How to Use Count in Update Statement condition? - sql

Question

We have two tables, Bucket and Student. We have to update Bucket_id in Student table having same country and city. But in my Bucket table I have two Bucket Ids against same country and city we can differentiate them by Desc.
When we update Bucket_id in Student table keep in mind when can only assign 70 students to same bucket Id. After 70 Students have been assigned, it will update with another bucket Id
Bucket Table
BucketId Desc Country City
1 SOUTH PAK KHI
2 NORTH PAK KHI
Initially Bucket_Id in Student is null
Expected Result Below: After Update Bucket_Id in Student Table
Student Table
StudentId Country City BucketId
101 PAK KHI 1
102 PAK KHI 1
109 PAK KHI 1
201 PAK KHI 2
I have tried using
update student std set Bucket_Id = (
select bucket_id from Bucket b where b.country=std.country and b.city = std.city and count(std.bucket_id) < 70
);
How can we do that? We are using oracle 11g.

Answer 1

You can assign the buckets using a query:
select s.*, b.bucketid
from (select s.*,
row_number() over (partition by country, city order by studentid) - 1 as seqnum
from students s
) s join
(select b.*,
row_number() over (partition by country, city order by bucketid) - 1 as seqnum
from buckets b
) b
on s.seqnum between b.seqnum * 70 and (b.seqnum + 1) * 70 - 1;
Oracle doesn't have very good support in update for handling updates across tables, so you then need to use merge:
merge into
students s using
(select s.*, b.bucketid
from (select s.*,
row_number() over (partition by country, city order by studentid) - 1 as seqnum
from students s
) s join
(select b.*,
row_number() over (partition by country, city order by bucketid) - 1 as seqnum
from buckets b
) b
on s.seqnum between b.seqnum * 70 and (b.seqnum + 1) * 70 - 1
) ss
on s.studentid = ss.studentid
when matched then update
set s.bucketid = ss.bucketid;

Answer 2

See if this works for you:
update students std set BucketId = (
select min(bucketid) from Bucket b where b.country=std.country and b.city = std.city)
where bucketid is null
and std.STUDENTID in (
SELECT x.STUDENTID FROM
(
select StudentId,
rank() over(partition by country, city order by StudentId) rn
from students
) x where rn <= 3);

Answer 3

You can also try this, if I understood the problem correctly:
MERGE INTO student t
USING (
SELECT
studentid
, CAST(
ROW_NUMBER() OVER(
PARTITION BY country,city
ORDER BY studentid
) / 70 + 0.5
AS NUMBER(9,0)
) AS bucketid
FROM student
) s
ON (t.studentid=s.studentid)
WHEN MATCHED THEN UPDATE
SET bucketid = s.bucketid