Ignore the entire record if a matching record is found

Question

Ignore the entire record if a matching record is found

SQL_Hunt

SSC-Dedicated

Points: 33518
More actions
September 26, 2020 at 1:56 am

#3791046

Hello,
Kindly guide me with the following code in T-SQL.
I need to only extract those KeyId where the applicationname = 'Flower' does not exist. So Flower is present in keyId = 100, 101, so in the output I should not have 100 and 101. Just need to display 102.
Thanks.

Viewing 7 posts - 1 through 6 (of 6 total)

You must be logged in to reply to this topic. Login to reply

Steve Collins SSCrazy Points: 2169 More actions · Answer 1

select distinct a.KeyId
from applications a
where not exists (select 1
                  from applications app
                  where a.KeyId=app.KeyId
                        and app.applicationname='Flower');

Aus dem Paradies, das Cantor uns geschaffen, soll uns niemand vertreiben können

SQL_Hunt SSC-Dedicated Points: 33518 More actions · Answer 2

Thank you, Steve. Working like a charm.

By the way, can we tune this to be used like a self join?

Thanks.

Steve Collins SSCrazy Points: 2169 More actions · Answer 3

There are a couple of ways.

The most preferred imo would be to create 'count_grp' using a windowing function and exclude in the WHERE clause of the outer query.

with flower_cte(KeyId, applicationname, applicationId, count_grp) as (
    select *, max(case when applicationname='Flower' then 1 else 0 end) 
              over (partition by KeyId order by (select null))
    from applications) 
select KeyId, count(*) app_count, max(applicationId) max_app_id, 
                min(applicationId) min_app_id
from flower_cte fc
where count_grp=0
group by KeyId;

It could be done using a LEFT JOIN

with flower_cte(KeyId) as (
    select distinct KeyId
    from applications
    where applicationname='Flower') 
select a.KeyId, count(*) app_count, max(a.applicationId) max_app_id, 
                min(a.applicationId) min_app_id
from applications a
     left join flower_cte fc on a.KeyId=fc.KeyId
where fc.KeyId is null
group by a.KeyId;

Or it could be done using NOT IN (which I do not prefer)

with flower_cte(KeyId) as (
    select distinct KeyId
    from applications
    where applicationname='Flower') 
select a.KeyId, count(*) app_count, max(a.applicationId) max_app_id, 
                min(a.applicationId) min_app_id
from applications a
where a.KeyId not in(select * from flower_cte)
group by a.KeyId;

Aus dem Paradies, das Cantor uns geschaffen, soll uns niemand vertreiben können

Jeff Moden SSC Guru Points: 1004401 More actions · Answer 4

Heh... as a bit of a sidebar, it's nice to see someone use the term "Record" correctly. 😀

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)

Steve Collins SSCrazy Points: 2169 More actions · Answer 5

Jeff Moden wrote:

Heh... as a bit of a sidebar, it's nice to see someone use the term "Record" correctly. 😀

Has a consensus been reached and record = row? I fear you've invited a taxonomy sermon upon us

Aus dem Paradies, das Cantor uns geschaffen, soll uns niemand vertreiben können

Jeff Moden SSC Guru Points: 1004401 More actions · Answer 6

Steve Collins wrote:

Jeff Moden wrote:
Heh... as a bit of a sidebar, it's nice to see someone use the term "Record" correctly. 😀
Has a consensus been reached and record = row? I fear you've invited a taxonomy sermon upon us

No. I share no such consensus. While a single row CAN be a record, not all rows are records by themselves. In a normalized database, it usually takes more than 1 row in multiple tables to form a "record".

For example, if you look at the OPS table, the 4 rows for KeyID 100 form a "record" as do the 3 rows for keyid 101 and the two rows for key id 102.

It's also one of those "pure" things that's not usually worth fighting over because it usually doesn't matter. I was just impressed to see someone actually use the term in a more correct manner than most folks do for the first time in over a decade.

--Jeff Moden

RBAR is pronounced "ree-bar" and is a "Modenism" for Row-By-Agonizing-Row.
First step towards the paradigm shift of writing Set Based code:
________Stop thinking about what you want to do to a ROW... think, instead, of what you want to do to a COLUMN.

Change is inevitable... Change for the better is not.

Helpful Links:
How to post code problems
How to Post Performance Problems
Create a Tally Function (fnTally)