我正在构建一个rake任务,以便在两个数据库之间迁移一些数据。那些具有完全相同的结构。我的任务是这样的:
namespace :oab_nexus_migration do
task :start, [:oab_user, :nexus_user] => :environment do |t, args|
oab_account = User.find_by_username(args[:oab_user]).main_account
oab_trials = oab_account.trials.includes(:parts)
oab_schedules = oab_account.schedules
oab_movements = oab_account.movements
oab_annotations = oab_account.annotations
oab_hearings = oab_account.hearings
oab_publications = oab_account.publications
oab_tasks = oab_account.tasks
oab_people = oab_account.people.includes(:addresses, :internet_addresses, :phones)
ActiveRecord::Base.establish_connection(:other_database)
...
More code here
end
end
任务执行时:
RAILS_ENV=production rake oab_nexus_migration:start[user_one, user_two]
从数据库一中检索到我需要的所有信息后,我需要将其插入到数据库二中。但是真的很奇怪。例如,如果我在p oab_trials
之前调用establish_connection
(或任何其他变量),则所有值都在那里,由大数组表示。但是如果我尝试在establish_connection
之后调用它,则返回的值是一个空数组。好像ActiveRecord正在重置我以前定义的所有变量。
这里发生了什么?
答案 0 :(得分:2)
oab_account.trials.includes(:parts)
不会存储值,它会存储对延迟评估的ActiveRecord::Relation
的引用。
因此,ActiveRecord
不是重置变量的值,而是当您连接到第二个数据库时,您将引用更改为空集合。
答案 1 :(得分:2)
您引用的相关集合是ActiveRecord::Relation
的惰性求值实例。如果您的数据库连接在流中更改,则延迟评估不起作用。
在切换连接之前,您需要先将这些关系加载到内存中。在关系上调用.to_a
以强制执行此操作。
oab_account = User.find_by_username(args[:oab_user]).main_account
oab_trials = oab_account.trials.includes(:parts).to_a
oab_schedules = oab_account.schedules.to_a
oab_movements = oab_account.movements.to_a
oab_annotations = oab_account.annotations.to_a
oab_hearings = oab_account.hearings.to_a
oab_publications = oab_account.publications.to_a
oab_tasks = oab_account.tasks.to_a
oab_people = oab_account.people.includes(:addresses, :internet_addresses, :phones).to_a