Question

I’ve got a date property for a model called ‘transaction’, when I save it I want to check if it’s within a range and if not then set the date to the previous transaction.

It’s working but I cannot seem to get my test working, it keeps bombing out when adding the error. Is there a way to stub the ‘errors.add’ line out so I can isolate my test to check if the tran_date value has changed correctly?

  def tran_date_within_financial_year
    range = ledger.fin_start..ledger.fin_end
    unless range === tran_date
      tran = Transaction.all.order("created_at").last
      self.tran_date = tran.tran_date
      errors.add(:tran_date, "is outside financial year")
    end
  end

Here is my test

context "date is empty" do
  it "should set date to previous transaction" do
    @prev_tran = Transaction.make!(:tran_date => Date.new(2012,02,03))
    @tran = Transaction.make!(:tran_date => "")
    @tran.save
    @tran.reload.tran_date.should eq(@prev_tran.tran_date)
  end
end

The result when running rspec is:

  1) Transaction Saving a transaction date is empty should set date to previous transaction
     Failure/Error: @tran = Transaction.make!(:ledger => ledger, :tran_date => "")
     ActiveRecord::RecordInvalid:
       Validation failed: Tran date is outside financial year

However if I comment out the error.add function then my test passes. So it is setting the value correctly to the previous transaction date. The problem is the errors.add stops the test all together.

Answer 1

Well, the spec simply emphasises that you can’t save an invalid object.

And the object is invalid because you add errors to it.

These specs should pass:

context "date is empty" do
  it "should set date to previous transaction" do
    @prev_tran = Transaction.make!(:tran_date => Date.new(2012,02,03))
    @tran = Transaction.make(:tran_date => "")
    @tran.should_not be_valid
    @tran.errors.should have_key :tran_date
    @tran.tran_date.should eq(@prev_tran.tran_date)
  end
end