In Bryan Helmkamp's excellent blog post called "7 Patterns to Refactor Fat ActiveRecord Models", he mentions using Form Objects
to abstract away multi-layer forms and stop using accepts_nested_attributes_for
.
Edit: see below for a solution.
I've almost exactly duplicated his code sample, as I had the same problem to solve:
class Signup
include Virtus
extend ActiveModel::Naming
include ActiveModel::Conversion
include ActiveModel::Validations
attr_reader :user
attr_reader :account
attribute :name, String
attribute :account_name, String
attribute :email, String
validates :email, presence: true
validates :account_name,
uniqueness: { case_sensitive: false },
length: 3..40,
format: { with: /^([a-z0-9\-]+)$/i }
# Forms are never themselves persisted
def persisted?
false
end
def save
if valid?
persist!
true
else
false
end
end
private
def persist!
@account = Account.create!(name: account_name)
@user = @account.users.create!(name: name, email: email)
end
end
One of the things different in my piece of code, is that I need to validate the uniqueness of the account name (and user e-mail). However, ActiveModel::Validations
doesn't have a uniqueness
validator, as it's supposed to be a non-database backed variant of ActiveRecord
.
I figured there are three ways to handle this:
- Write my own method to check this (feels redundant)
- Include ActiveRecord::Validations::UniquenessValidator (tried this, didn't get it to work)
- Or add the constraint in the data storage layer
I would prefer to use the last one. But then I'm kept wondering how I would implement this.
I could do something like (metaprogramming, I would need to modify some other areas):
def persist!
@account = Account.create!(name: account_name)
@user = @account.users.create!(name: name, email: email)
rescue ActiveRecord::RecordNotUnique
errors.add(:name, "not unique" )
false
end
But now I have two checks running in my class, first I use valid?
and then I use a rescue
statement for the data storage constraints.
Does anyone know of a good way to handle this issue? Would it be better to perhaps write my own validator for this (but then I'd have two queries to the database, where ideally one would be enough).