are no values in child
for a group, I want that group completely omitted (as shown where group_id
= 3)
I was originally thinking of using GROUP BY 1, 2
and aggregating by something like MAX(child)
, but then I came across edge cases in my data where there may be >1 child. I also tried using CROSS JOIN
but I'm struggling with getting my desired output. Thanks in advance.
CodePudding user response:
Disclaimer, I don't use Redshift so there may be better options. However, a CROSS JOIN should work. Just grab the DISTINCT parent values for all groups. Then do the same for the child values, and JOIN the two results together
SELECT p.group_id, p.parent, c.child
FROM (
SELECT group_id, parent
FROM YourTable
GROUP BY group_id, parent
)
p CROSS JOIN
(
SELECT group_id, child
FROM YourTable
WHERE child IS NOT NULL
GROUP BY group_id, child
)
c
WHERE p.group_id = c.group_id
ORDER BY p.group_id, c.child, p.parent
Results:
group_id | parent | child |
---|---|---|
1 | A | E |
1 | B | E |
2 | C | F |
2 | D | F |
2 | E | F |
2 | C | G |
2 | D | G |
2 | E | G |
db<>fiddle here
CodePudding user response:
You can do it exploiting a variable. The variable will be set to the value of child
if its value is not null, otherwise it will copy over the value contained in the variable
SET @child := "";
SELECT
group_id,
parent,
IF(child IS NULL, @child, @child := child) AS child
FROM
tab
Here's a fiddle: https://www.db-fiddle.com/f/5KnoFogV19jc4nWa8rrzVx/0.
Does it work for you?