Force controller to use current_user with mocking - ruby-on-rails

I am trying to specify in my RSpec tests that my controller should use current_user.projects.find() instead of Project.find() I am using the Mocha mocking framework and was trying something like this:
controller.current_user.projects.expects(:find).returns(#project)
I have already mocked out controller.stubs(:current_user).returns(#profile)
This test passes with this even when I use the Project.find() implementation. How can I test that my controller is calling off of the correct object?
Edit (adding additional code):
I have Projects and Tasks, Project have many tasks. This is the show method for displaying a task in a project that is owned by current_user
Action in the controller:
def show
#project = current_user.projects.find_by_id(params[:cardset_id])
if #project.nil?
flash[:notice] = "That project doesn't exist. Try again."
redirect_to(projects_path)
else
#task = #project.tasks.find_by_id(params[:id])
end
end
This is the test that is not checking that the cardsets method was called off the current_user object.
Current Test:
context "with get to show" do
context "with valid project" do
before(:each) do
#project = Factory(:project)
#task = Factory(:task)
#profile = #project.profile
ApplicationController.stubs(:require_user).returns(true)
controller.stubs(:current_user).returns(#profile)
Project.stubs(:find_by_id).returns(#project)
#project.tasks.stubs(:find_by_id).returns(#task)
get :show, :project_id => #project.id, :id => #task.id
end
it "should assign task" do
assigns[:task].should_not be_nil
end
it "should assign project" do
assigns[:project].should_not be_nil
end
end
context "with invalid project" do
before(:each) do
Project.stubs(:find_by_id).returns(nil)
get :show, :project_id => #project.id, :id => #task.id
end
it "should set flash" do
flash[:notice].should match(/doesn't exist/i)
end
it "should redirect" do
response.should redirect_to(cardsets_url)
end
end
end

Based on the little you've told us, I think you need:
#profile.expects(:find).returns(#project)

Related

Rspec method is being called 2X, but can't find second time

Here is my controller spec
before do
#order = Order.new
end
it "should call find & assign_attributes & test delivery_start methods" do
Order.should_receive(:find).with("1").and_return(#order)
Order.any_instance.should_receive(:assign_attributes).with({"id"=>"1", "cancel_reason" => "random"}).and_return(#order)
Order.any_instance.should_receive(:delivery_start).and_return(Time.now)
post :cancel, order: {id:1, cancel_reason:"random"}
end
The failure is this:
Failure/Error: Unable to find matching line from backtrace
(#<Order:0x007fdcb03836e8>).delivery_start(any args)
expected: 1 time with any arguments
received: 2 times with any arguments
# this backtrace line is ignored
But I'm not sure why delivery_start is being called twice based on this controller action:
def cancel
#order = Order.find(cancel_params[:id])
#order.assign_attributes(cancel_params)
if (#order.delivery_start - Time.now) > 24.hours
if refund
#order.save
flash[:success] = "Your order has been successfully cancelled & refunded"
redirect_to root_path
else
flash[:danger] = "Sorry we could not process your cancellation, please try again"
render nothing: true
end
else
#order.save
flash[:success] = "Your order has been successfully cancelled"
redirect_to root_path
end
end
I would suggest you test the behavior and not the implementation. While there are cases where you would want to stub out the database doing it in a controller spec is not a great idea since you are testing the integration between your controllers and the model layer.
In addition your test is only really testing how your controller does its job - not that its actually being done.
describe SomeController, type: :controller do
let(:order){ Order.create } # use let not ivars.
describe '#cancel' do
let(:valid_params) do
{ order: {id: '123', cancel_reason: "random"} }
end
context 'when refundable' do
before { post :cancel, params }
it 'cancels the order' do
expect(order.reload.cancel_reason).to eq "random"
# although you should have a model method so you can do this:
# expect(order.cancelled?).to be_truthy
end
it 'redirects and notifies the user' do
expect(response).to redirect_to root_path
expect(flash[:success]).to eq 'Your order has been successfully cancelled & refunded'
end
end
end
end
I would suggest more expectations and returning true or false depending on your use. Consider the following changes
class SomeController < ApplicationController
def cancel
...
if refundable?
...
end
end
private
def refundable?
(#order.delivery_start - Time.now) > 24.hours
end
end
# spec/controllers/some_controller_spec.rb
describe SomeController, type: :controller do
describe '#cancel' do
context 'when refundable' do
it 'cancels and refunds order' do
order = double(:order)
params = order: {id: '123', cancel_reason: "random"}
expect(Order).to receive(:find).with('123').and_return(order)
expect(order).to receive(:assign_attributes).with(params[:order]).and_return(order)
expect(controller).to receive(:refundable?).and_return(true)
expect(controller).to receive(:refund).and_return(true)
expect(order).to receive(:save).and_return(true)
post :cancel, params
expect(response).to redirect_to '/your_root_path'
expect(session[:flash]['flashes']).to eq({'success'=>'Your order has been successfully cancelled & refunded'})
expect(assigns(:order)).to eq order
end
end
end
end
Sorry, this is a very unsatisfactory answer, but I restarted my computer and the spec passed...
One thing that has been a nuisance for me before is that I've forgotten to save the code, i.e., the old version of the code the test is running against called delivery_start twice. But in this case, I definitely checked that I had saved. I have no idea why a restart fixed it...

How can I determine the subject of an rspec controller test?

So I don't have a great reason for needing to know this other than curiosity - the BEST reason - but I'm not sure what's going on here.
Background:
I'm working through the RSpec book and updating the examples.
On Chapter 24 - Rails Controllers there's a test for a messages controller.
## spec/controllers/messages_controller_spec.rb ##
require 'spec_helper'
describe MessagesController do
describe "POST create" do
let(:message) { mock_model(Message).as_null_object }
before do
Message.stub(:new).and_return(message)
end
# Then a bunch of Tests...
context "when the message fails to save" do
before do
message.stub(:save).and_return(false)
post :create
end
it "assigns #message" do
assigns[:message].should eq(message)
end
it "renders the new template" do
response.should render_template("new")
end
end
end
end
This goes along with the messages controller:
## app/controllers/messages_controller.rb ##
class MessagesController < ApplicationController
def create
#message = Message.new(params[:message])
if #message.save
flash[:notice] = "The message was saved successfully"
redirect_to action: "index"
else
render "new"
end
end
end
When I run the tests:
The test passes with response.
it "renders the new template" do
response.should render_template("new")
end
The test also passes with subject.
it "renders the new template" do
subject.should render_template("new")
end
The test Also passes with page
it "renders the new template" do
page.should render_template("new")
end
The test ALSO passes with NOTHING
it "renders the new template" do
should render_template("new")
end
In case it helps anyone make heads or tails of this, the config/routes.rb just has resources :messages
Why do all those tests pass? What am I actually testing? Are 'page', 'subject', and ' ' just synonyms for response?
Does it matter as long as my tests pass?
By default, the subject would be referencing the class, which is the MessagesController.
Not defining a subject in the last test example, will implicitly set the subject to be MessagesController.
From a binding.pry, it appears that subject is an instance of the controller class:
[2] pry(#<RSpec::ExampleGroups::MyController::DescribeString::ContextString>)> subject.is_a? Class
=> false
[3] pry(#<RSpec::ExampleGroups::MyController::DescribeString::ContextString>)> subject.is_a? Users::SessionsController
=> true

Can I have some feedback with rspec when writing controllers specs?

I was wondering if i could have some feedbacks with the controller spec bellow. In fact i'm new when writing specs and controller's spec are way different from model's spec ! So i'm wondering if i may not go in the wrong direction...
subjects_controller.rb
def show
#subject = Subject.find(params[:id])
if #subject.trusted?(current_user)
#messages = #subject.messages
else
#messages = #subject.messages.public
#messages = #messages + #subject.messages.where(:user_ids => current_user.id)
#messages.uniq!
end
# sort the list
#messages = #messages.sort_by(&:created_at).reverse
if !#subject.company.id == current_user.company.id
redirect_to(subjects_path, :notice => "Invalid subject")
end
end
subjects_controller_spec.rb
require 'spec_helper'
describe SubjectsController do
before(:each) do
#subject = mock_model(Subject)
end
context "for signed users" do
before(:each) do
#current_user = sign_in Factory(:user)
end
context "GET #show" do
before(:each) do
Subject.stub!(:find, #subject).and_return(#subject)
end
context "when current_user is trusted" do
before(:each) do
messages = []
company = mock_model(Company)
#subject.should_receive(:trusted?).and_return(true)
#subject.should_receive(:messages).and_return(messages)
#subject.should_receive(:company).and_return(company)
end
it "should render success" do
get :show, :id => #subject
response.should be_success
end
end
context "when current_user is not trusted" do
before(:each) do
messages = []
company = mock_model(Company)
#subject.should_receive(:trusted?).and_return(false)
#subject.should_receive(:messages).and_return(messages)
messages.should_receive(:public).and_return(messages)
#subject.should_receive(:messages).and_return(messages)
messages.should_receive(:where).and_return(messages)
#subject.should_receive(:company).and_return(company)
end
it "should render success" do
get :show, :id => #subject
response.should be_success
end
end
context "when subject's company is not equal to current_user's company" do
# I have no idea of how to implement ==
end
end
end
end
Factories.rb
Factory.define :user do |u|
u.first_name 'Test User' #
u.username 'Test User' #
u.surname 'TheTest' #
u.email 'foo#foobar.com' #
u.password 'please' #
u.confirmed_at Time.now #
end
As far as I can tell you're on the right path. The basic idea is to completely isolate your controller code from model and view in these tests. You appear to be doing that--stubbing and mocking model interaction.
Don't write RSpec controller specs at all. Use Cucumber stories instead. Much easier, and you get better coverage.

How to DRY up RSpec tests shared by different actions in same controller

I have the following tests that I want tested from various actions in the same controller. How can I DRY this up? In the comments below you'll see that the test should call a different method and action depending on which action I'm testing.
shared_examples_for "preparing for edit partial" do
it "creates a new staff vacation" do
StaffVacation.should_receive(:new)
get :new
end
it "assigns #first_day_of_week" do
get :new
assigns(:first_day_of_week).should == 1
end
end
describe "GET new" do
# i want to use 'it_behaves_like "preparing for edit partial"'
# and it should use 'get :new'
end
describe "GET edit" do
# i want to use 'it_behaves_like "preparing for edit partial"'
# but it should use 'get :edit' instead
end
describe "POST create" do
# on unsuccessful save, i want to use 'it_behaves_like "preparing for edit partial"'
# but it should use 'post :create' instead
end
You could do something like this:
shared_examples_for "preparing for edit partial" do
let(:action){ get :new }
it "creates a new staff vacation" do
StaffVacation.should_receive(:new)
action
end
it "assigns #first_day_of_week" do
action
assigns(:first_day_of_week).should == 1
end
end
context 'GET new' do
it_should_behave_like 'preparing for edit partial' do
let(:action){ get :new }
end
end
context 'GET edit' do
it_should_behave_like 'preparing for edit partial' do
let(:action){ get :edit }
end
end
context 'POST create' do
it_should_behave_like 'preparing for edit partial' do
let(:action){ post :create }
end
end
Or, you could use some kind of loop for the examples:
['get :new', 'get :edit', 'post :create'].each do |action|
context action do
it "creates a new staff vacation" do
StaffVacation.should_receive(:new)
eval(action)
end
it "assigns #first_day_of_week" do
eval(action)
assigns(:first_day_of_week).should == 1
end
end
end
One option might be to provide a module mix-in with a method that has your spec inside it.
include Auth # This is your module with your generalized spec inside a method
it "redirects without authentication" do
unauthorized_redirect("get", "new")
end
Then, in our method, we could do a loop through different types of authorization:
module Auth
def unauthorized_redirect(request, action)
[nil, :viewer, :editor].each do |a|
with_user(a) do
eval "#{request} :#{action}"
response.should redirect_to login_path
# whatever other expectations
end
end
end
end

rspec stub current_company

I have a Rails 3 project in which I want to store the current company selected in a session variable.
I'm working with the staff controller spec and would like to stub out current_company for now as I'm isolating my spec example for the staff new controller action.
it "should call current_company" do
company = mock_model(Company, :id => "1")
controller.should_receive(:current_company).and_return(company)
get :new
end
Here is my new action for the staff controller
def new
#staff = Staff.new
#staff.company_id = current_company.id
end
I keep getting error
Failure/Error: get :new
NameError:
undefined local variable or method `current_company' for #<StaffsController:0x000000028d6ad8>
I've also tried just stubbing it out instead of using should_receive
controller.stub!(:current_company).and_return(company)
I get the same error.
Your code looks fine to me, it should work. There must be some other problem we are not seeing. I notice the controller name is "StaffsController" -- is that correct? Double-check the names of the controller and the corresponding spec -- they should be the same.
I think it was bombing out on the 'should be successful' example/test, so I've put my stubbing in a before block.
require 'spec_helper'
describe StaffsController do
describe "GET 'new'" do
let(:staff) { mock_model(Staff, :company_id= => nil)}
let(:company) { mock_model(Company, :id => 1)}
before do
Staff.stub!(:new).and_return(staff)
controller.stub!(:current_company).and_return(company)
end
it "should be successful" do
get :new
response.should be_success
end
it "should call current_company" do
controller.should_receive(:current_company).and_return(company)
get :new
end
end
end
This works for:
class StaffsController < ApplicationController
def new
#staff = Staff.new
current_company.id
end
end

Resources