Date-based multiple group by T-SQL query - sql-server

Question

First of all, execuse the longer question, but I will try to put it as simply as possible...
I'm trying to write a kind of a reporting query, but I'm having a problem getting the desired results. The problem:
Employee table
Id | Name
---------------
1 | John Smith
2 | Alan Jones
3 | James Jones
Task table
Id | Title | StartDate | EmployeeId | Estimate (integer - ticks)
----------------------------------------------------------------------------
1 | task1 | 21.08.2011 | 1 | 90000000000
2 | task2 | 21.08.2011 | 1 | 150000000
3 | task3 | 22.08.2011 | 2 | 1230000000
Question:
How to get the estimate summary per day, grouped, but to include all the employees?
Like this:
Date | EmployeeId | EmployeeName | SummaryEstimate
-------------------------------------------------------------
19.08.2011 | 1 | John Smith | NULL
19.08.2011 | 2 | Alan Jones | NULL
19.08.2011 | 3 | James Jones | NULL
20.08.2011 | 1 | John Smith | NULL
20.08.2011 | 2 | Alan Jones | NULL
20.08.2011 | 3 | James Jones | NULL
21.08.2011 | 1 | John Smith | 90150000000
21.08.2011 | 2 | Alan Jones | NULL
21.08.2011 | 3 | James Jones | NULL
22.08.2011 | 1 | John Smith | NULL
22.08.2011 | 2 | Alan Jones | 1230000000
22.08.2011 | 3 | James Jones | NULL
What I currently do is I have a "dates" table with 30years of days. I left join and group by that table to get other dates included too. Well, here is the query:
SELECT dates.value, employee.Id, employee.Name, sum(task.Estimate)
FROM TableOfDates as dates
left join Tasks as task on (dates.value = convert(varchar(10), task.StartTime, 101))
left join Employees as employee on (employee.Id = task.EmployeeId)
WHERE dates.value >= '2011-08-19' and dates.value < '2011-08-22'
GROUP BY dates.value, employee.Id, employee.Name
ORDER BY dates.value, employee.Id
The convert call is to get the date part of the DateTime column.
The result that I get is:
Date | EmployeeId | EmployeeName | SummaryEstimate
-------------------------------------------------------------
19.08.2011 | NULL | NULL | NULL
20.08.2011 | NULL | NULL | NULL
21.08.2011 | 1 | John Smith | 90150000000
22.08.2011 | 2 | Alan Jones | 1230000000
I am there half of the way, I get dates that are not in the two base joined tables (Employees and Tasks) but I cannot also have all the employees included as in the table shown before this one.
I've tried cross-joining, then subqueries, but little luck there. Any help would be very much appreciated ! Thank you for having the time to go through all of this, I hope I was clear enough...

Answer 1

SELECT DE.DateValue, DE.EmployeeId, DE.EmployeeName, sum(task.Estimate)
FROM
( SELECT
D.value AS DateValue
, E.Id AS EmployeeId
, E.Name AS EmployeeName
FROM
TableOfDates D
CROSS JOIN Employees E ) DE
left join Tasks as task on DE.DateValue = convert(varchar(10), task.StartTime, 101)
AND DE.EmployeeId = task.EmployeeId
WHERE DE.DateValue >= '2011-08-19' and DE.DateValue < '2011-08-22'
GROUP BY DE.DateValue, DE.EmployeeId, DE.EmployeeName
ORDER BY DE.DateValue, DE.EmployeeId

Answer 2

Note that this solution offers the possibility to drop the day-table as you may use a dynamic recursive CTE instead.
The other CTE:s (Employees and Tasks) can be substituted with the real tables.
DECLARE #startDate DATETIME = '2011-08-01'
DECLARE #endDate DATETIME = '2011-09-01'
;WITH Employees(Id,Name)
AS
(
SELECT 1, 'John Smith'
UNION ALL
SELECT 2, 'Alan Jones'
UNION ALL
SELECT 3, 'James Jones'
)
,Tasks (Id, Title, StartDate, EmployeeId, Estimate)
AS
(
SELECT 1, 'task1', '2011-08-21', 1, 90000000000
UNION ALL
SELECT 2, 'task2', '2011-08-21', 1, 150000000
UNION ALL
SELECT 3, 'task3', '2011-08-22', 2, 1230000000
)
,TableOfDates(value)
AS
(
SELECT DATEADD(DAY, DATEDIFF(DAY, 0, #startDate), 0)
UNION ALL
SELECT DATEADD(DAY, 1, value)
FROM TableOfDates
WHERE value < #endDate
)
SELECT dates.value
,employee.Id
,employee.Name
,SUM(task.Estimate) AS SummaryEstimate
FROM TableOfDates dates
CROSS JOIN Employees employee
LEFT JOIN Tasks task
ON dates.value = task.StartDate
AND (employee.Id = task.EmployeeId)
WHERE dates.value >= '2011-08-19'
AND dates.value < '2011-08-26'
GROUP BY
dates.value
,employee.Id
,employee.Name
ORDER BY
dates.value
,employee.Id

Answer 3

use this query:
create table #T_dates (id_date int identity(1,1),inp_date datetime)
create table #T_tasks (id_task int identity(1,1),key_date int, key_emp int, est int)
create table #T_emp (id_emp int identity(1,1),name varchar(50))
insert #T_dates (inp_date) values ('08.19.2011')
insert #T_dates (inp_date) values ('08.20.2011')
insert #T_dates (inp_date) values ('08.21.2011')
insert #T_dates (inp_date) values ('08.22.2011')
insert #T_dates (inp_date) values ('08.23.2011')
insert #T_dates (inp_date) values ('08.24.2011')
--select * from #T_dates
insert #T_emp (name) values ('John Smith')
insert #T_emp (name) values ('Alan Jones')
insert #T_emp (name) values ('James Jones')
--select * from #T_emp
insert #T_tasks (key_date,key_emp,est) values (4,1,900000)
insert #T_tasks (key_date,key_emp,est) values (4,1,15000)
insert #T_tasks (key_date,key_emp,est) values (5,2,123000)
--select * from #T_tasks
select inp_date,id_emp,name,EST
from #T_emp
cross join #T_dates
left join
(
select key_date,key_emp,SUM(est) 'EST' from #T_tasks group by key_date,key_emp
) Gr
ON Gr.key_emp = id_emp and Gr.key_date = id_date
where inp_date >= '2011-08-19' and inp_date <= '2011-08-22'
order by inp_date,id_emp