Select only records where column values are not all equal to zero

Question

Hi everyone, this seems so easy but it's got me stumped on finding a clean, easy way to accomplish this. I'm sure when someone responds, I'll kick myself. So, assume the following is what I've got:
with mytable as (
select 'Type 1' as itemtype, 'JAN' as monthname, 0 as theval from dual union all
select 'Type 1' as itemtype, 'FEB' as monthname, 1 as theval from dual union all
select 'Type 1' as itemtype, 'MAR' as monthname, 5 as theval from dual union all
select 'Type 1' as itemtype, 'APR' as monthname, 1 as theval from dual union all
select 'Type 1' as itemtype, 'MAY' as monthname, 4 as theval from dual union all
select 'Type 1' as itemtype, 'JUL' as monthname, 0 as theval from dual union all
select 'Type 1' as itemtype, 'AUG' as monthname, 0 as theval from dual union all
select 'Type 1' as itemtype, 'SEP' as monthname, 1 as theval from dual union all
select 'Type 1' as itemtype, 'OCT' as monthname, 7 as theval from dual union all
select 'Type 1' as itemtype, 'NOV' as monthname, 1 as theval from dual union all
select 'Type 1' as itemtype, 'DEC' as monthname, 2 as theval from dual union all
select 'Type 2' as itemtype, 'JAN' as monthname, 0 as theval from dual union all
select 'Type 2' as itemtype, 'FEB' as monthname, 0 as theval from dual union all
select 'Type 2' as itemtype, 'MAR' as monthname, 0 as theval from dual union all
select 'Type 2' as itemtype, 'APR' as monthname, 0 as theval from dual union all
select 'Type 2' as itemtype, 'MAY' as monthname, 0 as theval from dual union all
select 'Type 2' as itemtype, 'OCT' as monthname, 0 as theval from dual union all
select 'Type 2' as itemtype, 'NOV' as monthname, 0 as theval from dual union all
select 'Type 2' as itemtype, 'DEC' as monthname, 0 as theval from dual
select
  itemtype,
  sum (case monthname when 'JAN' then theval else 0 end) as JAN,
  sum (case monthname when 'FEB' then theval else 0 end) as FEB,
  sum (case monthname when 'MAR' then theval else 0 end) as MAR,
  sum (case monthname when 'APR' then theval else 0 end) as APR,
  sum (case monthname when 'MAY' then theval else 0 end) as MAY,
  sum (case monthname when 'JUN' then theval else 0 end) as JUN,
  sum (case monthname when 'JUL' then theval else 0 end) as JUL,
  sum (case monthname when 'AUG' then theval else 0 end) as AUG,
  sum (case monthname when 'SEP' then theval else 0 end) as SEP,
  sum (case monthname when 'OCT' then theval else 0 end) as OCT,
  sum (case monthname when 'NOV' then theval else 0 end) as NOV,
  sum (case monthname when 'DEC' then theval else 0 end) as DEC
from mytable
group by itemtype
order by itemtypeI need an outer query around this one or something that will only select "Type 1"... i.e., if all of the months are each equal to zero, don't include the record in the result set.
Summing them to get a total of zero is not an option, as I could have -15 and +15 in different columns, in which case, that record would need to be displayed.
Something as simple as... "where not (oct = 0 and nov = 0 and dec = 0...) at the end is all I seem to need. I've thought about adding a case clause for each column, but that doesn't seem very efficient. Ideas?
Thanks in advance!
Mark
Edit... I know the following will work using the MINUS operator, but my real query is really huge, and I don't want to have to write it twice...
{code}
select
itemtype,
sum (case monthname when 'JAN' then theval else 0 end) as JAN,
sum (case monthname when 'FEB' then theval else 0 end) as FEB,
sum (case monthname when 'MAR' then theval else 0 end) as MAR,
sum (case monthname when 'APR' then theval else 0 end) as APR,
sum (case monthname when 'MAY' then theval else 0 end) as MAY,
sum (case monthname when 'JUN' then theval else 0 end) as JUN,
sum (case monthname when 'JUL' then theval else 0 end) as JUL,
sum (case monthname when 'AUG' then theval else 0 end) as AUG,
sum (case monthname when 'SEP' then theval else 0 end) as SEP,
sum (case monthname when 'OCT' then theval else 0 end) as OCT,
sum (case monthname when 'NOV' then theval else 0 end) as NOV,
sum (case monthname when 'DEC' then theval else 0 end) as DEC
from mytable
group by itemtype
minus
select
itemtype,
jan,feb,mar,apr,may,jun,jul,aug,sep,oct,nov,dec
from (
select
itemtype,
sum (case monthname when 'JAN' then theval else 0 end) as JAN,
sum (case monthname when 'FEB' then theval else 0 end) as FEB,
sum (case monthname when 'MAR' then theval else 0 end) as MAR,
sum (case monthname when 'APR' then theval else 0 end) as APR,
sum (case monthname when 'MAY' then theval else 0 end) as MAY,
sum (case monthname when 'JUN' then theval else 0 end) as JUN,
sum (case monthname when 'JUL' then theval else 0 end) as JUL,
sum (case monthname when 'AUG' then theval else 0 end) as AUG,
sum (case monthname when 'SEP' then theval else 0 end) as SEP,
sum (case monthname when 'OCT' then theval else 0 end) as OCT,
sum (case monthname when 'NOV' then theval else 0 end) as NOV,
sum (case monthname when 'DEC' then theval else 0 end) as DEC
from mytable
group by itemtype
where ( oct = 0 and nov = 0 and dec = 0 and jan = 0 and feb = 0 and mar = 0
and apr = 0 and may = 0 and jun = 0 and jul = 0 and aug = 0 and sep = 0
order by itemtype
{code}
Edit again... Ok, I guess I'm answering my own question here, but I think using a WITH clause to write the main query once and then selecting * from it twice using the MINUS operator in between where the second query has where (oct = 0, etc.) is what I need. If anyone else has better suggestions, please do let me know! Here's the pseudo logic for what I've come up with so far...
{code}
WITH mainquery as (select everything)
select * from mainquery
minus
select * from mainquery where (oct = 0, nov = 0, etc...)
{code}
Thanks again!
Mark
Edited by: user455268 on Mar 1, 2012 7:13 PM
Edited by: user455268 on Mar 1, 2012 7:16 PM

Answer 1

Hi,
You can do that with a HAVING clause:
select
  itemtype,
  sum (case monthname when 'JAN' then theval else 0 end) as JAN,
  sum (case monthname when 'FEB' then theval else 0 end) as FEB,
  sum (case monthname when 'MAR' then theval else 0 end) as MAR,
  sum (case monthname when 'APR' then theval else 0 end) as APR,
  sum (case monthname when 'MAY' then theval else 0 end) as MAY,
  sum (case monthname when 'JUN' then theval else 0 end) as JUN,
  sum (case monthname when 'JUL' then theval else 0 end) as JUL,
  sum (case monthname when 'AUG' then theval else 0 end) as AUG,
  sum (case monthname when 'SEP' then theval else 0 end) as SEP,
  sum (case monthname when 'OCT' then theval else 0 end) as OCT,
  sum (case monthname when 'NOV' then theval else 0 end) as NOV,
  sum (case monthname when 'DEC' then theval else 0 end) as DEC
from mytable
group by itemtype
HAVING      MIN (theval)     != 0
OR      MAX (theval)     != 0
order by itemtype
;If the values are all 0, then the MIN and the MAX will both be 0.
If either the MIN or the MAX is not 0, then the values are not all 0.
This assumes that the combination (itemtype, monthname) is unique, as it is in your sample data.
If that's not the case, start with a subquery that GROUPs BY itemtype, monthname, so that, when you get to the main query, that combination will be unique.