merging multiple queries that have the same core syntax - sql

Question

Sorry for the vague title!
I'm building a web filter where users will be able to click on options and filter down results using MS-SQL 2012
The problem I have is i'm running 4 queries on every selection made to build the filter and 1 for the results.
Ignoring how it can be best coded so I don't have to reload the filter queries every time, I need help in how i can merge the 4 queries that produce the filter options and counts into 1.
The core of the syntax is the same where most of the logic goes to extract the results based on the filtered selection. However, i need to still produce the filters with their counts.
select datePart(yy,p.startDate) year, count(p.personId) itemCount
from person p
where p.field1 = something
and p.field2 = somethingElse...
Is there a way of running the core and producing the filter list and counts (see below) all in one rather than doing each individually?
Below is an example of 2 of the filters, but there are others that do similar things, either produce a list from existing data or produce a list from based on before, between and after certain dates.
--Filter 1 to get years and counts
with annualList as
(
select a.year
from table a
where a.year > 2000
)
select al.year, isnull(count,0) itemCount
from annualList al
left join (
select datePart(yy,p.startDate) year, count(p.personId) itemCount
from person p
where p.field1 = something
and p.field2 = somethingElse...
) group by datePart(yy,p.startDate) persons
on al.year = persons.year
order by al.year desc;
--filter 2 to get group stats
select anotherList.groupStatus, groupStatusCounts.itemCount
from (
select 'Child' as groupStatus
union all
select 'Adult' as groupStatus
union all
select 'Pensioner' as groupStatus
) anotherList
left join (
SELECT personStatus.groupStatus, count(personStatus.personId) itemCount
FROM ( select p.personId
case when (p.age between 1 and 17) then 'Child'
case when (p.age between 18 and 67) then 'Adult'
case when (p.age > 65) then 'Pensioner'
end as groupStatus
FROM person p
--and some other syntax to calculate the age...
where p.field1 = something
and p.field2 = somethingElse exactly as above query...
) personStatus
GROUP BY personStatus.groupStatus
) groupStatusCounts
ON anotherList.groupStatus = groupStatusCounts.groupStatus
As an example, using the dataset below, from regDate and using a group by I will be able to to get a list of years from 2010-2014 (filter1 code above).
Using the dataofbirth i'll need to 'calculate' the groupStatus using a case. As you can see from the dataofbirth data, I don't have any records where I can identify a pensioner hence the way I wrote filter 2 (see code above), it will give me the filters I need even if I don't have the data to represent it.
Tried to add the code on SQL Fiddle unfortunately it was down last night
INSERT INTO personTable
([PersonID], [dateOfBirth], [regDate])
VALUES
(1, '1979-01-01 00:00:00', '2010-01-01 00:00:00'),
(2, '1979-01-01 00:00:00', '2010-01-01 00:00:00'),
(3, '1979-01-01 00:00:00', '2011-01-01 00:00:00'),
(4, '1979-01-01 00:00:00', '2011-01-01 00:00:00'),
(5, '1979-01-01 00:00:00', '2012-01-01 00:00:00'),
(6, '1979-01-01 00:00:00', '2012-01-01 00:00:00'),
(7, '1984-01-01 00:00:00', '2012-01-01 00:00:00'),
(8, '1992-01-01 00:00:00', '2012-01-01 00:00:00'),
(9, '2000-01-01 00:00:00', '2013-01-01 00:00:00'),
(10, '2010-01-01 00:00:00', '2014-01-01 00:00:00')
my example of SQL Fiddle
The result want to end up with I need it to look like this based on the results shown in Fiddle
filter | Year | groupStatus
-----------+-----------+--------
2010 | 0 | null
2011 | 2 | null
2012 | 3 | null
2013 | 0 | null
2014 | 3 | null
child | null | 2
adult | null | 6
pensioner | null | 0
Thanking you in advance

Answer 1

I may have misunderstood the requirement massively, however I suspect you want to use GROUPING SETS. As a simple example (ignoring your specific filters) imagine the following simple data set:
PersonID | GroupStatus | Year
---------+-------------+--------
1 | Adult | 2013
2 | Adult | 2013
3 | Adult | 2014
4 | Adult | 2014
5 | Adult | 2014
6 | Child | 2012
7 | Child | 2014
8 | Pensioner | 2012
9 | Pensioner | 2013
10 | Pensioner | 2013
From what I gather you are trying to get 2 different summaries out of this data without repeating the query, e.g.
GroupStatus | ItemCount
------------+-----------
Adult | 5
Child | 2
Pensioner | 3
And
Year | ItemCount
------------+-----------
2012 | 2
2013 | 4
2014 | 4
You can get this in a single data set using:
SELECT Year,
GroupStatus,
ItemCount = COUNT(*)
FROM T
GROUP BY GROUPING SETS ((Year), (GroupStatus));
Which will yield a single data set:
YEAR GROUPSTATUS ITEMCOUNT
------------------------------------
NULL Adult 5
NULL Child 2
NULL Pensioner 3
2012 NULL 2
2013 NULL 4
2014 NULL 4
If you wanted to include the full data as well, you can just include a grouping set with all fields e.g. (PersonID, Year, GroupStatus).
Examples on SQL Fiddle
EDIT
The below query gives the output you have requested:
WITH AllValues AS
( SELECT TOP (DATEPART(YEAR, GETDATE()) - 2009)
Filter = CAST(2009 + ROW_NUMBER() OVER(ORDER BY object_id) AS VARCHAR(15)),
FilterType = 'Year'
FROM sys.all_objects o
UNION ALL
SELECT Filter, 'GroupStatus'
FROM (VALUES ('child'), ('Adult'), ('Pensioner')) T (Filter)
)
SELECT v.Filter,
[Year] = CASE WHEN v.FilterType = 'Year' THEN ISNULL(data.ItemCount, 0) END,
GroupStatus = CASE WHEN v.FilterType = 'GroupStatus' THEN ISNULL(data.ItemCount, 0) END
FROM AllValues AS v
LEFT JOIN
( SELECT [Year] = DATENAME(YEAR, t.RegDate),
Age = a.Name,
ItemCount = COUNT(*)
FROM T
LEFT JOIN
(VALUES
(0, 18, 'Child'),
(18, 65, 'adult'),
(65, 1000000, 'pensioner')
) a (LowerValue, UpperValue, Name)
ON a.LowerValue <= DATEDIFF(HOUR, T.dateofbirth, GETDATE()) / 8766.0
AND a.UpperValue > DATEDIFF(HOUR, T.dateofbirth, GETDATE()) / 8766.0
WHERE T.PersonID > 2
GROUP BY GROUPING SETS ((DATENAME(YEAR, t.RegDate)), (a.Name))
) AS data
ON ISNULL(data.[Year], data.Age) = v.Filter;
Example on SQL Fiddle