train_archival/analysis.rb

#! /usr/bin/env ruby

require 'json'

$datestr = ARGV[0] || $stdin.gets.strip

# @param [String] station_name
def normalize_station station_name
	station_name
		.downcase
		.gsub(/\s/, ' ')
		.gsub('ă', 'a')
		.gsub('â', 'a')
		.gsub('î', 'i')
		.gsub('ș', 's')
		.gsub('ț', 't')
end

def get_station station_name
	# Try getting from file
	begin
		JSON.parse(File.read(File.join($datestr, 'stations', "#{normalize_station(station_name)}.json")), symbolize_names: true)
	rescue
		nil
	end
end

# @param [String] train_number
def get_train train_number
	# Try getting from file
	begin
		JSON.parse(File.read(File.join($datestr, 'trains', "#{train_number}.json")), symbolize_names: true)
	rescue
		nil
	end
end

visited_stations = Dir.entries(File.join $datestr, 'stations')
	.filter { |entry| entry != '.' and entry != '..' }
	.map { |name| if name.end_with? ".json" then name[0...-5] else name end }
	.map { |name| get_station(name)[:stationName] }
visited_trains = Dir.entries(File.join $datestr, 'trains')
	.filter { |entry| entry != '.' and entry != '..' }
	.map { |name| if name.end_with? ".json" then name[0...-5] else name end }

trains = visited_trains.map { |train_number| get_train train_number }
r_trains = trains.filter { |train| train[:rank] == 'R' }
re_trains = trains.filter { |train| train[:rank] == 'R-E' }
ir_trains = trains.filter { |train| train[:rank] == 'IR' }
irn_trains = trains.filter { |train| train[:rank] == 'IRN' }
ic_trains = trains.filter { |train| train[:rank] == 'IC' }

def filter_by_arrival_delay trains
	trains.filter do |train|
		arrival_station = train[:groups][0][:stations][-1]
		arrival = arrival_station[:arrival]
		if arrival[:status] and arrival[:status][:real] 
			yield arrival[:status][:delay]
		else
			false
		end
	end
end

def get_early_arrival trains
	filter_by_arrival_delay(trains) { |delay| delay < 0 }
end

def get_on_time_arrival trains
	filter_by_arrival_delay(trains) { |delay| delay == 0 }
end

def get_late_arrival trains
	filter_by_arrival_delay(trains) { |delay| delay > 0 }
end

def get_late_gt5_arrival trains
	filter_by_arrival_delay(trains) { |delay| delay > 5 }
end

on_time_departure = trains.filter do |train|
	departure_station = train[:groups][0][:stations][0]
	departure = departure_station[:departure]
	departure[:status] and departure[:status][:real] and departure[:status][:delay] == 0
end
late_departure = trains.filter do |train|
	departure_station = train[:groups][0][:stations][0]
	departure = departure_station[:departure]
	departure[:status] and departure[:status][:real] and departure[:status][:delay] > 0
end

early_arrival = get_early_arrival trains
on_time_arrival = get_on_time_arrival trains
late_arrival = get_late_arrival trains
late_arrival_gt5 = get_late_gt5_arrival trains
late_gt5_arrival = late_arrival_gt5

gained_delay = []
recovered_delay = []

early_on_route = []

ongoing = trains.filter do |train|
	arrival_station = train[:groups][0][:stations][-1]
	arrival_station[:arrival][:status] and not arrival_station[:arrival][:status][:real]
end

arrival_delays = {}
trains.each do |train|
	arrival_station = train[:groups][0][:stations][-1]
	arrival = arrival_station[:arrival]
	if arrival[:status] and arrival[:status][:real]
		delay = arrival[:status][:delay]
		arr = arrival_delays.fetch(delay, [])
		arr << train
		arrival_delays[delay] = arr
	end
end

puts "#{on_time_departure.count} trains departed on time"
puts "#{late_departure.count} trains departed late"
puts
printf "%d trains arrived early\n", early_arrival.count
# printf "    %4d (%3d%%) R, %4d (%3d%%) R-E, %4d (%3d%%) IR, %4d (%3d%%) IRN, %4d (%3d%%) IC\n", get_early_arrival(r_trains).count, (get_early_arrival(r_trains).count * 100 / early_arrival.count), get_early_arrival(re_trains).count, (get_early_arrival(re_trains).count * 100 / early_arrival.count), get_early_arrival(ir_trains).count, (get_early_arrival(ir_trains).count * 100 / early_arrival.count), get_early_arrival(irn_trains).count, (get_early_arrival(irn_trains).count * 100 / early_arrival.count), get_early_arrival(ic_trains).count, (get_early_arrival(ic_trains).count * 100 / early_arrival.count)
printf "%d trains arrived on time\n", on_time_arrival.count
# printf "    %4d (%3d%%) R, %4d (%3d%%) R-E, %4d (%3d%%) IR, %4d (%3d%%) IRN, %4d (%3d%%) IC\n", get_on_time_arrival(r_trains).count, (get_on_time_arrival(r_trains).count * 100 / on_time_arrival.count), get_on_time_arrival(re_trains).count, (get_on_time_arrival(re_trains).count * 100 / on_time_arrival.count), get_on_time_arrival(ir_trains).count, (get_on_time_arrival(ir_trains).count * 100 / on_time_arrival.count), get_on_time_arrival(irn_trains).count, (get_on_time_arrival(irn_trains).count * 100 / on_time_arrival.count), get_on_time_arrival(ic_trains).count, (get_on_time_arrival(ic_trains).count * 100 / on_time_arrival.count)
printf "%d trains arrived late\n", late_arrival.count
# printf "    %4d (%3d%%) R, %4d (%3d%%) R-E, %4d (%3d%%) IR, %4d (%3d%%) IRN, %4d (%3d%%) IC\n", get_late_arrival(r_trains).count, (get_late_arrival(r_trains).count * 100 / late_arrival.count), get_late_arrival(re_trains).count, (get_late_arrival(re_trains).count * 100 / late_arrival.count), get_late_arrival(ir_trains).count, (get_late_arrival(ir_trains).count * 100 / late_arrival.count), get_late_arrival(irn_trains).count, (get_late_arrival(irn_trains).count * 100 / late_arrival.count), get_late_arrival(ic_trains).count, (get_late_arrival(ic_trains).count * 100 / late_arrival.count)
printf "%d trains arrived with a delay greater than 5 minutes\n", late_arrival_gt5.count
# printf "    %4d (%3d%%) R, %4d (%3d%%) R-E, %4d (%3d%%) IR, %4d (%3d%%) IRN, %4d (%3d%%) IC\n", get_late_gt5_arrival(r_trains).count, (get_late_gt5_arrival(r_trains).count * 100 / late_gt5_arrival.count), get_late_gt5_arrival(re_trains).count, (get_late_gt5_arrival(re_trains).count * 100 / late_gt5_arrival.count), get_late_gt5_arrival(ir_trains).count, (get_late_gt5_arrival(ir_trains).count * 100 / late_gt5_arrival.count), get_late_gt5_arrival(irn_trains).count, (get_late_gt5_arrival(irn_trains).count * 100 / late_gt5_arrival.count), get_late_gt5_arrival(ic_trains).count, (get_late_gt5_arrival(ic_trains).count * 100 / late_gt5_arrival.count)
puts
puts "Arrivals at destination by rank:"
printf "  R: %4d (%3d%%) early, %4d (%3d%%) on time, %4d (%3d%%) late [%4d (%3d%%) >5min late], %4d total\n", get_early_arrival(  r_trains).count, (get_early_arrival(  r_trains).count * 100 /   r_trains.count), get_on_time_arrival(  r_trains).count, (get_on_time_arrival(  r_trains).count * 100 /   r_trains.count), get_late_arrival(  r_trains).count, (get_late_arrival(  r_trains).count * 100 /   r_trains.count), get_late_gt5_arrival(  r_trains).count, (get_late_gt5_arrival(  r_trains).count * 100 /   r_trains.count),   r_trains.count
printf "R-E: %4d (%3d%%) early, %4d (%3d%%) on time, %4d (%3d%%) late [%4d (%3d%%) >5min late], %4d total\n", get_early_arrival( re_trains).count, (get_early_arrival( re_trains).count * 100 /  re_trains.count), get_on_time_arrival( re_trains).count, (get_on_time_arrival( re_trains).count * 100 /  re_trains.count), get_late_arrival( re_trains).count, (get_late_arrival( re_trains).count * 100 /  re_trains.count), get_late_gt5_arrival( re_trains).count, (get_late_gt5_arrival( re_trains).count * 100 /  re_trains.count),  re_trains.count
printf " IR: %4d (%3d%%) early, %4d (%3d%%) on time, %4d (%3d%%) late [%4d (%3d%%) >5min late], %4d total\n", get_early_arrival( ir_trains).count, (get_early_arrival( ir_trains).count * 100 /  ir_trains.count), get_on_time_arrival( ir_trains).count, (get_on_time_arrival( ir_trains).count * 100 /  ir_trains.count), get_late_arrival( ir_trains).count, (get_late_arrival( ir_trains).count * 100 /  ir_trains.count), get_late_gt5_arrival( ir_trains).count, (get_late_gt5_arrival( ir_trains).count * 100 /  ir_trains.count),  ir_trains.count
printf "IRN: %4d (%3d%%) early, %4d (%3d%%) on time, %4d (%3d%%) late [%4d (%3d%%) >5min late], %4d total\n", get_early_arrival(irn_trains).count, (get_early_arrival(irn_trains).count * 100 / irn_trains.count), get_on_time_arrival(irn_trains).count, (get_on_time_arrival(irn_trains).count * 100 / irn_trains.count), get_late_arrival(irn_trains).count, (get_late_arrival(irn_trains).count * 100 / irn_trains.count), get_late_gt5_arrival(irn_trains).count, (get_late_gt5_arrival(irn_trains).count * 100 / irn_trains.count), irn_trains.count
printf " IC: %4d (%3d%%) early, %4d (%3d%%) on time, %4d (%3d%%) late [%4d (%3d%%) >5min late], %4d total\n", get_early_arrival( ic_trains).count, (get_early_arrival( ic_trains).count * 100 /  ic_trains.count), get_on_time_arrival( ic_trains).count, (get_on_time_arrival( ic_trains).count * 100 /  ic_trains.count), get_late_arrival( ic_trains).count, (get_late_arrival( ic_trains).count * 100 /  ic_trains.count), get_late_gt5_arrival( ic_trains).count, (get_late_gt5_arrival( ic_trains).count * 100 /  ic_trains.count),  ic_trains.count
puts
puts "Delays at arrival:"
arrival_delays.to_a.sort {|a, b| b[1].count <=> a[1].count}.each do |delay, trains|
	printf "%3d min: %4d train%s", delay, trains.count, if trains.count == 1 then " " else "s" end
	if trains.count <= 5
		printf " (%s)", trains.map { |train|
			sprintf "%3s %5s", train[:rank], train[:number]
		}.join(", ")
	end
	puts
end
puts
puts "#{ongoing.count} trains still travelling"

puts
puts "#{on_time_departure.count} trenuri plecate la timp"
puts "#{late_departure.count} trenuri plecate cu întârziere"
puts
puts "#{early_arrival.count} trenuri sosite mai devreme"
puts "#{on_time_arrival.count} trenuri sosite la timp"
puts "#{late_arrival.count} trenuri sosite cu întârziere"
puts "#{late_arrival_gt5.count} trenuri sosite cu întârziere mai mare de 5 minute"
puts
puts "Întârzieri la destinație:"
arrival_delays.to_a.sort {|a, b| b[1].count <=> a[1].count}.each do |delay, trains|
	printf "%3d min: %4d tren%s", delay, trains.count, if trains.count == 1 then "   " else "uri" end
	if trains.count <= 5
		printf " (%s)", trains.map { |train|
			sprintf "%3s %5s", train[:rank], train[:number]
		}.join(", ")
	end
	puts
end