Question

不确定为什么Ecto会返回错误的结果。我可能会搞乱连接

这是我的Ecto.Query

from u in User, 
        join: c in assoc(u, :companies), 
        join: r in assoc(c, :roles), 
        join: a in assoc(r, :assets), 
        where: u.id == ^2 and a.id == ^1, 
        group_by: [c.id, u.id, r.id], 
        select: {u.email, u.id, r.name, c.name}

它生成：

    SELECT u0."email", u0."id", r2."name", c1."name" 
FROM "users" AS u0 
INNER JOIN "access_lists" AS a4 ON a4."user_id" = u0."id" 
INNER JOIN "companies" AS c1 ON a4."company_id" = c1."id" 
INNER JOIN "access_lists" AS a5 ON a5."company_id" = c1."id" 
INNER JOIN "roles" AS r2 ON a5."role_id" = r2."id" 
INNER JOIN "access_lists" AS a6 ON a6."role_id" = r2."id" 
INNER JOIN "assets" AS a3 ON a6."asset_id" = a3."id"
 WHERE ((u0."id" = $1) AND (a3."id" = $2)) 
GROUP BY c1."id", u0."id", r2."id" [2, 1]

此查询的结果不正确

这是正确的查询：

SELECT u0.email, u0.id, r2.name, c1.name, a6.asset_name , a6.id as asset_id
FROM users AS u0 
INNER JOIN access_lists AS a4 ON a4.user_id = u0.id 
INNER JOIN assets AS a6 ON a4.asset_id = a6.id 
INNER JOIN companies AS c1 ON a4.company_id = c1.id 
INNER JOIN roles AS r2 ON a4.role_id = r2.id 
WHERE ((u0.id = 2) AND (a6.id = 1)) 
Group By u0.id, r2.id, c1.id, a6.id

这是架构：

accesslist中：

  schema "access_lists" do
        belongs_to :user, Db.User
        belongs_to :role, Db.Role
        belongs_to :asset, Db.Asset
        belongs_to :project, Db.Project
        belongs_to :company, Db.Company

        timestamps()
    end

用户：

many_to_many :roles, Db.Role, join_through: Db.AccessList
many_to_many :assets, Db.Asset, join_through: Db.AccessList
many_to_many :projects, Db.Project, join_through: Db.AccessList        
many_to_many :companies, Db.Company, join_through: Db.AccessList

资产：

many_to_many :users, Db.User, join_through: Db.AccessList
many_to_many :companies, Db.Company, join_through: Db.AccessList 
many_to_many :roles, Db.Role, join_through: Db.AccessList 
many_to_many :projects, Db.Project, join_through: Db.AccessList

公司：

    many_to_many :users, Db.User, join_through: Db.AccessList
    many_to_many :assets, Db.Asset, join_through: Db.AccessList 
    many_to_many :roles, Db.Role, join_through: Db.AccessList 
    many_to_many :projects, Db.Project, join_through: Db.AccessList

作用：

    many_to_many :users, Db.User, join_through: Db.AccessList
    many_to_many :assets, Db.Asset, join_through: Db.AccessList
    many_to_many :projects, Db.Project, join_through: Db.AccessList  
    many_to_many :companies, Db.Company, join_through: Db.AccessList

这是一张图：

Answer 1

我用过

from u in User, 
        join: acl in AccessList, on: acl.user_id == u.id,
        join: c in Company, on: acl.company_id == c.id, 
        join: r in Role, on: acl.role_id == r.id, 
        join: a in Asset, on: acl.asset_id == a.id, 
        where: u.id == ^2 and a.id == ^1, 
        group_by: [c.id, u.id, r.id], 
        select: {u.email, u.id, r.name, c.name}

这有效，我得到了正确的结果，但我不确定为什么。

为什么我必须定义每个连接和

为什么我不能使用join assoc

Answer 2

没有看到模式，很难说，但我可以做一些猜测。由于AccessList是User与公司，角色和资产之间的联接表，因此Ecto不太了解在后续联接中重复使用先前的AccessList查询。

尝试显式映射出来：

from u in User, 
  join: acl in assoc(u, :access_lists),
  join: c in assoc(acl, :companies),
  join: r in assoc(acl, :roles),
  join: a in assoc(acl, :assets),
  where: u.id == ^2 and a.id == ^1, 
  group_by: [c.id, u.id, r.id], 
  select: {u.email, u.id, r.name, c.name}

Ecto.Query连接生成了太多的INNER JOIN

2 个答案: