Putting stuff into date ranges in SQL Server 2005 - sql-server-2005

Question

I have a table with week ranges (week number,start date, end date) and a table with tutorial dates (for writing tutors (tutor ID, tutorial_date, tutorial type(A or B).
I want to create two query that shows the week ranges (week 1, week 2) across the top with the tutor names on the side with count of tutorials (of type "A") in that week's date range in each block for that week.
The result should look like this:
Counts of Tutorials of Type "A"
Tutor|Week One|Week Two|Week Three|Week Four|Total
Joe | 3 | 5 | 7 | 8 | 23
Sam | 2 | 4 | 3 | 8 | 17
Meaning that Joe completed 3 tutorials in week one, five in week two, 7 in week three, and 8 in week 4.
The second query should show totals for tutorial type "A" and type "B"
Tutor|Week One|Week Two|Week Three|Week Four|Total |
Joe | 3/1 | 5/3 | 7/2 | 8/2 | 23/8 |
Sam | 2/3 | 4/4 | 3/2 | 8/3 | 17/12 |
Here, in Week One, Joe has done 3 tutorials of type A and 1 of type B.
Sample table data for tutorials (week one)
Tutor | Tutorial_ID | Tutorial Date |Type|
------------------------------------------
Joe | 1 | 2011-01-01 | A |
Joe | 2 | 2011-01-02 | A |
Joe | 3 | 2011-01-03 | A |
Joe | 4 | 2011-01-03 | B |
Sam | 5 | 2011-01-01 | A |
Sam | 6 | 2011-01-02 | A |
Sam | 7 | 2011-01-03 | B |
The week table looks like this:
weekNumber |startDate |endDate
1 |2011-01-01|2011-01-15
I'd like to gen this in SQL Server 2005

Answer 1

There are a few ways to do this.
For query one, where you only need to PIVOT on type 'A' then you can do just a PIVOT
select *
from
(
select w1.tutor
, w1.type
, wk.weeknumber
from w1
inner join wk
on w1.tutorialdate between wk.startdate and wk.enddate
where w1.type = 'a'
) x
pivot
(
count(type)
for weeknumber in ([1])
)p
See SQL Fiddle with Demo
Or you can use a Count() with a CASE statement.
select w1.tutor
, COUNT(CASE WHEN w1.type = 'A' THEN 1 ELSE null END) [Week One]
from w1
inner join wk
on w1.tutorialdate between wk.startdate and wk.enddate
group by w1.tutor
See SQL Fiddle with Demo
But for the second query, I would just use a Count() with a CASE
select w1.tutor
, Cast(COUNT(CASE WHEN w1.type = 'A' AND wk.weeknumber = 1 THEN 1 ELSE null END) as varchar(10))
+ ' / '
+ Cast(COUNT(CASE WHEN w1.type = 'B' AND wk.weeknumber = 1 THEN 1 ELSE null END) as varchar(10)) [Week One]
, Cast(COUNT(CASE WHEN w1.type = 'A' AND wk.weeknumber = 2 THEN 1 ELSE null END) as varchar(10))
+ ' / '
+ Cast(COUNT(CASE WHEN w1.type = 'B' AND wk.weeknumber = 2 THEN 1 ELSE null END) as varchar(10)) [Week Two]
from w1
inner join wk
on w1.tutorialdate between wk.startdate and wk.enddate
group by w1.tutor
See SQL Fiddle with Demo
Edit as AndriyM pointed out the second could be done with a PIVOT here is a solution for the Second query:
SELECT *
FROM
(
select distinct w1.tutor
, wk.weeknumber
, left(total, len(total)-1) Totals
FROM w1
inner join wk
on w1.tutorialdate between wk.startdate and wk.enddate
CROSS APPLY
(
SELECT cast(count(w2.type) as varchar(max)) + ' / '
from w1 w2
inner join wk wk2
on w2.tutorialdate between wk2.startdate and wk2.enddate
WHERE w2.tutor = w1.tutor
AND wk2.weeknumber = wk.weeknumber
group by w2.tutor, wk2.weeknumber, w2.type
FOR XML PATH('')
) D ( total )
) x
PIVOT
(
min(totals)
for weeknumber in ([1], [2])
) p
See SQL Fiddle with Demo