Logic to get only one row in result set - sql

Question

am stuck at point to build the logic. Can anyone help me ?
requirement:
For privacyType = 'Primary Address' if there is >1 row where SI0_ADDR.ADDR_TYPE_CODE = 'M', display the row where currentDate() is between SI0_ADDR.ADDR_EFF_DATE and SI0_ADDR.ADDR_EXPR_DATE.
my query is :
select STU_ID
,case when Privacyflag = '' then 'N'
else Privacyflag
end Privacyflag
,type from (
select a.STU_ID,Privacyflag,a.type
,ROW_NUMBER() OVER ( ORDER BY ADDR_EXPR_DATE DESC) disp_nm from (
select ad.STU_ID, case when ad.ADDR_TYPE_CODE = 'M' then ad.ADDR_PRIVACY_FLAG
when ad.ADDR_TYPE_CODE='' then 'N'
end Privacyflag, 'Primary Phone' type
, case when ADDR_EXPR_DATE = '1900-01-01' then '2100-12-31'
else ADDR_EXPR_DATE
end as ADDR_EXPR_DATE
from SI0_ADDR ad
where ad.STU_ID = #studentid ) a
where Privacyflag is not null
) ab
where ab.disp_nm = '1'
This logic is not working in some of the cases

Answer 1

You're not qualifying by the date in any way, which was part of your requirements. Your query doesn't need so many nested sub-queries. The PrivacyFlag doesn't seem to be part of your question, so I didn't work it into this example. And your example shows Primary Phone, but your question talks about Primary Address, so I'm not sure how to work that out.
In any case, here is a very simple example that shows the data, and how to pull the one record with the latest expiration date, and also the latest effective date (in case there are two records with the same expiration date).
create table #temp (stu_id int, type varchar(10), address varchar(50), eff_date datetime, expr_date datetime)
insert into #temp values
(1, 'primary', '123 NW 52nd', '1/1/2016', '1/1/1900'),
(1, 'primary', '942 SE 33rd', '1/2/2016', '12/31/2016'),
(1, 'primary', '721 SW 22nd', '4/1/2015', '1/1/1900')
select top 1 *
from (
select stu_id
,type
,address
,eff_date
,case when expr_date = '1/1/1900' then '12/31/2100' else expr_date end as expr_date
from #temp
where stu_id = 1
and type = 'primary'
) as a
where getdate() between eff_date and expr_date
order by a.expr_date desc, a.eff_date desc
drop table #temp
You could even do it with zero sub-queries, but then you'd need to duplicate the case statement into the where and order by parts of the query:
select top 1
stu_id
,type
,address
,eff_date
,case when expr_date = '1/1/1900' then '12/31/2100' else expr_date end as expr_date
from #temp
where stu_id = 1
and type = 'primary'
and getdate() between eff_date and case when expr_date = '1/1/1900' then '12/31/2100' else expr_date end
order by case when expr_date = '1/1/1900' then '12/31/2100' else expr_date end desc
,eff_date desc
With such a small dataset, the showplan didn't indicate which query would be most efficient; it showed the same identical query plan for both queries, except the first select statement took 7 CPU cycles to compile the plan and the second select statement took 2 CPU cycles to compile the plan. You may want to test against a larger dataset on real tables with indexes, etc., and see which performs the best.