Question

我有一长串的关联，连接，排序等，最终从我的一个导轨模型中选择。在一天结束时，我需要结果是独特的和分类的。我不关心SELECT语句中使用哪些列，ORDER BY中出现的内容等等（这些都根据用户选择的过滤选项而改变），我只关心顶级模型/表在查询中是唯一的（基于id）。

对于后台，widgets是主要表格，我们正在加入widget_steps，这是在Rails 3中（公司正在努力尽快升级，但这就是他们所做的＆＃39} ;现在再次坚持）

以下是正在生成的查询和错误：

PG::InvalidColumnReference: ERROR:  for SELECT DISTINCT, ORDER BY expressions must appear in select list
LINE 1: ...completed_at" IS NULL)) ORDER BY sequential DESC, widget_s...
                                                             ^
: SELECT  DISTINCT "widgets".* FROM "widgets" INNER JOIN "widget_steps" ON "widget_steps"."widget_id" = "widgets"."id" INNER JOIN "widget_steps" "active_steps_widgets" ON "active_steps_widgets"."id" = "widgets"."active_widget_step_id" WHERE "widgets"."account_id" = 1 AND "widgets"."completed_at" IS NULL AND (("widgets"."sequential" = 't' AND "widget_steps"."assigned_to" = 5 AND "widget_steps"."id" = "widgets"."active_widget_step_id" AND "widget_steps"."completed_at" IS NULL) OR ("widgets"."sequential" = 'f' AND "widget_steps"."assigned_to" = 5 AND "widget_steps"."completed_at" IS NULL)) ORDER BY sequential DESC, widget_steps.name ASC LIMIT 10 OFFSET 0
Completed 500 Internal Server Error in 52.3ms

ActiveRecord::StatementInvalid - PG::InvalidColumnReference: ERROR:  for SELECT DISTINCT, ORDER BY expressions must appear in select list
LINE 1: ...completed_at" IS NULL)) ORDER BY sequential DESC, widget_s...
                                                             ^
: SELECT  DISTINCT "widgets".* FROM "widgets" INNER JOIN "widget_steps" ON "widget_steps"."widget_id" = "widgets"."id" INNER JOIN "widget_steps" "active_steps_widgets" ON "active_steps_widgets"."id" = "widgets"."active_widget_step_id" WHERE "widgets"."account_id" = 1 AND "widgets"."completed_at" IS NULL AND (("widgets"."sequential" = 't' AND "widget_steps"."assigned_to" = 5 AND "widget_steps"."id" = "widgets"."active_widget_step_id" AND "widget_steps"."completed_at" IS NULL) OR ("widgets"."sequential" = 'f' AND "widget_steps"."assigned_to" = 5 AND "widget_steps"."completed_at" IS NULL)) ORDER BY sequential DESC, widget_steps.name ASC LIMIT 10 OFFSET 0:

为什么这是一件事？ postgres认为这是多么含糊不清？为什么像这样的查询在MySQL中总能正常工作，但会使postgres窒息。

我试过了：

在链的末尾指定.select([everything mentioned in the order by]).uniq
在链的末尾指定.uniq而不进行自定义选择
编写一些自定义AREL以尝试将所有这些嵌入到子查询中，然后在此之外执行.uniq或.order（无法使其正常工作）
在postgres之外做.uniq（这会因为分页而中断...你最终会得到一些页面上只有1或2个项目，因为重复删除了）
哭闹

Answer 1

您需要将Int添加到所选列的列表中：

widget_steps.name

这不应该改变查询的逻辑，并且可以正常工作。

在Rails中，您可以使用SELECT DISTINCT "widgets".*, "widget_steps.name" FROM "widgets" INNER JOIN "widget_steps" ON "widget_steps"."widget_id" = "widgets"."id" INNER JOIN "widget_steps" "active_steps_widgets" ON "active_steps_widgets"."id" = "widgets"."active_widget_step_id" WHERE "widgets"."account_id" = 1 AND "widgets"."completed_at" IS NULL AND (("widgets"."sequential" = 't' AND "widget_steps"."assigned_to" = 5 AND "widget_steps"."id" = "widgets"."active_widget_step_id" AND "widget_steps"."completed_at" IS NULL) OR ("widgets"."sequential" = 'f' AND "widget_steps"."assigned_to" = 5 AND "widget_steps"."completed_at" IS NULL)) ORDER BY sequential DESC, widget_steps.name ASC LIMIT 10 OFFSET 0方法设置所选列的列表：

select

希望这会有所帮助。

Answer 2

建议：

您可以在Widget类中添加has_many :steps，然后使用Widget.includes（：steps）执行带有Order子句的查询，该子句按widget_steps列进行排序

Answer 3

在某些情况下（例如has_and_belongs_to_many类型关系）可能会发生此错误的另一个选项是取消范围限制：

#Code
data <- data[!data$id %in% id_to_remove,]

这应该让您完成所有这些操作

class Resource < ApplicationRecord
  has_and_belongs_to_many :things
  scope :with_blue_things, -> { joins(:things).where(:things => {:color => :blue}).unscope(:order).distinct }

PG :: InvalidColumnReference：错误：对于SELECT DISTINCT，ORDER BY表达式必须出现在选择列表中

3 个答案: