Here is a way to do this in SQL, for just two-day bookings, for room_1
:
SELECT
avail1.date start_date,
"room_1" room_name
FROM vg_booking_availability avail1
/* This row is the next day to the first one */
INNER JOIN vg_booking_availability avail2 ON (avail1.date + 1 = avail2.date)
WHERE
avail1.room_1 = 0
AND avail2.room_1 = 0
/* Add in a clause to check future dates only here */
;
You could add all the rooms in this as bracketed OR
statements, but I'd be inclined to run that as a separate query (otherwise you'd have to re-search your result in PHP to determine which room returned as available)
We are getting into a bit of trouble here because all the rooms are denormalised - they would be better in another table where they can be treated much more generically.
This example can be expanded by adding more aliased rows for 4-day and 7-day searches respectively. The unavailability of rooms on a Thursday (or whatever other rules) is not directly relevant to the problem, since you can just create future rows of availability (based on how far into the future people book) and then make rooms unavailable according to those rules. That's a separate (and trivial) problem.
I'd also be inclined to change this, so you use NULL
as available and a foreign key to a customer table as unavailable. That will then give you useful information about why a room is unavailable, and will allow you to make it available again easily if a specific customer cancels their booking.
Lastly, this solution has the capacity for a large number of joins, and so testing this against data sets is essential. If you were to run this on a 10K row table it'd be fine, but it might not be if you have 1M rows (depending on your hardware and load, of course). So, once you have this working, I recommend you create a data generator (PHP is good for this) and ensure you can get the performance you require.