我得到了那些模型(简化):
User(id: Int, name: String)
Restaurant(id: Int, ownerId: Int, name: String)
Employee(userId: Int, restaurantId: Int)
当我使用此查询时:
for {
r <- Restaurants
e <- Employees
if r.ownerId === userId || (e.userId === userId && e.restaurantId === r.id)
} yield r
转换为:
select x2."id", x2."owner_id", x2."name" from "restaurants" x2, "employees" x3 where (x2."owner_id" = 2) or ((x3."user_id" = 2) and (x3."restaurant_id" = x2."id"))
到目前为止没有问题。但是当我插入这些数据时:
User(1, "Foo")
User(2, "Fuu")
Restaurant(1, 2, "Fuu")
Restaurant(2, 1, "Foo")
Restaurant(3, 1, "Bar")
Employee(2, 2)
Employee(2, 3)
然后尝试查询,我得到了这个结果:
List(Restaurant(1, 2, "Fuu"), Restaurant(1, 2, "Fuu"), Restaurant(2, 1, "Foo"), Restaurant(3, 1, "Bar))
我不明白为什么Restaurant(1, 2, "Fuu")
出现2次。
(我正在使用org.h2.Driver
和网址jdbc:h2:mem:play
)
我错过了什么吗?
答案 0 :(得分:1)
交叉连接很难;您对SQL查询的要求是:
-- A Cartesian product of all of the rows in restaurants and employees
Employee.user_id | Employee.restaurant_id | Restaurant.name | Restaurant.owner_id
2 | 2 | Fuu | 2
2 | 3 | Fuu | 2
2 | 2 | Foo | 1
2 | 3 | Foo | 1
2 | 2 | Bar | 1
2 | 3 | Bar | 1
-- Filtering out those where the owner != 2
Employee.user_id | Employee.restaurant_id | Restaurant.name | Restaurant.owner_id
2 | 2 | Fuu | 2
2 | 3 | Fuu | 2
-- And combining that set with the set of those where the employee's user_id = 2
-- and the restaurant's ID is equal to the employee's restaurant ID
Employee.user_id | Employee.restaurant_id | Restaurant.name | Restaurant.owner_id
2 | 2 | Foo | 1
2 | 2 | Bar | 1
改为使其显式为左连接:
for {
(r, e) <- Restaurants leftJoin Employees on (_.id = _.restaurantId)
if r.ownerId === userId || e.userId === userId
} yield r
或者,使用exists
使其更清晰:
for {
r <- Restaurants
if r.ownerId === userId ||
Employees.filter(e => e.userId === userId && e.restaurantId === r.id).exists
} yield r