计算 MySQL 5.6 中重叠日期范围的最大数量

Question

I am creating a vehicle rental application. I was trying find overlapping booking in given dates. I come across a similar question Count maximum number of overlapping date ranges in MySQL but this only answered for MySQL 8.0.

I modified above question for my problem. I need solution for MySQL 5.6 without window function.

create table if not exists BOOKING
(
    start datetime null,
    end datetime null,
    vehicle_id varchar(255),
    id int auto_increment
        primary key
);

INSERT INTO BOOKING (start, end, vehicle_id)
VALUES
('2020-02-06 10:33:55', '2020-02-07 10:34:41', 111),
('2020-02-08 10:33:14', '2020-02-10 10:33:57', 111),
('2020-02-06 10:32:55', '2020-02-07 10:33:32', 222),
('2020-08-06 10:33:03', '2020-02-11 10:33:12', 111),
('2020-02-12 10:31:38', '2020-02-15 10:32:41', 111),
('2020-02-09 09:48:44', '2020-02-10 09:50:37', 222);

Suppose If I give start as 2020-02-05 and end as 2020-02-11, this should return 2, as maximum usage of vehicle 111, is 2 from 2020-02-06 to 2020-02-10

5  6  7  8  9  10  11 
   <-->  <------>
   <---------------->     (Vehicle Id 111, ANSWER should be 2)

for vehicle id 222, (For same query)
5  6  7  8  9  10  11 
   <-->     <--->         (Vehicle Id 222, ANSWER should be 1)

So overall output I am expecting for input start(2020-02-05) and end(2020-02-11)

+---------+-------+
| vehicle | usage |
+---------+-------+
| 111     | 2     |
| 222     | 1     |
+---------+-------+

I need solution which covers followings

• on passing start_date and end_date my query will return data only for that range
• If no data found should return vehicle_id 0

Answer 1

SELECT vehicle_id vehicle, MAX(cnt) `usage`
FROM ( SELECT booking.vehicle_id, timepoints.dt, COUNT(*) cnt
       FROM booking
       JOIN ( SELECT start dt FROM booking
              UNION ALL
              SELECT `end` FROM booking ) timepoints ON timepoints.dt BETWEEN booking.start AND booking.`end`
       GROUP BY booking.vehicle_id, timepoints.dt ) subquery
GROUP BY vehicle_id;

fiddle

PS. Misprint in 4th row is corrected.

Answer 2

The maximum number of overlaps occurs when a rental starts (although it might persist for a period of time, this is all you care about).

You can calculate this for each start using:

SELECT b.vehicle_id, b.start, COUNT(*)
FROM booking b JOIN
     booking b2
     ON b2.vehicle_id = b.vehicle_id AND
        b2.start <= b.start AND
        b2.end > b.start
WHERE b.start <= $end and b.end >= $start
GROUP BY b.vehicle_id, b.start;

Then for the maximum:

SELECT vehicle_id, MAX(overlaps)
FROM (SELECT b.vehicle_id, b.start, COUNT(*) as overlaps
      FROM booking b JOIN
           booking b2
           ON b2.vehicle_id = b.vehicle_id AND
              b2.start <= b.start AND b2.end > b.start
      GROUP BY b.vehicle_id, b.start
     ) b
GROUP BY vehicle_id;

Here is a db<>fiddle.

Performance on this type of query is never going to be as good as using window functions. However, an index on (vehicle_id, start, end) would help.