data.gov.ro infofer parser

infofer.rb

require 'nokogiri'
require 'sqlite3'
require 'FileUtils'

def seconds_to_hhmm (total_seconds)
  # From http://stackoverflow.com/a/9916691
  seconds = total_seconds % 60
  minutes = (total_seconds / 60) % 60
  hours = total_seconds / (60 * 60)

  hhmm = format("%02d:%02d", hours, minutes)
  return hhmm
end

agency_map_id = {
  "236037" => "interregional",
  "227098" => "regiotrans",
  "6100826" => "sntfc",
  "236025" => "softrans",
  "228389" => "transferoviar",
}

data_stops = {}
data_trips = []

Dir["#{Dir.pwd}/data.gov.ro/2014-2015/*.xml"].each do |file_in|
  file_content = IO.read(file_in)
  doc = Nokogiri::XML(file_content)  

  print "Parsing #{file_in}\n"
  
  doc.xpath('/XmlIf/XmlMts/Mt/Trenuri/Tren').each_with_index do |row_train, k_train|
    data_trip = {
      "type" => row_train.attr('CategorieTren'),
      "no" => row_train.attr('Numar'),
      "stops" => [],
      "agency_id" => row_train.attr('Operator'),
    }

    if data_trip['no'] != "9481"
      # next
    end

    km_cumulated = 0

    row_stations = row_train.xpath('Trase/Trasa/ElementTrasa')
    row_stations.each_with_index do |row_station, k|
      st_a_id = row_station.attr('CodStaOrigine')
      st_a_name = row_station.attr('DenStaOrigine')
      st_a_sec = row_station.attr('OraP').to_i
      st_a_hms = seconds_to_hhmm(st_a_sec)

      st_b_id = row_station.attr('CodStaDest')
      st_b_name = row_station.attr('DenStaDestinatie')
      st_b_sec = row_station.attr('OraS').to_i
      st_b_hms = seconds_to_hhmm(st_b_sec)

      if data_stops[st_a_id].nil?
        data_stop = {
          "id" => st_a_id,
          "name" => st_a_name
        }
        data_stops[st_a_id] = data_stop
      end

      trip_stop = {
        "stop_id" => st_a_id
      }

      if k == 0
        trip_stop["arr"] = ""
        trip_stop["km_cumulated"] = 0
      else
        trip_stop["arr"] = seconds_to_hhmm(row_stations[k-1].attr('OraS').to_i)
      end

      if k == (row_stations.length - 1)
        trip_stop["dep"] = ""
      else
        trip_stop["dep"] = st_a_hms
      end

      trip_stop['km_cumulated'] = km_cumulated
      km_cumulated += (row_station.attr('Km').to_i / 1000).to_i

      data_trip['stops'].push(trip_stop)

      # p "#{st_a_name}(#{st_a_hms})-#{st_b_name}(#{st_b_hms})"
      # p "#{st_a_sec} -- #{st_b_sec}"
    end

    data_trips.push(data_trip)

    if k_train > 10
      # break
    end
  end  
end

p data_trips