Here's my try

[gniquil]

No description provided.

[gniquil]

@twosevenzero, good call on the duplication. I refactored a bit and added a bit of comments. Hopefully the tests will be more clear

[briantemple]

@gniquil I really liked your solution - very slick!

I was just curious about doing the analysis in two phases. It looks like the first phase collects the speeches by line to the speakers, and then the second phase counts the lines and aggregates them all up. Instead of copying over all the lines, it seems like speeches and analyze could be combined in to something like:

  def analyze
    @output = Hash.new(0)
    Nokogiri::XML(contents).css("SPEECH").map do |speech|
      @output[speech.at_css("SPEAKER").content] += speech.css("LINE").count
    end

    @output
  end

[gniquil]

@briantemple you are absolutely right! The only reason I did it this way was TDD. I wrote the line parser first to just get an understanding. Then I realized all i need was an inject method. BTW, I think inject might be even better here:

def analyze
  Nokogiri::XML(contents).css("SPEECH").inject(Hash.new(0)) do |recorder, speech|
    recorder[speech.at_css("SPEAKER").content] += speech.css("LINE").count
    recorder
  end
end

Wow, didn't know Hash.new(0) is so cool! Learned something new today! Thanks!

[runkmc]

I think this block is nicely done. You've made what's happening here extremely clear.

[gniquil]

Hi All,

I guess I am over-thinking too much. Anyway, check out the analyze method again. XPATH is pretty scarily good sometimes. I think this is about the best I can do.

Hope you guys like it.

Frank

[r00k]

What about pulling this into a named constant?

[r00k]

Good effort! See my comments inline.

[gniquil]

@r00k Yoda at work!

Thanks Ben for the reply. Your comment on the stub is really valuable. The way I was testing it always leaves me a funny uneasy feeling. I will try the responsibility-double approach.