In Ruby 1.9.2 on Rails 3.0.3, I'm attempting to test for object equality between two Friend (class inherits from ActiveRecord::Base) objects.
The objects are equal, but the test fails:
Failure/Error: Friend.new(name: 'Bob').should eql(Friend.new(name: 'Bob'))
expected #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
got #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
(compared using eql?)
Just for grins, I also test for object identity, which fails as I'd expect:
Failure/Error: Friend.new(name: 'Bob').should equal(Friend.new(name: 'Bob'))
expected #<Friend:2190028040> => #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
got #<Friend:2190195380> => #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
Compared using equal?, which compares object identity,
but expected and actual are not the same object. Use
'actual.should == expected' if you don't care about
object identity in this example.
Can someone explain to me why the first test for object equality fails, and how I can successfully assert those two objects are equal?
Rails deliberately delegates equality checks to the identity column. If you want to know if two AR objects contain the same stuff, compare the result of calling #attributes on both.
Take a look at the API docs on the == (alias eql?) operation for ActiveRecord::Base
Returns true if comparison_object is the same exact object, or comparison_object is of the same type and self has an ID and it is equal to comparison_object.id.
Note that new records are different from any other record by definition, unless the other record is the receiver itself. Besides, if you fetch existing records with select and leave the ID out, you’re on your own, this predicate will return false.
Note also that destroying a record preserves its ID in the model instance, so deleted models are still comparable.
If you want to compare two model instances based on their attributes, you will probably want to exclude certain irrelevant attributes from your comparison, such as: id, created_at, and updated_at. (I would consider those to be more metadata about the record than part of the record's data itself.)
This might not matter when you are comparing two new (unsaved) records (since id, created_at, and updated_at will all be nil until saved), but I sometimes find it necessary to compare a saved object with an unsaved one (in which case == would give you false since nil != 5). Or I want to compare two saved objects to find out if they contain the same data (so the ActiveRecord == operator doesn't work, because it returns false if they have different id's, even if they are otherwise identical).
My solution to this problem is to add something like this in the models that you want to be comparable using attributes:
def self.attributes_to_ignore_when_comparing
[:id, :created_at, :updated_at]
end
def identical?(other)
self. attributes.except(*self.class.attributes_to_ignore_when_comparing.map(&:to_s)) ==
other.attributes.except(*self.class.attributes_to_ignore_when_comparing.map(&:to_s))
end
Then in my specs I can write such readable and succinct things as this:
Address.last.should be_identical(Address.new({city: 'City', country: 'USA'}))
I'm planning on forking the active_record_attributes_equality gem and changing it to use this behavior so that this can be more easily reused.
Some questions I have, though, include:
Does such a gem already exist??
What should the method be called? I don't think overriding the existing == operator is a good idea, so for now I'm calling it identical?. But maybe something like practically_identical? or attributes_eql? would be more accurate, since it's not checking if they're strictly identical (some of the attributes are allowed to be different.)...
attributes_to_ignore_when_comparing is too verbose. Not that this will need to be explicitly added to each model if they want to use the gem's defaults. Maybe allow the default to be overridden with a class macro like ignore_for_attributes_eql :last_signed_in_at, :updated_at
Comments are welcome...
Update: Instead of forking the active_record_attributes_equality, I wrote a brand-new gem, active_record_ignored_attributes, available at http://github.com/TylerRick/active_record_ignored_attributes and http://rubygems.org/gems/active_record_ignored_attributes
META = [:id, :created_at, :updated_at, :interacted_at, :confirmed_at]
def eql_attributes?(original,new)
original = original.attributes.with_indifferent_access.except(*META)
new = new.attributes.symbolize_keys.with_indifferent_access.except(*META)
original == new
end
eql_attributes? attrs, attrs2
I created a matcher on RSpec just for this type of comparison, very simple, but effective.
Inside this file:
spec/support/matchers.rb
You can implement this matcher...
RSpec::Matchers.define :be_a_clone_of do |model1|
match do |model2|
ignored_columns = %w[id created_at updated_at]
model1.attributes.except(*ignored_columns) == model2.attributes.except(*ignored_columns)
end
end
After that, you can use it when writing a spec, by the following way...
item = create(:item) # FactoryBot gem
item2 = item.dup
expect(item).to be_a_clone_of(item2)
# True
Useful links:
https://relishapp.com/rspec/rspec-expectations/v/2-4/docs/custom-matchers/define-matcher
https://github.com/thoughtbot/factory_bot
If like me you're looking for a Minitest answer to this question then here's a custom method that asserts that the attributes of two objects are equal.
It assumes that you always want to exclude the id, created_at, and updated_at attributes, but you can override that behaviour if you wish.
I like to keep my test_helper.rb clean so created a test/shared/custom_assertions.rb file with the following content.
module CustomAssertions
def assert_attributes_equal(original, new, except: %i[id created_at updated_at])
extractor = proc { |record| record.attributes.with_indifferent_access.except(*except) }
assert_equal extractor.call(original), extractor.call(new)
end
end
Then alter your test_helper.rb to include it so you can access it within your tests.
require 'shared/custom_assertions'
class ActiveSupport::TestCase
include CustomAssertions
end
Basic usage:
test 'comments should be equal' do
assert_attributes_equal(Comment.first, Comment.second)
end
If you want to override the attributes it ignores then pass an array of strings or symbols with the except arg:
test 'comments should be equal' do
assert_attributes_equal(
Comment.first,
Comment.second,
except: %i[id created_at updated_at edited_at]
)
end
Related
In Ruby 1.9.2 on Rails 3.0.3, I'm attempting to test for object equality between two Friend (class inherits from ActiveRecord::Base) objects.
The objects are equal, but the test fails:
Failure/Error: Friend.new(name: 'Bob').should eql(Friend.new(name: 'Bob'))
expected #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
got #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
(compared using eql?)
Just for grins, I also test for object identity, which fails as I'd expect:
Failure/Error: Friend.new(name: 'Bob').should equal(Friend.new(name: 'Bob'))
expected #<Friend:2190028040> => #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
got #<Friend:2190195380> => #<Friend id: nil, event_id: nil, name: 'Bob', created_at: nil, updated_at: nil>
Compared using equal?, which compares object identity,
but expected and actual are not the same object. Use
'actual.should == expected' if you don't care about
object identity in this example.
Can someone explain to me why the first test for object equality fails, and how I can successfully assert those two objects are equal?
Rails deliberately delegates equality checks to the identity column. If you want to know if two AR objects contain the same stuff, compare the result of calling #attributes on both.
Take a look at the API docs on the == (alias eql?) operation for ActiveRecord::Base
Returns true if comparison_object is the same exact object, or comparison_object is of the same type and self has an ID and it is equal to comparison_object.id.
Note that new records are different from any other record by definition, unless the other record is the receiver itself. Besides, if you fetch existing records with select and leave the ID out, you’re on your own, this predicate will return false.
Note also that destroying a record preserves its ID in the model instance, so deleted models are still comparable.
If you want to compare two model instances based on their attributes, you will probably want to exclude certain irrelevant attributes from your comparison, such as: id, created_at, and updated_at. (I would consider those to be more metadata about the record than part of the record's data itself.)
This might not matter when you are comparing two new (unsaved) records (since id, created_at, and updated_at will all be nil until saved), but I sometimes find it necessary to compare a saved object with an unsaved one (in which case == would give you false since nil != 5). Or I want to compare two saved objects to find out if they contain the same data (so the ActiveRecord == operator doesn't work, because it returns false if they have different id's, even if they are otherwise identical).
My solution to this problem is to add something like this in the models that you want to be comparable using attributes:
def self.attributes_to_ignore_when_comparing
[:id, :created_at, :updated_at]
end
def identical?(other)
self. attributes.except(*self.class.attributes_to_ignore_when_comparing.map(&:to_s)) ==
other.attributes.except(*self.class.attributes_to_ignore_when_comparing.map(&:to_s))
end
Then in my specs I can write such readable and succinct things as this:
Address.last.should be_identical(Address.new({city: 'City', country: 'USA'}))
I'm planning on forking the active_record_attributes_equality gem and changing it to use this behavior so that this can be more easily reused.
Some questions I have, though, include:
Does such a gem already exist??
What should the method be called? I don't think overriding the existing == operator is a good idea, so for now I'm calling it identical?. But maybe something like practically_identical? or attributes_eql? would be more accurate, since it's not checking if they're strictly identical (some of the attributes are allowed to be different.)...
attributes_to_ignore_when_comparing is too verbose. Not that this will need to be explicitly added to each model if they want to use the gem's defaults. Maybe allow the default to be overridden with a class macro like ignore_for_attributes_eql :last_signed_in_at, :updated_at
Comments are welcome...
Update: Instead of forking the active_record_attributes_equality, I wrote a brand-new gem, active_record_ignored_attributes, available at http://github.com/TylerRick/active_record_ignored_attributes and http://rubygems.org/gems/active_record_ignored_attributes
META = [:id, :created_at, :updated_at, :interacted_at, :confirmed_at]
def eql_attributes?(original,new)
original = original.attributes.with_indifferent_access.except(*META)
new = new.attributes.symbolize_keys.with_indifferent_access.except(*META)
original == new
end
eql_attributes? attrs, attrs2
I created a matcher on RSpec just for this type of comparison, very simple, but effective.
Inside this file:
spec/support/matchers.rb
You can implement this matcher...
RSpec::Matchers.define :be_a_clone_of do |model1|
match do |model2|
ignored_columns = %w[id created_at updated_at]
model1.attributes.except(*ignored_columns) == model2.attributes.except(*ignored_columns)
end
end
After that, you can use it when writing a spec, by the following way...
item = create(:item) # FactoryBot gem
item2 = item.dup
expect(item).to be_a_clone_of(item2)
# True
Useful links:
https://relishapp.com/rspec/rspec-expectations/v/2-4/docs/custom-matchers/define-matcher
https://github.com/thoughtbot/factory_bot
If like me you're looking for a Minitest answer to this question then here's a custom method that asserts that the attributes of two objects are equal.
It assumes that you always want to exclude the id, created_at, and updated_at attributes, but you can override that behaviour if you wish.
I like to keep my test_helper.rb clean so created a test/shared/custom_assertions.rb file with the following content.
module CustomAssertions
def assert_attributes_equal(original, new, except: %i[id created_at updated_at])
extractor = proc { |record| record.attributes.with_indifferent_access.except(*except) }
assert_equal extractor.call(original), extractor.call(new)
end
end
Then alter your test_helper.rb to include it so you can access it within your tests.
require 'shared/custom_assertions'
class ActiveSupport::TestCase
include CustomAssertions
end
Basic usage:
test 'comments should be equal' do
assert_attributes_equal(Comment.first, Comment.second)
end
If you want to override the attributes it ignores then pass an array of strings or symbols with the except arg:
test 'comments should be equal' do
assert_attributes_equal(
Comment.first,
Comment.second,
except: %i[id created_at updated_at edited_at]
)
end
I am creating an active record object like so:
shop = ShopifyShop.create(shopify_domain: shopDomain, primary_domain: shopObj.domain, shopify_token: token, user_id: userId) if !shop
When the an object with the same primary_domain is created, the record is not stored in the database and the object is created with no primary key
so when i do:
p shop
i get #<ShopifyShop id: nil, shopify_domain: "***.myshopify.com", shopify_token: "***", created_at: nil, updated_at: nil, user_id: 45, primary_domain: "***.myshopify.com">
I checked to see if it was related to constrain maybe but theres only a primary key constraint in the table, i am not sure what is causing this.
Anyone know what could be happening?
One of the validations you have for your shop caused your object to not be saved. create returns the object regardless of whether or not it was saved.
In order to investigate further:
shop.persisted?
# => false
shop.errors
# => will display errors of your object
I'm pretty sure that your object is simply invalid and therefore was not saved to database. That's why it has no id. You can check if there are any errors by:
shop.errors.any?
shop.errors.messages # => display errors
If you want to raise RecordInvalid error if validations fail, use create! instead of create.
I am manually creating objects in the rails console using Model.new(<attributes here>). Is there an easy way to list out which attributes a model will require me to include in order for the .save call to succeed?
I am running rails 4.2.3
You can get an array of validators using Model.validators. You'll have to parse this in some way to extract those validations for presence, something like:
presence_validated_attributes = Model.validators.map do |validator|
validator.attributes if validator.is_a?(ActiveRecord::Validations::PresenceValidator)
end.compact.flatten
I found a simpler way to accomplish the same thing:
When you do a failed create you can check the error message on the object.
# app/models/price.rb
class Price < ActiveRecord::Base
validates_presence_of :value
end
# in console
p = Price.new()
=> #<Price id: nil, created_at: nil, updated_at: nil, value: nil>
p.save
=> false
p.errors.messages
=> {:value=>["can't be blank"]}
In case you the mandatory attributes with error messages
book = Book.new
book.valid?
book.errors.messages
In case you just want the name of attributes without an error message
book = Book.new
book.valid?
book.errors.messages.keys
I have these factories setup:
FactoryGirl.define do
factory :product do
name { Faker::Commerce.product_name }
price { Faker::Commerce.price }
image { Faker::Internet.url }
end
factory :new_product, parent: :product do
name nil
price nil
image nil
end
factory :string_product, parent: :product do
price { Faker::Commerce.price.to_s }
end
end
Why do I want to use :string_product? Well, although the price attribute is of datatype float at the database level, occasionally I want to build a Factory with all of the attributes as strings.
This is so I can build the factory and then run expectations against its attributes when they are passed into the params hash. (All params from the URL are strings)
However, in the rails console:
> FactoryGirl.build :string_product
=> #<Product:0x00000007279780 id: nil, name: "Sleek Plastic Hat", price: 43.54, image: "http://blick.name/moie", created_at: nil, updated_at: nil>
As you can see, price is still being saved as a string.
An experiment to attempt to see what's going on:
...
factory :string_product, parent: :product do
price { "WHY ARE YOU NOT A STRING?" }
end
...
results in:
=> #<Product:0x000000077ddfa0 id: nil, name: "Awesome Steel Pants", price: 0.0, image: "http://rogahn.com/kavon", created_at: nil, updated_at: nil>
My string is now converted to the float 0.0
How do I prevent this behavior? If I want to have one of my attributes as a string, especially when I'm only building it I should be allowed to. Is there a FactoryGirl configuration where I can stop this happening? Exactly the same thing happens with the Fabrication gem, so I'm guessing this is something to do with the model? My Product model is literally empty right now...no validations or anything, so how can that be? The only way FactoryGirl knows price is a float is because it has that datatype on the database level.
Anyway, this is really annoying, if someone could show me how to let me write strings to my Factory's attributes I would be very appreciative. I could use .to_s in the spec itself but I want to keep my specs clean as possible and thought factories would be a great place to keep this configuration...
Is there a fabrication library that would let me do this?
Just some more experimentation:
> "WHY ARE YOU NOT A STRING".to_f
=> 0.0
Okay, so somewhere, in rails or in factorygirl, to_f is being called on my beloved string. Where? And how do I stop it?
With fabrication you need to use attributes_for to generate a hash representation of your object. It will bypass the ActiveRecord model entirely so nothing should be coerced.
Fabricate.attributes_for(:string_product)
Don't mind me, I fricked up my attribute names :(
This is entirely possible, using the exact syntax I used - you just need to be able to spell!
I can't seem to get this to work, and it seems like a common enough scenario that there must be a solution, but I'm not having any luck with the correct terminology to get a helpful Google result.
I want to do this:
u = User.first
u.clients.find_or_create_by_email('example#example.com')
With the effect that a new Client is created with user_id = u.id.
Can I get the nice dynamic finders through a has_many relationship? If not, why?
Thanks :)
This
u = User.first
u.clients.find_or_create_by_email('example#example.com')
works if you have has_many relationship set. However, it won't raise validation error if you have any validations set on your Client object and it will silently fail if the validation fails.
You can check the output in your console when you do
u.clients.find_or_create_by_email('example#example.com') # => #<Client id: nil, email: 'example#example.com', name: nil, user_id: 1, another_attribute: nil, active: true, created_at: nil, updated_at: nil>
and the user_id will be set but not the id of client because the validation has failed and the client is not created
So this should create the client only if you pass all the required attributes of client object and the validation for client object has passed successfully.
So lets say your client model has validation on name as well apart from email then you should do
u.clients.find_or_create_by_email_and_name('example#example.com', 'my_name') #=> #<Client id: 1, email: 'example#example.com', name: 'my_name', user_id: 1, another_attribute: nil, active: true, created_at: "2009-12-14 11:08:23", updated_at: "2009-12-14 11:08:23">
This is entirely possible, using the exact syntax I used - you just need to be able to spell!