Finn.no real estate pages as JSON data

finn.rb

# encoding: UTF-8

require 'open-uri'
require 'uri'

def finn(id)
  doc = open("http://www.finn.no/finn/realestate/homes/object?finnkode=#{id}").read
  doc =~ /helios_parameters\s?:\s?'(.*)'/
  attrs = $1.split(';').inject({}) do |mem, e|
    key, value = e.split "="
    value = URI.unescape(value)
    if key !~ /bilde/
      value = value.gsub('_', ' ')
      case key
      when /mobil$/
        value.gsub!(" ", "")
      end
      mem[key] = URI.unescape(value)
    end
    mem
  end

  attrs['geo'] = { }
  doc =~ /utmx    : "(\d+)"/
    attrs['geo']['x'] = $1
  doc =~ /utmy    : "(\d+)"/
    attrs['geo']['y'] = $1

  if attrs['sold'] = !!(doc =~ /soldNoteHolder/)
    doc =~ /soldNoteHolder.*<p>\((.*),-\)/m
    attrs['sold_price'] = $1
  end
  doc = Nokogiri::HTML.parse doc
  attrs['images'] = doc.css('.gallery-image').map {|i| i.attribute('data-main').text}
  attrs
end

require 'nokogiri'
def today
  doc = Nokogiri::HTML.parse(open('http://www.finn.no/finn/realestate/homes/result?page=1&periode=1').read)
  page = 1
  links = []
  while(true)
    new_links = doc.css('ul#resultlist li').map do |li|
      li.css('a').attribute('href').to_s.split("=")[-1]
    end
    if new_links.size == 0
      break
    end
    links += new_links
    puts "Have #{links.size} links now"
    page += 1
    doc = Nokogiri::HTML.parse(open("http://www.finn.no/finn/realestate/homes/result?page=#{page}&periode=1").read)
  end
  links
end

require 'pp'
require 'json'
puts finn('31233480').to_json