Compute sum of hours/day with overlapping periods, SQL - sql

Question

I have a table with the start date(event.startdate), end date(event.enddate), and the hours/person (event.hrday) of an event. I have another table with weekdays listed which has another field for each person (calendar.name). I want to populate those columns with the total hours worked each day. I can't seem to figure out how to properly sum the hours if two events overlap in dates, I can only come up with a correct value for a single event in a time period.
I believe that in theory this question has the answer I need: compute sum of values associated with overlapping date ranges
But I am very new to SQL and I don't fundamentally understand the solution posted even after some additional research. I am using Access 2013. Apologies if this is a super elementary question, I was hoping what I wanted to do could be handled "visually" with Access...
What I Have: ("event" table)
Startdate | Enddate | Hrsday | Name
5/1/2015 5/12/2015 1.25 Joe
5/7/2015 5/8/2015 8 Joe
What I'm looking for:("calendar" table, days already filled in first column)
Weekdays | Joe | name2 | name3 | ....
5/1/2015 1.25
5/4/2015 1.25
5/5/2015 1.25
5/6/2015 1.25
5/7/2015 9.25
5/8/2015 9.25
5/11/2015 1.25
5/12/2015 1.25
I've tried using the query builder within access to build an UPDATE query, but my result either does not appear at all (no updates, all null) or will only fill in one event with no overlaps. (5/1-5/12 all have 1.25).

Answer 1

I think you will need to create a "date table" if you want to be able to achieve this sort of result in MS-Access (without using windowed functions).
Here is a quick example of how this might work in SQL Server, but only using syntax available to MS-Access (I hope).
--Load the test data into a table variable
DECLARE #event TABLE (
[start_date] DATE,
end_date DATE,
hrsperday NUMERIC(19,2),
name VARCHAR(20));
INSERT INTO #event SELECT '20150401', '20150412', 1.25, 'Joe';
INSERT INTO #event SELECT '20150407', '20150408', 8, 'Joe';
--Add some more test data, to make it more "interesting"
INSERT INTO #event SELECT '20150401', '20150405', 0.1, 'Bill';
INSERT INTO #event SELECT '20150401', '20150430', 7.5, 'Bill';
INSERT INTO #event SELECT '20150412', '20150415', 0.5, 'Bill';
--Make a date table, this creates one on the fly but wouldn't work in MS-Access
--I store a date for each day in 2015/Apr, obviously I would want more dates eventually
DECLARE #dates TABLE (
[date] DATE);
WITH cte AS (
SELECT CONVERT(DATE, '20150401') AS [date]
UNION ALL
SELECT DATEADD(DAY, 1, [date]) FROM cte WHERE [date] < '20150430')
INSERT INTO
#dates
SELECT
[date]
FROM
cte OPTION (MAXRECURSION 0);
--Now the answer is trivial
SELECT
e.name,
d.[date],
SUM(hrsperday) AS hrs
FROM
#dates d
LEFT JOIN #event e ON d.[date] BETWEEN e.[start_date] AND e.end_date
GROUP BY
e.name,
d.[date]
ORDER BY
e.name,
d.[date];
--Note the format you want, but a PIVOT would give you this
--(I don't think PIVOT is supported by MS-Access though)
Results for this are:
name date hrs
Bill 2015-04-01 7.60
Bill 2015-04-02 7.60
Bill 2015-04-03 7.60
Bill 2015-04-04 7.60
Bill 2015-04-05 7.60
Bill 2015-04-06 7.50
Bill 2015-04-07 7.50
Bill 2015-04-08 7.50
Bill 2015-04-09 7.50
Bill 2015-04-10 7.50
Bill 2015-04-11 7.50
Bill 2015-04-12 8.00
Bill 2015-04-13 8.00
Bill 2015-04-14 8.00
Bill 2015-04-15 8.00
Bill 2015-04-16 7.50
Bill 2015-04-17 7.50
Bill 2015-04-18 7.50
Bill 2015-04-19 7.50
Bill 2015-04-20 7.50
Bill 2015-04-21 7.50
Bill 2015-04-22 7.50
Bill 2015-04-23 7.50
Bill 2015-04-24 7.50
Bill 2015-04-25 7.50
Bill 2015-04-26 7.50
Bill 2015-04-27 7.50
Bill 2015-04-28 7.50
Bill 2015-04-29 7.50
Bill 2015-04-30 7.50
Joe 2015-04-01 1.25
Joe 2015-04-02 1.25
Joe 2015-04-03 1.25
Joe 2015-04-04 1.25
Joe 2015-04-05 1.25
Joe 2015-04-06 1.25
Joe 2015-04-07 9.25
Joe 2015-04-08 9.25
Joe 2015-04-09 1.25
Joe 2015-04-10 1.25
Joe 2015-04-11 1.25
Joe 2015-04-12 1.25