Question

在activeadmin中，我们需要在Score模型的索引页上应用一种复杂的过滤器。即，我们希望在Score表中获得结果，以使他们的ExportOrder具有选定的DistributionChain。我们从以下模型开始：

class Score < ApplicationRecord
  has_and_belongs_to_many :export_orders, join_table: :scores_export_orders
end

class ExportOrder < ApplicationRecord
  belongs_to :distribution_chain
  has_and_belongs_to_many :scores, join_table: :scores_export_orders
end

class DistributionChain < ApplicationRecord
  has_many :export_orders
end

在schema.rb中（摘录）：

  create_table "scores_export_orders", id: false, force: :cascade do |t|
    t.integer "score_id"
    t.integer "export_order_id"
    t.index ["export_order_id"], name: "index_scores_export_orders_on_export_order_id", using: :btree
    t.index ["score_id"], name: "index_scores_export_orders_on_score_id", using: :btree
  end

  create_table "export_orders", force: :cascade do |t|
    t.integer  "distribution_chain_id"
  end

在得分Activeadmin中：

ActiveAdmin.register Score
  filter :delivery_distr_chain, as: :select, collection: DistributionChain.all
end

过滤器的范围在得分模型中定义：

class Score < ApplicationRecord
  ...
  scope :delivery_distr_chain, -> (input) {
    self.joins(:export_orders).where('export_orders.distribution_chain_id = ?', input)
  }

  def self.ransackable_scopes(auth_object = nil)
    [:delivery_distr_chain]
  end
end

以下是错误：

ActionView::Template::Error (PG::DuplicateAlias: ERROR:  table name "scores_export_orders" specified more than once
: SELECT COUNT(DISTINCT count_column) FROM (SELECT  DISTINCT "scores"."id" AS count_column FROM "scores" INNER JOIN "scores_export_orders" ON "scores_export_orders"."score_id" = "scores"."id" INNER JOIN "export_orders" ON "export_orders"."id" = "scores_export_orders"."export_order_id" LEFT JOIN scores_export_orders ON scores.id = scores_export_orders.score_id WHERE (NOT (scores_export_orders.score_id IS NULL)) AND (export_orders.distribution_chain_id = '2') LIMIT $1 OFFSET $2) subquery_for_count):
    1: insert_tag renderer_for(:index)

activerecord (5.0.7) lib/active_record/connection_adapters/postgresql_adapter.rb:600:in `async_exec'

我感觉到查询未正确编写。您能告诉我们我们哪里失败了吗？

编辑：值得注意的是，在rails控制台中执行查询与在Score模型的范围内执行查询之间的区别。

如果我在Rails控制台中执行相同的查询，那么它将正常工作而不会出错。 Score.joins(:export_orders).where('export_orders.distribution_chain_id = ?', '2').to_sql的输出是：

"SELECT \"scores\".* FROM \"scores\" INNER JOIN \"scores_export_orders\" ON \"scores_export_orders\".\"score_id\" = \"scores\".\"id\" INNER JOIN \"export_orders\" ON \"export_orders\".\"id\" = \"scores_export_orders\".\"export_order_id\" WHERE (export_orders.distribution_chain_id = '2') ORDER BY \"scores\".\"created_at\" DESC"

在Score模型的范围内，完全相同的查询失败，而.to_sql的输出为：

"SELECT DISTINCT \"scores\".* FROM \"scores\" INNER JOIN \"scores_export_orders\" ON \"scores_export_orders\".\"score_id\" = \"scores\".\"id\" INNER JOIN \"export_orders\" ON \"export_orders\".\"id\" = \"scores_export_orders\".\"export_order_id\" LEFT JOIN scores_export_orders ON scores.id = scores_export_orders.score_id WHERE (NOT (scores_export_orders.score_id IS NULL)) AND (export_orders.distribution_chain_id = '2') ORDER BY \"scores\".\"id\" desc"

主要区别在于范围LEFT JOIN scores_export_orders ON scores.id = scores_export_orders.score_id引入的左联接，它可以再次指定scores_export_orders表。我不知道为什么要在这里而不是在控制台上引入它。

Answer 1

已解决。我不知道我已经在ActiveAdmin for Scores中声明了一个作用域，如下所示：

scope :exported, -> { joins("LEFT JOIN scores_export_orders ON scores.id = scores_export_orders.score_id").where.not("scores_export_orders.score_id IS NULL").distinct

在ActiveAdmin中称为

  controller do
    def scoped_collection
      end_of_association_chain.exported
    end
    ...
  end

现在，我已经重写了分数模型中的代码：

class Score < ApplicationRecord
  ...
  scope :exported, -> { joins("INNER JOIN scores_export_orders ON scores.id = scores_export_orders.score_id INNER JOIN export_orders ON export_orders.id = scores_export_orders.export_order_id").where.not("scores_export_orders.score_id IS NULL").distinct }
  ...
  scope :delivery_distr_chain, -> (input) {
    self.where('export_orders.distribution_chain_id = ?', input)
  }

  def self.ransackable_scopes(auth_object = nil)
    [:delivery_distr_chain]
  end
end

这样我就可以在exported范围内进行双重联接，而无需在过滤器范围内再次联接。

Activeadmin自定义过滤器失败

1 个答案: