finder_sql不使用Rails解析字符串

时间:2011-12-17 14:23:16

标签: ruby-on-rails ruby-on-rails-3 postgresql activerecord finder-sql

我遇到的问题是我用于finder_sql的查询在被移交给PostgreSQL之前没有被正确解析,导致数据库语法错误。

为了说明问题我刚才使用了示例代码:

  

http://api.rubyonrails.org/classes/ActiveRecord/Associations/ClassMethods.html

我只将class_name更改为"User",因为我没有人模型,但这并不重要。

has_many :subscribers, :class_name => "User", :finder_sql =>
'SELECT DISTINCT people.* ' +
'FROM people p, post_subscriptions ps ' +
'WHERE ps.post_id = #{id} AND ps.person_id = p.id ' +
'ORDER BY p.first_name'

当我使用它时,我收到以下错误:

User Load (0.3ms)  SELECT DISTINCT people.* FROM people p, post_subscriptions ps WHERE
ps.post_id = #{id} AND ps.person_id = p.id ORDER BY p.first_name
PGError: ERROR:  Syntaxerror near »{« 
LINE 1: ...ople p, post_subscriptions ps WHERE ps.post_id = #{id} AND p...
                                                         ^

正如您所看到的,#{id}未被对象的id替换,从而引发PostgreSQL错误。

环境

  • Rails 3.1
  • RVM
  • PostgreSQL 9.1
  • Ubuntu 11.10
  • Ruby 1.9.2p290(2011-07-09修订版32553)[x86_64-linux]

3 个答案:

答案 0 :(得分:10)

我认为你真正想要的是:

has_many :posts, :finder_sql =>
    proc {"SELECT p.* from posts p join topics t on p.topic_id = t.id where t.id=#{id}"}

从Rails 3.1开始,你必须使用proc而不是字符串来使用像#{id}这样的字段。

请在此处查看问题:https://github.com/rails/rails/issues/3920

答案 1 :(得分:7)

:finder_sql的文档非常不完整,示例代码已损坏。正如您所发现的那样:

has_many :subscribers, :class_name => "User", :finder_sql =>
  'SELECT DISTINCT people.* ' +
  'FROM people p, post_subscriptions ps ' +
  'WHERE ps.post_id = #{id} AND ps.person_id = p.id ' +
  'ORDER BY p.first_name'

将无法工作,并且基于ActiveRecord源,无法正常工作。如果您查看来源,则会see things like this

def custom_finder_sql
  interpolate(options[:finder_sql])
end

然后interpolate执行此操作:

def interpolate(sql, record = nil)
  if sql.respond_to?(:to_proc)
    owner.send(:instance_exec, record, &sql)
  else
    sql
  end
end

所以如果你的:finder_sql只是一个字符串(例如在例子中),那么它按原样使用,根本没有插值,你最终会得到破坏的SQL。如果你想要插值,那么你必须让interpolate进入第一个分支,这样你就需要一个lamba用于:finder_sql和一个双引号的字符串在lambda中,以便{{1}将工作:

#{id}

这应该进入has_many :subscribers, :class_name => "User", :finder_sql => ->(record) do "SELECT DISTINCT people.* " + "FROM people p, post_subscriptions ps " + "WHERE ps.post_id = #{id} AND ps.person_id = p.id " + "ORDER BY p.first_name" end 内的第一个分支,以便评估instance_exec调用,并在相关对象的上下文中插入字符串。我不确定interpolate何时不会record,所以您可能需要这样做:

nil

虽然我们在这里,但请使用显式连接条件而不是隐式连接条件:

has_many :subscribers, :class_name => "User", :finder_sql => ->(record) do
      record = self if(record.nil?)
      "SELECT DISTINCT people.* " +
      "FROM people p, post_subscriptions ps " +
      "WHERE ps.post_id = #{record.id} AND ps.person_id = p.id " +
      "ORDER BY p.first_name"
end

您发现的有关单/双引号和has_many :subscribers, :class_name => "User", :finder_sql => ->(record) do record = self if(record.nil?) "SELECT DISTINCT people.* " + "FROM people p " + "JOIN post_subscriptions ps on p.id = ps.person_id " + "WHERE ps.post_id = #{record.id} " + "ORDER BY p.first_name" end 的博客:

  

http://tamersalama.com/2007/05/17/finder_sql-single-vs-double-quotes/

已过期,似乎不适用于Rails 3+。上面的摘录来自3.1,但您看到的行为表明代码和行为可能在3.0中发生了变化,但文档未更新。

答案 2 :(得分:0)

我知道这不是你希望听到的,但问题是你应该让ActiveRecord为你做这项工作。

你真正想要解决的问题是拥有这三个文件:

# user.rb
class User < ActiveRecord::Base
  self.table_name = 'people'
  has_many :post_subscriptions
end
# post_subscription.rb
class PostSubscription < ActiveRecord::Base
  belongs_to :user
  belongs_to :post
end
# post.rb
class Post < ActiveRecord::Base
  has_many :post_subscriptions
  has_many :subscribers, :through => :post_subscriptions, :source => :user
end

然后你根本不需要编写任何SQL。只需致电@post.subscribers即可获得订阅用户的完整列表。