dubravy parser

require 'rubygems'
require 'nokogiri'
require 'open-uri'

class ParsingRobots
  #returns array of hashes {:apartment =>, :project_property =>, :outside_area =>, :attachments =>}
  def parse_dubravy_apartments(url, block)
    doc = Nokogiri::HTML open(url)
    apartments = []
    urls=[]

    #download links to estates for block given in argument
    doc.css('table.sortable')[4].css('tr').each do |tr|
      id = tr.css('td')[1]
      urls.push "http://dubravy.eu/jednoizbovy-byt/#{id.content}" unless id.nil? or id.content[0]!=block
    end

    #parsing concrete estates
    urls.each do |url|
      doc = Nokogiri::HTML open(url)
      #f name is fully deserved
      f_table = doc.css('div.floor-info div')[0].content.tr("\t", '').split("\n").select { |line| line =~ /[0-1a-z]+/i }
      #appartment overlay table
      overlay_table = doc.css('div.floor-info table tr')
      #nope zone
      app = {
          project_property: {
              name: f_table[0],
              csl_category: 'Byty',
              price: f_table.select { |line| line =~ /.*Cena:.*/ }[0].split(':')[-1].split(' ')[0..1].join.to_i,
              area: f_table.select { |line| line =~ /.*Interi.r:.*/ }[0].split(' ')[1].to_i,
              property_type: 'apartment',
              rooms: f_table.select { |line| line =~ /[1-4] izb[ay]/ }[0].split(' ')[0].to_i,
              status: dubravy_match_status(f_table[6]),
              identification: "dubrava-#{block}-#{f_table[0]}",
              offer_type_code: 'sale',
              orientation: dubravy_orientation(doc.css('div.floor-info img.compas').attr('class').value.
                  split(' ')[1]&.tr('^0-9', '').to_i)
          },
          apartment: {
              construction_code: 'reinforced_concrete',
              floor: f_table[0].split(' ')[1][1].to_i,
              floor_count: 7,
              state_code: 'approved',
              garden_area: 0
          },
          outside_area: {
              balconies: overlay_table.select { |row| row.css('td')[1].content =~ /balk.n/i unless row.css('td')[1].nil? }.
                  map { |row| row.css('td')[2].content.to_i },
              cellars: overlay_table.select { |row| row.css('td')[1].content =~ /komora/i unless row.css('td')[1].nil? }.
                  map { |row| row.css('td')[2].content.to_i }
          },
          attachments: {
              blueprint: 'http://dubravy.eu' + doc.css('div.floor-info a').
                  select { |link| link.attr('href') =~ /.pdf$/ }[0].attr('href')
          }
      }
      apartments.push app
    end
    apartments
  end

  def dubravy_match_status(status)
    case status
      when /rezervovan./
        rv = 'reserved'
      when /predrezervovan./
        rv = 'prereserved'
      when /vo.n./
        rv = 'free'
      when /predan./
        rv = 'sold'
      else
        rv = nil
    end
    rv
  end

  def dubravy_orientation(compass_rotation)
    case compass_rotation
      when 0..5
        rv = 'N'
      when 6...85
        rv = 'NE'
      when 85..95
        rv = 'E'
      when 96...175
        rv = 'ES'
      when 175..185
        rv = 'S'
      when 186...265
        rv = 'SW'
      when 266..275
        rv = 'W'
      when 276...355
        rv = 'WN'
      when 356..360
        rv = 'N'
      else
        rv = nil
    end
    rv
  end
end