如何急切加载从 rake 任务调用的方法

Question

我有这样的任务：

namespace :company do
  task :update, [:code] => :environment do |t, args|
    company = Company.find_or_create_by(code: args[:code])
    company.update_from_local_data
  end
end

这是Company 类。

class Company < ActiveRecord::Base
  has_many :items
  def update_from_local_data
    data = YAML.load(File.read(ENV['COMPANY_DATA_FILE']))
    update_items(data)
  end
  def update_items(item_array)
    item_array.each do |value|
      item = items.find_or_initialize_by(name: value[:name])
      item.update_attributes(value)
    end
  end
end

我确认此代码有很多SELECT SQL 查询。

在控制器中我可以处理它，但我如何从 rake 任务中使用预先加载？

编辑

感谢 Uri 的 cmets 我了解了如何提高将多个数据保存到数据库的性能，但我仍然无法为多个项目调用 find_or_initialize_by。

我找到ActiveRecord.import 的:on_duplicate_key_update 选项，但它只能与MySQL 一起使用，而我正在使用PostgreSQL。

编辑 2

为了解释问题是什么，我创建了一个example project。

这是Company#update_from_local_data 的结果。 我不想SELECT 查询每个Items。

我怎样才能更有效地编写它？

c = Company.first
c.update_from_local_data
  Item Load (0.2ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item0' LIMIT 1  [["company_id", 1]]
   (0.1ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item1' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item2' LIMIT 1  [["company_id", 1]]
   (0.1ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item3' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item4' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item5' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item6' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item7' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item8' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
  Item Load (0.1ms)  SELECT  "items".* FROM "items"  WHERE "items"."company_id" = ? AND "items"."name" = 'item9' LIMIT 1  [["company_id", 1]]
   (0.0ms)  begin transaction
   (0.0ms)  commit transaction
=> [{:name=>"item0"}, {:name=>"item1"}, {:name=>"item2"}, {:name=>"item3"}, {:name=>"item4"}, {:name=>"item5"}, {:name=>"item6"}, {:name=>"item7"}, {:name=>"item8"}, {:name=>"item9"}]

Answer 1

您说要加快 find_or_initialize_by 步骤。

def update_items( items ) 
  # 'items' is an array of attributes hashes

  ActiveRecord::Base.transaction do

    names_array = items.map{ |attributes| attributes[:name] }
    existing_records = Company.where(name: names_array)

    records_by_name = existing_record.each_with_object({}) do |record, hash|
      name = record.name
      hash[name] = record
    end

    items.each do |attributes|
      name = attributes[:name]
      record = records_by_name[name] || Company.new

      # with validations and callbacks:
      #   record.update_attributes(attributes)

      # without validations:
      #   attributes.each{ |k, v| record[k] = v }
      #   record.save(validate: false)


      # without validations or callbacks:
      #   If you're using an older version of Rails,
      #   you can use record.save(:update_without_callbacks)
      #   For recent versions, you'll need to either write SQL-
      #   or disable all callbacks with skip_callbacks and then re-enable-
      #   them with set_callbacks
    end
  end
end

基本上，您可以一次性找到所有现有记录，而不是为每个名称执行单独的搜索查询。

Answer 2

1. 您可以使用update_all，它将返回更新的条目数。如果更新了 0 个条目，则创建新记录

   def update_items(item_array)
     item_array.each do |value|
       entries_updated = items.where(name: value[:name]).update_all(value)
       if entries_updated == 0
         items.create!(value)
       end
   end
   

   请注意，如果create! 无法创建记录，它将引发错误。您可能希望仅使用 create 并自行处理验证错误。

2. 根据您在聊天items = load_all(item_array); items.update_all 上建议的界面，您可以采取的另一种方法是

   def update_items(item_array)
     grouped = item_array.group_by {|i| i[:name] }
     items.where(name: grouped.keys).each do |item|
       data = grouped[item.name]
       item.assign_attributes(data)
       item.save! if item.changed? 
     end
   end
   

   如果不是所有项目都经常更改，这将减少查询，但如果Company 有数千个项目，则可能会很慢，但您可以将item_array 分成更小的组，然后执行该操作。请注意，无法生成一个更新语句来根据不同的条件更改多条记录。