Question

I’m attempting to create a report for an accommodation service with the following information:

Number of Bookings (Easy, use the COUNT function)
Revenue Amount (Kind of easy).
Number of Room nights. (Rather Hard it seems)

Broken down into each month of the year.

Limitations – I’m currently using PHP/MySQL to create this report.
I’m pulling the data out of the booking system 1 month at a time, then using an ETL process to put it into MySQL.
Because of this, I have duplicate records, when a booking splits across the end of the Month. (eg BookingID = 9216 below – This is because for Revenue purposes we need to split the percentage of the revenue into the corresponding month).

The Question.

How do I write some SQL that will:
Calculate the number of room nights that was booked into a Property and Group it by the month. Taking into account that if a booking spans across the end of the month, that the room nights that are inside of the same month, as the checkin are counted towards that month, and room nights which the same month as checkout are in the same month as checkout.

At first I used this: DATEDIFF(Checkout, Checkin).
But that lead to one month having 48 room nights in a 31 day month. (because a) it counted 1 booking as 11 nights, even through it was split across the 2 months, and b) because it appears twice).

Then once I have the statement I need to integrate it back into my CrossTab SQL for the entire year.

Some resources that I have found, but can’t seem to make work (MySql Query- Date Range within a Date Range & php mysql double date range)

Here is a Sample of the Table: (There are ~100,000 rows of similar data).

CREATE TABLE IF NOT EXISTS `bookingdata` (
`idBookingData` int(11) NOT NULL AUTO_INCREMENT,
`PropertyID` int(10) NOT NULL,
`Checkin` date DEFAULT NULL,
`Checkout` date DEFAULT NULL,
`Rent` decimal(10,2) DEFAULT NULL,
`BookingID` int(11) DEFAULT NULL,
PRIMARY KEY (`idBookingData`),
UNIQUE KEY `idBookingData_UNIQUE` (`idBookingData`)
) ENGINE=InnoDB  DEFAULT CHARSET=latin1 AUTO_INCREMENT=10472 ;

INSERT INTO `bookingdata` (`idBookingData`, `PropertyID`, `Checkin`, `Checkout`, `Rent`, `BookingID`) VALUES
(5148, 2, '2011-07-02', '2011-07-05', 1105.00, 10612),
(5149, 2, '2011-07-05', '2011-07-13', 2155.00, 10184),
(5151, 2, '2011-07-14', '2011-07-17', 1105.00, 11102),
(5153, 2, '2011-07-22', '2011-07-24', 930.00, 14256),
(5154, 2, '2011-07-24', '2011-08-04', 1832.73, 9216),
(5907, 2, '2011-07-24', '2011-08-04', 687.27, 9216),
(5910, 2, '2011-08-11', '2011-08-14', 1140.00, 13633),
(5911, 2, '2011-08-15', '2011-08-16', 380.00, 17770),
(5915, 2, '2011-08-25', '2011-08-29', 1350.00, 17719),
(5916, 2, '2011-08-30', '2011-09-01', 740.00, 16813);

Answer 1

You’re on the right lines. You need to join your query with a table of the months for which you want data, which can either be permanent or (as shown in my example below) created dynamically in a UNION subquery:

SELECT   YEAR(month.d),
         MONTHNAME(month.d),
         SUM(1 + DATEDIFF( -- add 1 because start&finish on same day is still 1 day
           LEAST(Checkout, LAST_DAY(month.d)), GREATEST(Checkin, month.d)
         )) AS days
FROM     bookingdata
  RIGHT JOIN (
                   SELECT 20110101 AS d
         UNION ALL SELECT 20110201 UNION ALL SELECT 20110301
         UNION ALL SELECT 20110401 UNION ALL SELECT 20110501
         UNION ALL SELECT 20110601 UNION ALL SELECT 20110701
         UNION ALL SELECT 20110801 UNION ALL SELECT 20110901
         UNION ALL SELECT 20111001 UNION ALL SELECT 20111101
         UNION ALL SELECT 20111201
  ) AS month ON
             Checkin <= LAST_DAY(month.d)
         AND month.d <= Checkout
GROUP BY month.d

See it on sqlfiddle.