numnumnum is a stupid Krak.dk scraper API - http://numnumnum.herokuapp.com

config.ru

require 'bundler/setup'
Bundler.require :default
require 'json'
require 'open-uri'
require 'cgi'

class Info < Struct.new(:type, :name, :phone, :address, :postal_code, :city)

  def self.from_person handler
    name = handler.css('.hit-name-ellipsis').text.strip
    phone = handler.css('.hit-phone-number').text.strip
    street = handler.css('.hit-street-address').text.strip
    village = handler.css('.hit-postal-place-name')
    street << ", #{village.text.strip}" if village.any?
    postal_code = handler.css('.hit-postal-code').text.strip
    city = handler.css('.hit-address-locality').text.strip
    new *[:person, name, phone, street, postal_code, city]
  end

  def self.from_company handler
    name = handler.css('.hit-company-name-ellipsis').text.strip
    phone = handler.css('.hit-phone-number').text.strip
    street = handler.css('.street-address').text.strip
    postal_code = handler.css('.postal-code').text.strip
    city = handler.css('.locality').text.strip
    new *[:company, name, phone, street, postal_code, city]
  end

  def to_json *args
    to_h.to_json *args
  end

end

class Scraper

  def self.get_all str
    scraper = new str
    [scraper.get_people, scraper.get_companies].flatten.compact
  end

  def initialize str
    @number = parse str
  end

  def get_people
    handler = Nokogiri::XML(get "http://www.krak.dk/person/resultat/#{CGI.escape @number}")
    handler.css('article.hit').map do |info|
      Info.from_person info
    end
  end

  def get_companies
    handler = Nokogiri::XML(get "http://www.krak.dk/#{CGI.escape @number}/s%C3%B8g.cs")
    handler.css('article.hit').map do |info|
      Info.from_company info
    end
  end

  private

  def get url
    open(url).read
  rescue OpenURI::HTTPError => e
    nil
  end

  def parse str
    str.to_s
      .gsub(/^\+45/, '')
  end

end

class Num < Sinatra::Base

  get '/' do
    content_type "text/plain"
    host = env["REQUEST_URI"]
    <<-TEXT
usage: $ curl --silent #{host}12345678

https://gist.github.com/mikker/26ad0a071e9482021f17
TEXT
  end

  get '/favicon.ico' do
    status 404
  end

  get '/:number' do
    content_type 'application/json; charset=utf-8'
    if result = Scraper.get_all(params[:number])
      result.to_json
    else
      status 404
      { error: "not found" }.to_json
    end
  end

end

run Num.new

A wild Krak API appears