twitter_vs_hpricot_test.rb

require 'rubygems'
require 'open-uri'
require 'hpricot'
require 'test/unit'

class HpricotTest < Test::Unit::TestCase
  def assert_selector_is_found(page, selector, message = nil)
    message ||= "Expected to find elements to match selector: #{selector}"
    elements = (page / selector)
    assert(elements.any?, message)
  end
  
  def assert_selector_not_found(page, selector, message = nil)
    message ||= "Expected no elements to match selector: #{selector}"
    elements = (page / selector)
    assert(elements.empty?, message)
  end
  
  def test_url_based_page_should_find_element
    page = Hpricot(open("http://twitter.com/Squeegy"))
    assert_selector_is_found(page, 'a[@class="tweet-url web"]')
    assert_selector_is_found(page, 'a[@class~="tweet-url"]')
    assert_selector_is_found(page, 'a[@class~="web"]')
    assert_selector_is_found(page, 'a.tweet-url')
    assert_selector_is_found(page, 'a.web')
    assert_selector_is_found(page, 'a.tweet-url.web')
    
    assert_selector_not_found(page, 'a[@class="web tweet-url"]')
  end
  
  def test_string_based_page_should_find_element
    page = Hpricot(<<-HTML)
      <div>
        <div>
          <a href="http://foobar.com" class="tweet-url web">Foobar</a>
        </div>
        <div>
          <a href="http://barbaz.com" class="tweet-url web">Barbaz</a>
        </div>
        <div>
          <a href="http://bazfoo.com" class="tweet-url web">Bazfoo</a>
        </div>
      </div>
    HTML
    
    assert_selector_is_found(page, 'a[@class="tweet-url web"]')
    assert_selector_is_found(page, 'a[@class~="tweet-url"]')
    assert_selector_is_found(page, 'a[@class~="web"]')
    assert_selector_is_found(page, 'a.tweet-url')
    assert_selector_is_found(page, 'a.web')
    assert_selector_is_found(page, 'a.tweet-url.web')
    
    assert_selector_not_found(page, 'a[@class="web tweet-url"]')
  end
  
  def test_should_list_all_attributes
    page = Hpricot(open("http://twitter.com/Squeegy"))
    link = page.at('a.web.tweet-url')
    attributes = link.attributes
    
    assert_kind_of(Hash, attributes)
    assert_equal(attributes['class'], 'tweet-url web')
    assert(attributes.has_key?('href'))
    assert(attributes.has_key?('rel'))
    assert(attributes.has_key?('target'))
  end
end