On Thu, Sep 8, 2011 at 3:03 PM, Justin Beck <justinb...@mac.com> wrote:
> I have a class that expects some input via 'gets'. I found that, when
> testing, I can do something as simple as this:
>
> @foo.stub!(:gets) { "stuff\n" }
>
> However that only works in the most simple case. In my case the class asks
> a series of questions that require a yes/no answer culminating in some end
> state. The class then asks if the user would like to go through the
> questions again and waits for a response of yes/no. If I want to test a
> single run through the questions (answering 'yes' to all of them) and I stub
> 'gets' to always reply with 'yes' then I have no way of breaking out of the
> loop. The test I'm running dies and complains of the stack being too deep.
>
> So, my question is this:
>
> How do folks typically go about testing classes that take input from stdin?
> I can post more code if necessary but this is a pretty simple question to
> which, I'm sure, there is a reasonable solution.
>
> Thanks!
>
> Justin
> _______________________________________________
> rspec-users mailing list
> rspec-users@rubyforge.org
> http://rubyforge.org/mailman/listinfo/rspec-users
Two ways to approach this. On the unit level, you can stub "gets" and return
a different value each time:
stubs(:gets).and_return('yes', 'no', 'yes', 'yes', 'no')
Each time "gets" is called it will return the next value.
On a more high level, I would recommend using the aruba gem with RSpec. Here
is how I've done it in my own code:
https://github.com/justinko/reverse-polish-calculator/tree/master/spec/acceptance
Let us know if any of that works.
_______________________________________________
rspec-users mailing list
rspec-users@rubyforge.org
http://rubyforge.org/mailman/listinfo/rspec-users
