Comments for bmabey's email-spec  

Hey Rajan,

Issues should generally be in https://github.com/bmabey/email-spec/issues , not comments (people can find and help you a lot faster as it could get lost here).

But to answer your question: Assuming you have the latest email spec gem installed (from github, not rubyforge, run these to check:

gem sources -a https://gems.github.com
sudo gem uninstall email_spec
sudo gem install bmabey-email_spec

, and you have updated RAILS_ROOT/features/steps/email_steps.rb (an old file can cause problems, so run the following

script/generate email_spec

That done, using the latest version, when you type 'And I...' , the 'I' assumes the email example@example.com , so if you're email was sent to a different email account, then you'll need to use this syntax:

'And "someone123@mydomain.com" should receive 1 email'

and further commands should use 'When they...', 'And they...', 'Then they...' (third person).

Alternatively, you can edit RAILS_ROOT/features/step_definitions/email_steps.rb and change the value of current_email_address to that of the email you expect to send most emails too. You might also adjust the method to use the last logged in staff account, like I have.

last_email_address || (@current_staff && @current_staff.email) || "example@example.com"

Hope this helps. I have no problems using the latest email_spec if everything is configured correctly.

And I should receive 1 email # features/step_definitions/email_steps.rb:51

  expected: 1,
       got: 0 (using ==)
   Diff:
  @@ -1,2 +1,2 @@
  -1
  +0
   (Spec::Expectations::ExpectationNotMetError)
  features/register.feature:15:in `And I should receive 1 email'

This thing fails no matter what I do. Any ideas?

Regards
Rajan

Awesome – I was just about to provide this fix. Now I don’t have to. Thanks

Actually, I’m wrong. It was intended to do both. My features use the way that was commented out.

For some reason github won’t let me fork atm, but here is what I propose.

    def parse_email_for_link(email, text_or_regex)
      email.should have_body_text(text_or_regex)
      
      url = parse_email_for_explicit_link(email, text_or_regex) 
      url ||= parse_email_for_anchor_text_link(email, text_or_regex)
      raise "No link found matching #{text_or_regex.inspect} in #{email}" unless url
      url
    end
    
    # e.g. confirm in https://confirm
    def parse_email_for_explicit_link(email, regex)
      regex = /#{Regexp.escape(regex)}/ unless regex.is_a?(Regexp)
      url = links_in_email(email).detect { |link| link =~ regex }
      URI::parse(url).request_uri if url
    end
    
    # e.g. Click here in  <a href="https://confirm">Click here</a>
    def parse_email_for_anchor_text_link(email, link_text)
      email.body =~ %r{<a[^>]*href=['"]?([^'"]*)['"]?[^>]*?>[^<]*?#{link_text}[^<]*?</a>}
      URI.split($~[1])[5..-1].compact!.join("?").gsub("&amp;", "&")
      # sub correct ampersand after rails switches it (https://dev.rubyonrails.org/ticket/4002) 
    end

This is not how this method is intended to work.

The goal was for it to look in the link text, not in the url itself. links_in_email just returns the urls.

Ahh.. nm.. I see what you mean.. I’ll remove the dupe now. Thanks for pointing that out.

I’ve actually switched to using Jeweler.. so I’m not sure if this still applies. Mind taking a look at the current commit?

One of the files in here needs to be removed. Dupe of email_spec.gemspec

I would change that to amount.to_i Have had issues where it was giving me expected 1 got “1” and the like

Your Gem appears to be working now.

1. => Successfully installed bmabey-email_spec-0.0.5

Yeah, having it in the README may confuse people. I’ll go ahead and remove it from there. That said, I think the actual step may be worth leaving in in case people want to clear it mid-scenario. I will often have steps in features that do nothing except commicate intent better.. people may want to use it in this fashion as well.

By putting ‘current_email_address’ into a module and then extending the world object we prevent polluting Object. While probably not a big issue, since this is just test code, something about having that method on ALL of my objects bugs me. By using Cucumber’s world hook the only object that will have the method is the World object that the steps are executed in.

I agree about the ‘And no emails have been sent’ being unnecessary in the readme. Maybe just remove it altogether? I think it will be rare that anyone needs to use that step. The situation would involve having to clear the e-mails within a scenario, which I’m not sure why anyone would want to do.

Also, why put current_email_address in a module?