问题
I'm running the following MySQL query to find cars that don't have manuals (and have black wheels, etc.)
SELECT `cars`.* FROM `cars`
INNER JOIN wheels ON cars.id = wheels.car_id
LEFT OUTER JOIN manuals ON cars.id = manuals.car_id
WHERE (cars.created_at > '2010-09-09'
AND wheels.color = 'Black'
AND wheels.created_at < '2011-01-05'
AND manuals.car_id IS NULL)
The results of the query look correct, but it returns the car with id 27 twice. How do I change the query so that all results are unique (no duplicates)?
回答1:
You could try SELECT DISTINCT
instead of SELECT
回答2:
add group by cars.id
at the end of the query
EG:
SELECT `cars`.* FROM `cars`
INNER JOIN wheels ON cars.id = wheels.car_id
LEFT OUTER JOIN manuals ON cars.id = manuals.car_id
WHERE (cars.created_at > '2010-09-09'
AND wheels.color = 'Black'
AND wheels.created_at < '2011-01-05'
AND manuals.car_id IS NULL)
GROUP BY cars.id
回答3:
Presuming that cars.id
is a unique primary key, one of those joins is causing the Cartesian product. That is to say: either wheels
or manuals
contain more than one match for cars.id = 27
.
Subqueries are often a good tool for eliminating Cartesian products. The example query below shows two methods of using subqueries.
The first subquery ensures that we're only looking at cars with black wheels where that record was created before 01/05/2011. The
GROUP BY
clause ensures that we only return one record perw.car_id
.The second subquery (sometimes called a correlated subquery) ensures that there is no manual found for each car in the main query.
Not tested, but conveys the idea:
SELECT `cars`.*
FROM `cars`
JOIN (
SELECT w.car_id
FROM wheels w
WHERE w.color = 'Black'
AND w.created_at < '2011-01-05'
GROUP BY w.car_id
) wheels
ON cars.id = wheels.car_id
WHERE
cars.created_at > '2010-09-09'
AND
NOT EXISTS (SELECT m.car_id FROM manuals m WHERE m.car_id = cars.id)
来源:https://stackoverflow.com/questions/4663181/mysql-how-to-return-unique-distinct-results