Measures distance of any wikipedia article to Philosophy. Based on http://xkcd.org/903/ alt text

wikistance.rb

# WikiStance. A Wikipedia distance meter.
# Based on the alt text of http://xkcd.org/903/
# Gets a Wikipedia URL and measure the distance of this page to the Philosophy article, clicking on links not in parens
# neither italics
#
# Author::    Alejandro Fernández  (mailto:[email protected])
# Copyright:: Copyright (c) 2011 Alejandro Fernández
# License::   GPL
# 
# =Usage=
# 
# require 'wikistance'
# 
# url = 'http://en.wikipedia.org/wiki/Scrubs_%28TV_series%29'
# ws = WikiStance.new(url)
# ws.trace  # Go through all the pages until we reach Philosophy
# ws.distance # => 22
# ws.breadcrumbs # => ["List of characters on Scrubs", "NBC", "United States", ..., "Philosophy"]

require 'rubygems'
require 'mechanize'

class WikiStance
  
  attr_reader :title, :breadcrumbs

  def initialize(url)

    if url =~ /^http:\/\/en\.wikipedia\.org\/wiki\/(.*?)/
      @url = url

      # Wikipedia returns 403 with the default user agent
      @agent = Mechanize.new
      @agent.user_agent_alias = 'Mac Safari'
      
      self.reset
      
    else
      raise ArgumentError, "You should use a valid wikipedia link"
    end

  end
  
  # Resets the class
  def reset
    @page        = @agent.get(@url)
    @breadcrumbs = []
    @title       = page_title
    @breadcrumbs << @title
    true
  end
  
  # Gets the current @page title
  def page_title
    @page.at('#firstHeading').text()
  end
  
  # Go through the pages to calculate distance
  def trace
    
    while page_title != 'Philosophy'
      click_first_link
      title = page_title
      
      # Avoid entering in an infinite loop
      if @breadcrumbs.include?(title)
        raise "We are repeating ourselves! We already visited \"#{title}\""
      end
      @breadcrumbs << title
    end
    
    true
    
  end
  
  def distance
    # Breadcrumbs hold the initial page. If we start in philosophy the distance should be 0
    @breadcrumbs.length - 1
  end
  
  private
  
  def click_first_link
    first_link = nil
    
    # div#bodyContent is where wikipedia shows article's content
    # The starting text is direct child of div#bodyContent. This way we avoid <p> inside TOCs and other texts.
    # We also avoid Disambiguation and other wikipedia texts, (which all of them contains links in italics) because
    # they are in <div> instead of <p>
    @page.search('#bodyContent > p').each do |p|
      
      # Links between parens should not be clicked
      # I tried using a regex with lookbehind to know if a link has an opening parenthesis before, but ruby doesn't
      # support them, so I will just remove all text between parens...
      text = p.to_html.gsub(/\((?:.*?)\)/, '').gsub(/<i>(?:.*?)<\/i>/, '')

      # ...and then get the first link.
      first_link = text.match(/<a(?:.*?)href\=\"[^#](.*?)\"(?:.*?)\/a>/)
      break unless first_link.nil?
    end

    raise "Oops! seems that \"#{page_title}\" has no links" if first_link.nil?
    @page = @page.links_with(:href => /#{first_link[1]}/).first.click
  end


end

wikistance_spec.rb

require 'wikistance'

valid_url      = 'http://en.wikipedia.org/wiki/List_of_characters_on_Scrubs'
philosophy_url = 'http://en.wikipedia.org/wiki/Philosophy' 


describe WikiStance do
  
  it "should raise error with invalid URL" do
    lambda { WikiStance.new('wadus') }.should raise_error(ArgumentError)
  end
  
  it "should create a valid object with a valid URL" do
    wiki = WikiStance.new(valid_url)
    wiki.should be_instance_of WikiStance
  end

  it "should fetch the title of the page" do
    wiki = WikiStance.new(valid_url)
    wiki.title.should == "List of characters on Scrubs"
  end
  
  it "should have no distance from Philosophy" do
    wiki = WikiStance.new(philosophy_url)
    wiki.trace
    wiki.distance.should == 0
  end
  
  it "should have one of distance from Modern Philosophy" do
    wiki = WikiStance.new('http://en.wikipedia.org/wiki/Modern_philosophy')
    wiki.trace
    wiki.distance.should == 1
  end
  
  it "should have trails for Moder Philosophy" do
    wiki = WikiStance.new('http://en.wikipedia.org/wiki/Modern_philosophy')
    wiki.trace
    wiki.breadcrumbs.should == ['Modern philosophy', 'Philosophy']
  end
  
  # Greek_language comes back to itself becase of links between parens
  it "should not repeat pages" do
    wiki = WikiStance.new('http://en.wikipedia.org/wiki/Greek_language')
    lambda { wiki.trace }.should_not raise_error(RuntimeError)
  end
  
  # Psychologist first link is an anchor to the same page
  it "should avoid #anchor links" do
    wiki = WikiStance.new('http://en.wikipedia.org/wiki/Psychologist')
    wiki.trace.should be_true
    wiki.distance.should == 12
  end
  
end