Valgskrab

skrab.rb

require 'mechanize'
require 'logger'
require 'json'

ROOT = "http://www.kmdvalg.dk/main"

def get_agent
  agent = Mechanize.new
  agent.log = Logger.new "mech.log"
  agent.user_agent_alias = 'Mac Safari'
  agent
end

def get_all_voting_stations
  agent = get_agent
  all = agent.get ROOT
  results = []
  searched = {}
  searches = []
  searches = all.links.select {|l| l.href.match('fv')}.map {|l| l.href}
  link = searches.shift
  while link
    res = agent.get link
    good_link = link.match(/(F[0-9]+)/)
    prefix = good_link[1] if good_link
    if prefix
      searches += res.links.select {|l| l.href && l.href.match(prefix) && !searched[l.href]}.map {|l| searched[l.href] = true; l.href}
    end
    p prefix
    # sleep 1 # I gave up being nice
    if link.match(/F[0-9]{8}/)
      results.push [link, Time.now, get_tables(res)]
    end
    link = searches.shift
  end
  results
end

def get_tables(page)
  page.search('table').map {|t| t.search('tr').map {|r| r.search('td').map {|c| c.text}}}
end

res = get_all_voting_stations
File.open('stemmesteder.csv', 'w') {|f| f.write res.map {|s| s[2][1][1..-1].map {|r| r.join(';')}}.join("\n")}