Startup Weekend Hack

pro-finder.rb

require "mechanize"

BASE_URL = "http://www.societe.com"
SEARCH_URL = "#{BASE_URL}/cgi-bin/liste?ape={APE}&dep={DEP}"

APE = %W{
  4120A 4120B
  4311Z 4312A 4312B 4313Z 4321A 4321B 4322A 4322B 4329A 4329B 4331Z 4332A 4332B 4332C 4333Z 4334Z 4339Z 4391A 4391B 4399A 4399B 4399C 4399D 4399E
}

Artisan = Struct.new(:id, :name, :address, :postal_code, :city, :country, :ape_code, :created_at)

crawler = Mechanize.new

(1..99).each do |dep|
  puts "DEP: #{dep}"
  APE.each do |ape|
    puts "\t#{ape}: "
    search_url = SEARCH_URL.gsub("{DEP}", dep.to_s).gsub("{APE}", ape)
    crawler.get(search_url).search("#liste > a.linkresult").each do |link|
      societe_url = BASE_URL + link[:href]
      crawler.get(societe_url)
    end
  end
end