Querying a log table with nonlinear data - sql

Question

I'm trying to run some queries on a log-style table, which contains a bunch of nonlinear data. I have the following schema:
Signouts
+------------+----------------+------------+----------+
| signout_id | environment_id | date_start | date_end |
+------------+----------------+------------+----------+
| int | int | datetime | datetime |
+------------+----------------+------------+----------+
Environments
+-----+---------+
| id | name |
+-----+---------+
| int | varchar |
+-----+---------+
Signouts is the log table (and I say "log table" because records are never updated, only marked as "disabled" and added anew). When a user signs out an environment, their chosen start and end time is entered into the signouts table. Currently, to see if an environment is signed out, I simply check if the current date falls between date_start and date_end. If another user wants to sign out that environment, the minimum time they can choose is the ending date of the current signout.
I have a new challenge now, though. I now need to implement a reservation system. All of a sudden, dates can be anywhere in the future, and an environment can be reserved at any time. Now I need to know when an environment can still be signed out, and what those minimum (and now maximum) values are!
I've gotten it down to this naive plan, but I'm having trouble getting it into SQL:
get all signouts where start < curdate & end > curdate
if there is no current signout, get the min start of all signouts where start > curdate
if there is a signout, get the max end
Here is the closest I've gotten, among many other scrapped queries:
SELECT s.date_start_unavailable, s.date_available, e.id AS environment_id
FROM Environments AS e
LEFT OUTER JOIN (
SELECT TOP (100) PERCENT signout_id, environment_id, username, date_start, date_end, project, notes, in_use, max(date_end) as date_available, min(date_start) as date_start_unavailable
FROM dbo.Signouts
WHERE date_end >= GETDATE()
GROUP BY signout_id, environment_id, username, date_start, date_end, project, notes, in_use
ORDER BY date_start DESC
) AS s ON s.environment_id = e.id
This almost works. date_start_unavailable is the time at which the system becomes unavailable for a signout, and dave_available is the time when there are no more signouts. This still has problems, however; someone could reserve an environment years into the future for a month, and normal users wouldn't be able to see that most of the time is unallocated. I'll have to find a way to restrict this, but I can worry about that later.
The signouts last for arbitrary, user-entered amounts of time, otherwise implementing a time-block system would be trivial. If anybody can offer some DBA wisdom, it would be much appreciated!

Answer 1

Setting up my test-environment like this:
create table environment (id int, name varchar(255));
insert into environment values (1, 'DVD');
insert into environment values (2, 'BluRay');
create table signout (id int, environment_id int, date_start date, date_end date);
insert into signout values (1, 1, '01.11.2015', '09.11.2015');
insert into signout values (2, 1, '10.11.2015', '12.11.2015');
insert into signout values (3, 1, '01.12.2015', '24.12.2015');
insert into signout values (4, 2, '01.12.2015', '02.12.2015');
insert into signout values (5, 2, '04.12.2015', '07.12.2015');
insert into signout values (6, 2, '11.12.2015', '13.12.2015');
insert into signout values (7, 2, '14.12.2015', '23.12.2015');
Now, selecting the booked times is trivial:
select e.name, s.date_start d_start, s.date_end d_end, 'booked' as d_status FROM
signout s inner join environment e ON e.id = s.environment_id
Bu what about free times? These would be the times where no booking exists - so you join the table with itself in a given order:
select e.name, dateadd(DAY, 1, s.date_end) d_start,
COALESCE(dateadd(day, -1, s2.date_start), '31.12.2025') d_end, 'free'
FROM signout s
OUTER APPLY (
SELECT TOP 1 date_start, date_end from signout sx
WHERE sx.environment_id = s.environment_id
AND sx.date_start > s.date_end ORDER BY sx.date_start
) s2
inner join environment e ON e.id = s.environment_id
WHERE (s2.date_end is NULL OR s2.date_start > dateadd(DAY, 1, s.date_end))
Now UNION these together and add sort according to environment and date:
select e.name, s.date_start d_start, s.date_end d_end, 'booked' as d_status FROM
signout s inner join environment e ON e.id = s.environment_id
AND s.date_start > getdate()
UNION
select e.name, dateadd(DAY, 1, s.date_end) d_start,
COALESCE(dateadd(day, -1, s2.date_start), '31.12.2025') d_end, 'free'
FROM signout s
OUTER APPLY (
SELECT TOP 1 date_start, date_end from signout sx
WHERE sx.environment_id = s.environment_id
AND sx.date_start > s.date_end ORDER BY sx.date_start
) s2
inner join environment e ON e.id = s.environment_id
WHERE (s2.date_end is NULL OR s2.date_start > dateadd(DAY, 1, s.date_end))
AND s.date_start > getdate()
ORDER BY 1, 2
Here's what this gets me:
BluRay 2015-12-01 2015-12-02 booked
BluRay 2015-12-03 2015-12-03 free
BluRay 2015-12-04 2015-12-07 booked
BluRay 2015-12-08 2015-12-10 free
BluRay 2015-12-11 2015-12-13 booked
BluRay 2015-12-14 2015-12-23 booked
BluRay 2015-12-24 2025-12-31 free
DVD 2015-11-10 2015-11-12 booked
DVD 2015-11-13 2015-11-30 free
DVD 2015-12-01 2015-12-24 booked
DVD 2015-12-25 2025-12-31 free