Help with Writing Meaningful Specs

[andypearson]

Hello all,

In yet another attempt to learn Ruby on Rails and Rspec I have started writing a simple life streaming app with will aggregate feeds from several places and save them in a database for later use.

An internet search eventually led me to the following method for looping through the feeds in the database, getting the contents of the URL and then passing this into another model to prepare and save it.

def self.cache_all
  feeds = self.find(:all)
   
  for feed in feeds
    xml = REXML::Document.new Net::HTTP.get(URI.parse(feed.url))
    xml.elements.each '//item' do |item|
      Item.prepare_and_save(feed, item)
    end
  end
end

Problems are now beginning to arise when trying to write the specs for this method. I started from scratch slowly building up the specs but it has led to an awful amount of mocks and stubs, and I am not even sure whether they are asking the correct things of the method.

Can anyone give me some pointers on how to write useful, meaningful specs for this method?

The other thing I have found is that I seem to have incorrectly stubbed xml.elements.each meaning that the contents of the block are never called, how should I be specifying this behavior?

I have pastied the complete code for the Feed model, the spec and a little helper at: http://pastie.caboo.se/203941

Thanks in advance,

Andy

[Tom Stuart]

On 27 May 2008, at 12:44, andypearson wrote:
>    xml = REXML::Document.new Net::HTTP.get(URI.parse(feed.url))

It won't make the problem go away, but you can certainly reduce the  
blizzard of intermediate stubs by pulling this chain out into its own  
method (e.g. fetch_xml_from_url) and stubbing that method once in your  
cache_all spec.

> it has led to an awful amount of mocks and stubs, and I am not even  
> sure whether they
> are asking the correct things of the method.
> Can anyone give me some pointers on how to write useful, meaningful  
> specs
> for this method?

What part of its behaviour do you care about? (Refactoring into more  
fine-grained methods might help again here.) It seems as though all  
you're interested in is that 1. the feed URLs get fetched from the  
database, 2. the feed contents get fetched from the URLs, and 3.  
Item.prepare_and_save gets called for each item in each feed. So,  
ideally, that's what your specs for this method should be saying.

> The other thing I have found is that I seem to have incorrectly  
> stubbed
> xml.elements.each meaning that the contents of the block are never  
> called,
> how should I be specifying this behavior?

Use #and_yield, not #and_return. (And you might as well use an actual  
mock, rather than [@element], since you're stubbing the only method  
you plan to call on it.)

Cheers,
-Tom
_______________________________________________
rspec-users mailing list
rspec-users@...
http://rubyforge.org/mailman/listinfo/rspec-users

[Ashley Moran-4]

On 27 May 2008, at 12:44, andypearson wrote:

> def self.cache_all
>  feeds = self.find(:all)
>
>  for feed in feeds
>    xml = REXML::Document.new Net::HTTP.get(URI.parse(feed.url))
>    xml.elements.each '//item' do |item|
>      Item.prepare_and_save(feed, item)
>    end
>  end
> end
>
> Problems are now beginning to arise when trying to write the specs  
> for this
> method. I started from scratch slowly building up the specs but it  
> has led
> to an awful amount of mocks and stubs, and I am not even sure  
> whether they
> are asking the correct things of the method.
>
> Can anyone give me some pointers on how to write useful, meaningful  
> specs
> for this method?

Hi Andy

I think the problem comes from two things - trying to specify the  
details of an algorithm instead of checking it transforms the data  
correctly, and having too much logic in a class method.

To take some Feed-specific logic out, try making an accessor method,  
such as Feed#uri (I can't think of a good name for it), then you could  
replace one line with this:

   xml = REXML::Document.new = Net::HTTP.get(feed.uri)

Alternatively, go one step further and have a method in Feed that does  
the Net::HTTP.get, so you could write this:

   xml = feed.xml_source

Obviously you will still need the corresponding specs in the Feed  
instance methods, but at least you've encapsulated the logic related  
to fetching the URI/XML.

The other thing is the "item" elements.  I don't know what  
xml.elements.each yields, but I'm guessing it's either plain text  
(XML) or some REXML object.  If it's plain XML fine, but if it's an  
REXML object, you don't want your code depending on that (you're tying  
yourself down to a specific library).  Instead, consider parsing the  
XML in Feed.cache_all.  I'm imagining something like this:

   feed.prepare_and_save_item(:item_attr_1 => "foo",
                              :item_attr_2 => "bar",
                              ...)


> The other thing I have found is that I seem to have incorrectly  
> stubbed
> xml.elements.each meaning that the contents of the block are never  
> called,
> how should I be specifying this behavior?

Just noticed that Tom replied and pointed out the issue.  But if you  
refactor the code like I describe above, you can avoid the issue.  
Just stub feed.xml_source (or whatever) to return some sample valid  
feed XML, and check that Feed.cache_all correctly extracts the item  
data.

I can't say that my way is better or worse than the way you are going  
about it.  But I (where possible) I always test the output of third-
party libraries rather than my use of them.  It avoids loads of stubs  
that don't, in any case, prove you're using the library correctly so  
you can see immediately if your code is likely to work in the wild.

Hope this helps

Ashley


--
http://www.patchspace.co.uk/
http://aviewfromafar.net/



_______________________________________________
rspec-users mailing list
rspec-users@...
http://rubyforge.org/mailman/listinfo/rspec-users