如何在 PostgreSQL 中获取不同城市的最新日期

Question

I'm currently trying to do a task on PostgreSQL right now where I need to prepare a list of cities with their date of last reservation which also must include a hotel ID, and a photo.

Here's my SQL query thus far:

SELECT city.name, booking.booking_date as last_booking_date, hotel.id as hotel_id, hotel.photos ->> 0 as hotel_photo
FROM city 
INNER JOIN hotel
    ON city.id = hotel.city_id 
INNER JOIN booking 
    ON booking.hotel_id = hotel.id
WHERE 
    booking.booking_date = (SELECT MAX(booking_date) FROM booking WHERE hotel_id = hotel.id)
ORDER BY city.name ASC;

Now what this does is return the cities (sorted in alphabetical order, and can duplicate) which also contains their booking dates (also sorted from earliest to latest). If many hotels have the same amount of bookings, then I'll have to sort them by ID

Thanks to the answer below, and with some minor tweaks, I was able to return the distinct cities in alphabetical order, but not by their latest date for some cities.

SELECT distinct on (city.name) 
  city.name, 
  booking.booking_date as last_booking_date, 
  hotel.id as hotel_id, 
  hotel.photos ->> 0 as hotel_photo
FROM city 
INNER JOIN hotel
    ON city.id = hotel.city_id 
INNER JOIN booking 
    ON booking.hotel_id = hotel.id
WHERE 
    booking.booking_date = (SELECT MAX(booking_date) 
                            FROM booking
                            WHERE hotel_id = hotel.id)
    AND 
    booking.id = (SELECT MAX(id) FROM booking WHERE hotel_id = hotel.id)
ORDER BY city.name, hotel.id ASC, booking.booking_date ASC;

This has the following result, with the expected result shown as well.

As you can see, some cities are not sorted by their latest date of booking. I've tried variations of the SQL query above such as:

SELECT distinct on (city.name) 
  city.name, 
  booking.booking_date as last_booking_date, 
  hotel.id as hotel_id, 
  hotel.photos ->> 0 as hotel_photo
FROM city 
INNER JOIN hotel
    ON city.id = hotel.city_id 
INNER JOIN booking 
    ON booking.hotel_id = hotel.id
WHERE 
    booking.booking_date = (SELECT MAX(booking_date) 
                            FROM booking
                            WHERE hotel_id = hotel.id)
    AND 
    booking.hotel_id = (SELECT MAX(hotel_id) FROM booking WHERE hotel_id = hotel.id)
ORDER BY city.name, hotel.id ASC, booking.booking_date ASC;

and

SELECT distinct on (city.name) 
  city.name, 
  booking.booking_date as last_booking_date, 
  hotel.id as hotel_id, 
  hotel.photos ->> 0 as hotel_photo
FROM city 
INNER JOIN hotel
    ON city.id = hotel.city_id 
INNER JOIN booking 
    ON booking.hotel_id = hotel.id
WHERE 
    booking.booking_date = (SELECT MAX(booking_date) 
                            FROM booking
                            WHERE hotel_id = hotel.id)
    AND 
    booking.id = (SELECT MAX(id) FROM booking WHERE hotel_id = hotel.id)
    AND 
    booking.hotel_id = (SELECT MAX(hotel_id) FROM booking WHERE hotel_id = hotel.id)
ORDER BY city.name, hotel.id ASC, booking.booking_date ASC;

But they all show the same result as the one earlier.

With that, how do I get the row with the latest date? For context, this is what the first query returns.

Answer 1

Use DISTINCT ON . It will pick the first element for the subset, according to the ordering in the ORDER BY clause. For example:

SELECT distinct on (city.name) -- changed here
  city.name, 
  booking.booking_date as last_booking_date, 
  hotel.id as hotel_id, 
  hotel.photos ->> 0 as hotel_photo
FROM city 
INNER JOIN hotel
    ON city.id = hotel.city_id 
INNER JOIN booking 
    ON booking.hotel_id = hotel.id
WHERE 
    booking.booking_date = (SELECT MAX(booking_date) 
                            FROM booking
                            WHERE hotel_id = hotel.id)
ORDER BY city.name, booking.booking_date DESC; -- changed here

Answer 2

I believe, you do not need the WHEN subquery. Your ORDER BY statement is slightly off: it first sorts by city, then by hotel and only then choses a date. So, if you modify ORDER BY to be by city.name, date DESC and then hotel_id it should make the work.
Find the code below:

SELECT distinct on (city.name) 
  city.name, 
  booking.booking_date as last_booking_date, 
  hotel.id as hotel_id, 
  hotel.photos ->> 0 as hotel_photo
FROM city 
INNER JOIN hotel
    ON city.id = hotel.city_id 
INNER JOIN booking 
    ON booking.hotel_id = hotel.id
ORDER BY city.name, booking.booking_date DESC, hotel.id ASC;