Scrape the Spirit Halloween store locator and return a sensible array of hashes

spirit-locations.rb

require 'open-uri'
require 'nokogiri'
require 'yaml'

def aggressive_strip text
  text.strip.gsub(/[\s\r\n]+/, ' ')
end

postal_code = 10003
url = "http://checkout.spirithalloween.com/storelocation.aspx?zipPostalCode=#{ postal_code }"
doc = Nokogiri::HTML open(url)

results = []

doc.css('.vStoreLocatorMultipleLabelStyleLocationContainer').each do |node|
  result = {}

  result['location'] = aggressive_strip node.css('.locationName').text
  result['distance'] = aggressive_strip node.css('.locationDistance').text

  node.css('dl').each do |dl|
    term = aggressive_strip dl.css('dt').text.downcase
    value = aggressive_strip dl.css('dd').text

    class_name = dl.css('dt').first.attributes['class']

    if class_name and class_name.value
      value = term
      term = class_name.value
    elsif term == 'address'
      parts = aggressive_strip(dl.css('dd').inner_html).split('<br>').map{ |part| part.strip }
      result['name'] = parts.shift
      result['map'] = parts.pop
      value = parts
    elsif term == ''
      if value =~ /(now )?open(ing soon)?/i
        term = 'open'
        value = !!(value =~ /now open/i)
      end
    end

    if term == 'GalleryStore'
      term = 'type'
    end

    result[term] = value
  end

  if result['map']
    map = Nokogiri::HTML result['map']
    anchor = map.css('a').first
    result['map'] = anchor.attributes['href'].value rescue result['map']

    result['name'] = [result['name']]
    result['name'].unshift aggressive_strip(anchor.text).sub(/get directions to spirit - /i, '')

    begin
      coords = result['map'].scan(/\/place\/([0-9.,-]+)/).flatten.first.split(',')
      result['lat'] = coords[0].to_f
      result['lng'] = coords[1].to_f
      result.delete 'map'
    # rescue
    end
  end

  results << result
end

puts results.to_yaml