Question

I have a database that contains phone call records (I am NOT a telemarketer, this is emergency call data). Each row is a call record. Some of the columns contain timing information. There is also a priority column for each call. It looks like this:

  PRIORITY    TIMING1     TIMING2     TIMING3
  -----0----------22----------3-----------43 
  -----1----------42----------10----------12
  -----0----------12----------13----------6
  -----2----------23----------12----------37
  -----1----------12----------16----------23

The output that I need is basically this one row per timing with the sum of all the time grouped by priority as columns. There are a fixed number of timings and priorities that I know of ahead of time. It would look like this:

------------Priority 0-------Priority 1-------Priority 2
  Timing1----123--------------332--------------233
  Timing2----265--------------241--------------302
  Timing3----387--------------192--------------201

Is it possible to do something like this in a single query? The query also has to be runnable on a mobile platform that is fairly resource constrained, although time to run isn’t a huge deal. Given the size of this table and the memory issue I would rather avoid temporary tables.

Answer 1

You can use UNPIVOT and PIVOT for this:

Here is a Static Version of the query – a static version means that you hard-code the values.

select *
from
(
  select *
  from
  (
    select priority, timing1, timing2, timing3
    from yourtable
  ) x
  unpivot
  (
    value
    for field in (timing1, timing2, timing3)
  ) u
) x1
pivot
(
  sum(value)
  for priority in ([0], [1], [2])
) p

see SQL Fiddle with Demo

You can also do this Dynamically:

DECLARE @colsUnpivot AS NVARCHAR(MAX),
    @query  AS NVARCHAR(MAX),
    @colsPivot as  NVARCHAR(MAX),
    @colsPivotName as  NVARCHAR(MAX)

select @colsUnpivot = stuff((select ','+ quotename(C.name)
         from sys.columns as C
         where C.object_id = object_id('yourtable') and
               C.name LIKE 'timing%'
         for xml path('')), 1, 1, '')

select @colsPivot = 
    STUFF((SELECT distinct ',' + Quotename(priority)
             from yourtable
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

select @colsPivotName = 
    STUFF((SELECT distinct ',' 
           + Quotename(priority) + ' as Priority' + cast(priority as varchar(1))
             from yourtable
            FOR XML PATH(''), TYPE
            ).value('.', 'NVARCHAR(MAX)') 
        ,1,1,'')

set @query 
  = 'select col, '+ @colsPivotName +'
      from
      (
        select priority, val, col
        from 
        (
          select priority, ' + @colsUnpivot + '
          from yourtable
        ) x
        unpivot
        (
          val
          for col in ('+ @colsUnpivot +')
        ) u
      ) x1
      pivot
      (
        sum(val)
        for priority in ( '+ @colspivot + ')
      ) p'

exec(@query)

see SQL Fiddle with Demo

Edit #1, you stated in the comments that you are using SQLite which does not have a PIVOT function. But you can still do this using a UNION ALL and then an aggregate function with a CASE:

select col,
  sum(case when priority = 0 then value end) as Priority0,
  sum(case when priority = 1 then value end) as Priority1,
  sum(case when priority = 2 then value end) as Priority2
from
(
  select priority, timing1 as value, 'timing1' as col
  from yourtable
  union all
  select priority, timing2 as value, 'timing2' as col
  from yourtable
  union all
  select priority, timing3 as value, 'timing3' as col
  from yourtable
) x
group by col

see SQL Fiddle with Demo