Thread: Return count between timestamps
Hi
I have a car parking reservation table that stores 2 timestamps – entry_date and exit_date.
There is a maximum number of car parking places, and I want to check that on each day between the 2 requested reservation dates, the count of the existing records does not exceed the maximum.
So I need some ‘elegant’ system of getting a count of all the existing bookings for each of the days between the entry date and the exit date.
Unfortunately I have no idea how to GROUP by the dates between….
SELECT COUNT(id) as num_places FROM reservations r WHERE (('$entry_date' > r.entry_date) AND ('$entry_date' < r.exit_date)) OR (('$exit_date' > r.entry_date) AND ('$exit_date' < r.exit_date));
This obviously only returns the sum total – is there anyway I get this on a day by day basis?
Many thanks
Jonathan
- - - - - - - - - - - - - - - - - -
Nixon.
+44 (0)1736 758600
www.nixondesign.com
White’s Warehouse
Foundry Square
Hayle
Cornwall
TR27 4HH UK
Jonathan Davies <jonathan@nixondesign.com> schrieb: > > Hi > I have a car parking reservation table that stores 2 timestamps entry_date and > exit_date. > There is a maximum number of car parking places, and I want to check that on > each day between the 2 requested reservation dates, the count of the existing > records does not exceed the maximum. > So I need some elegantsystem of getting a count of all the existing bookings > for each of the days between the entry date and the exit date. > Unfortunately I have no idea how to GROUP by the dates between&. > SELECT COUNT(id) as num_places FROM reservations r WHERE (('$entry_date' > > r.entry_date) AND ('$entry_date' < r.exit_date)) OR (('$exit_date' > > r.entry_date) AND ('$exit_date' < r.exit_date)); > This obviously only returns the sum total is there anyway I get this on a day > by day basis? Yes. Example: test=# select * from parking ; id | entry | exit ----+------------+------------ 1 | 2006-03-01 | 2006-03-30 2 | 2006-03-15 | 2006-04-15 3 | 2006-03-30 | 2006-04-30 (3 rows) Now, i want to know all reservations between 2006/03/13 and the next 20 days, i create a table callend calendar: test=# create table calendar as select ('2006/03/13'::date + (generate_series(0,20) ||'days')::interval) as datum; SELECT Now, i can calculate the reservations for each day: test=# select a.datum, sum(case when a.datum between b.entry and b.exit then 1 else 0 end) from calendar a, parking b groupby a.datum order by 1; datum | sum ---------------------+----- 2006-03-13 00:00:00 | 1 2006-03-14 00:00:00 | 1 2006-03-15 00:00:00 | 2 2006-03-16 00:00:00 | 2 2006-03-17 00:00:00 | 2 2006-03-18 00:00:00 | 2 2006-03-19 00:00:00 | 2 2006-03-20 00:00:00 | 2 2006-03-21 00:00:00 | 2 2006-03-22 00:00:00 | 2 2006-03-23 00:00:00 | 2 2006-03-24 00:00:00 | 2 2006-03-25 00:00:00 | 2 2006-03-26 00:00:00 | 2 2006-03-27 00:00:00 | 2 2006-03-28 00:00:00 | 2 2006-03-29 00:00:00 | 2 2006-03-30 00:00:00 | 3 2006-03-31 00:00:00 | 2 2006-04-01 00:00:00 | 2 2006-04-02 00:00:00 | 2 (21 rows) HTH, Andreas -- Really, I'm not out to destroy Microsoft. That will just be a completely unintentional side effect. (Linus Torvalds) "If I was god, I would recompile penguin with --enable-fly." (unknow) Kaufbach, Saxony, Germany, Europe. N 51.05082°, E 13.56889°
Andreas Kretschmer <akretschmer@spamfence.net> schrieb: > Jonathan Davies <jonathan@nixondesign.com> schrieb: > > > > > Hi > > I have a car parking reservation table that stores 2 timestamps entry_date and > > exit_date. > > There is a maximum number of car parking places, and I want to check that on > > each day between the 2 requested reservation dates, the count of the existing > > records does not exceed the maximum. Btw.: you can create a trigger like this: (with a fixed limit of 6 places) create or replace function check_parking() returns trigger as $$ declare counter int; begin select into counter max(x.sum) from (select (NEW.entry + (g.n||'days')::interval)::date, sum(case when (NEW.entry + (g.n||'days')::interval)::date between a.entry and a.exit then 1 else 0 end) from parking a, generate_series(0,NEW.exit-NEW.entry) g(n) group by g.n order by 1) as x; if counter > 5 then -- limit -1!!! raise exception 'not possible'; else return NEW; end if; end; $$ language plpgsql; create trigger parking_check before insert or update on parking for each row execute procedure check_parking(); Now a test: test=# select * from parking ; id | entry | exit ----+------------+------------ 1 | 2006-03-01 | 2006-03-30 2 | 2006-03-15 | 2006-04-15 3 | 2006-03-30 | 2006-04-30 (3 rows) test=# insert into parking values (4,'2006/03/02','2006/03/20'); INSERT 0 1 test=# insert into parking values (5,'2006/03/12','2006/03/20'); INSERT 0 1 test=# insert into parking values (5,'2006/03/10','2006/03/20'); INSERT 0 1 test=# insert into parking values (6,'2006/03/10','2006/03/20'); INSERT 0 1 test=# insert into parking values (7,'2006/03/10','2006/03/20'); ERROR: not possible HTH, Andreas -- Really, I'm not out to destroy Microsoft. That will just be a completely unintentional side effect. (Linus Torvalds) "If I was god, I would recompile penguin with --enable-fly." (unknow) Kaufbach, Saxony, Germany, Europe. N 51.05082°, E 13.56889°