From: root Date: Tue, 20 Apr 2010 11:18:35 +0000 Subject: Begin to locate time points X-Git-Url: http://maxious.lambdacomplex.org/git/?p=bus.git&a=commitdiff&h=bea01b85a230725062457d751600fe986053fd9a --- Begin to locate time points --- --- a/maxious-canberra-transit-feed/extracttimes.rb +++ /dev/null @@ -1,95 +1,1 @@ -require 'rubygems' -require 'nokogiri' -require 'open-uri' -require 'pp' -def makeTimetable(table, period, short_name) - timetable = {"stop_times" => [], "between_stops" => [], "short_name" => short_name} - time_points = table.xpath('tr[1]//th').map do |tp| - if tp.content != "\302\240" && tp.content != "" && tp.content != "
" - timing_point = tp.content.squeeze(" ").gsub("\r\n Platform"," - Platform").gsub(" - "," - ").gsub("\n","").strip - end - end - time_points.delete(nil) - timetable["time_points"] = time_points - timetable["long_name"] = "To " + time_points.last - periodtimes = [] - table.css('tr').each do |row| - times = row.css('td').map do |cell| - #TODO convert to GTFS time ie. replace " AM" with a - time = cell.content.squeeze(" ").strip - end - if not times.empty? - if not (route = times.shift) - raise("TODO: account for shifting route numbers eg. intertown/redex 62/162") - end - periodtimes << times - end - end - if periodtimes.size < 1 - raise "No times for route " + short_name + " in period " + period - end - timetable["stop_times"] = { period => periodtimes } - # pp timetable - filename = timetable["short_name"] + "-" + timetable["long_name"].downcase.gsub(" ","-").gsub("/","") + "." + period + ".yml" - puts "Saving " + filename - File.open("#{File.dirname(__FILE__)}/output/"+filename, "w") do |f| - f.write timetable.to_yaml - end - timetable -end - -#TODO fix route 934 -Dir.glob("source-html/Route*.htm*") { |file| - puts "Opened " + file - doc = Nokogiri::HTML(open(file)) - # Search for nodes by css - timetables = [] - short_name = ""; - doc.xpath('//title').each do |title| - short_name = title.content.gsub("Route_","").gsub("Route ","").squeeze(" ").strip - end - if short_name == "" - raise "Route number(s) not found in tag" - end - - doc.xpath('//table[preceding::text()="Weekdays"]').each do |table| - timetables << makeTimetable(table, "weekday", short_name) - end - - #weekends - doc.xpath('//table[preceding::text()="Saturdays" and following::a]').each do |table| - timetables << makeTimetable(table, "saturday", short_name) - end - doc.xpath('//table[preceding::text()="Sundays"]').each do |table| - timetables << makeTimetable(table, "sunday", short_name) - end - #930/934 special cases - doc.xpath('//table[preceding::text()="Saturday" and following::h2]').each do |table| - timetables << makeTimetable(table, "saturday", short_name) - end - doc.xpath('//table[preceding::text()="Sunday"]').each do |table| - timetables << makeTimetable(table, "sunday", short_name) - end - #route 81 = Weekdays - School Holidays Only - doc.xpath('//table[preceding::text()="Weekdays - School Holidays Only "]').each do |table| - timetable = makeTimetable(table, "weekday", short_name) - #TODO set active date range to only be holidays - timetables << timetable; - end - - - if timetables.size > 2 - puts "WARNING: " + file + " more than 2 timetables (weekend split?):" + timetables.size.to_s - end - if timetables.size < 2 - puts "WARNING: " + file + " less than 2 timetables (weekday loop service?):" + timetables.size.to_s - elsif not (timetables[0]["time_points"] - timetables[1]["time_points"].reverse).empty? - puts "WARNING: first pair of timetable timing points are not complementary for "+ file - pp(timetables[0]["time_points"] - timetables[1]["time_points"].reverse) - end - if timetables.size < 1 - raise "No timetables extracted from " + file - end -} - --- a/maxious-canberra-transit-feed/substitutetimepoints.rb +++ /dev/null @@ -1,71 +1,1 @@ -require 'rubygems' -require 'pp' -require 'yaml' -time_points = [] -time_points_sources = Hash.new([]) -Dir.chdir("output") -Dir.glob("*.yml") { |file| - timetable = YAML::load_file(file) - time_points = time_points | timetable["time_points"] - timetable["time_points"].each do |timepoint| - time_points_sources[timepoint] = time_points_sources[timepoint] | [ file ] - end -} -pp time_points.sort! -#pp time_points_sources.sort - -time_point_corrections = {"North Lynehamham" => "North Lyneham", - "Lathlain St Platform 2" => "Lathlain St Bus Station - Platform 2", - "Lathlain St Sation - Platform 5" => "Lathlain St Bus Station - Platform 5", - "Lathlain Steet Station" => "Lathlain St Bus Station", - "Lathlain St - Platform 3" => "Lathlain St Bus Station - Platform 3", - "Lathlain Steet Station - Platform 3" => "Lathlain St Bus Station - Platform 3", - "Lathlain St Station" => "Lathlain St Bus Station", - "Lathlain St Station - Platform 1" => "Lathlain St Bus Station - Platform 1", - "Lathlain St Station - Platform 2" => "Lathlain St Bus Station - Platform 2", - "Lathlain St Station - Platform 3" => "Lathlain St Bus Station - Platform 3", - "Lathlain St Station - Platform 4" => "Lathlain St Bus Station - Platform 4", - "Lathlain St Station - Platform 5" => "Lathlain St Bus Station - Platform 5", - "Lathlain St Station - Platform 6" => "Lathlain St Bus Station - Platform 6", - "Manuka Captain Cook" => "Manuka, Captain Cook", - "Flemington Rd, Sandford St" => "Flemington Rd/Sandford St", - "Erindale Centre / - Sternberg Crescent" => "Erindale Drive/Sternberg", - "Canberra Hospita" => "Canberra Hospital", - "Cohen Str Station - Platform 1" => "Cohen St Bus Station - Platform 1", - "Cohen Street Station" => "Cohen St Bus Station", - "Cohen Street Station - Platform 2" => "Cohen St Bus Station - Platform 2", - "Cohn St Station - Platform 3" => "Cohen St Bus Station - Platform 3", - "Cohen St Station" => "Cohen St Bus Station", - "Cohen St Station - Platform 1" => "Cohen St Bus Station - Platform 1", - "Cohen St Station - Platform 2" => "Cohen St Bus Station - Platform 2", - "Cohen St Station - Platform 3" => "Cohen St Bus Station - Platform 3", - "Cohen St Station - Platform 4" => "Cohen St Bus Station - Platform 4", - "Cohen St Station - Platform 5" => "Cohen St Bus Station - Platform 5", - "City - Platform 7" => "City Interchange - Platform 7", - "Cameron Avenue Station" => "Cameron Ave Bus Station", - "Cameron Avenue Station - Platform 2" => "Cameron Ave Bus Station - Platform 2", - "Cameron Avenue Station - Platform 3" => "Cameron Ave Bus Station - Platform 3", - "Cameron Ave Station" => "Cameron Ave Bus Station", - "Cameron Ave Station - Platform 1" => "Cameron Ave Bus Station - Platform 1", - "Cameron Ave Station - Platform 2" => "Cameron Ave Bus Station - Platform 2", - "Cameron Ave Station - Platform 3" => "Cameron Ave Bus Station - Platform 3", - "Cameron Ave Station - Platform 4" => "Cameron Ave Bus Station - Platform 4", - "Cameron Ave Station - Platform 5" => "Cameron Ave Bus Station - Platform 5", - "Burton & Garranan Hall, Daley Road ANU" => "Burton & Garran Hall, Daley Road ANU", - "Burton & Garranan Hall,Daley Road ANU" => "Burton & Garran Hall, Daley Road ANU" - "Newcastle Street after Isa St" => "Newcastle / Isa Street Fyshwick", - "National Circ/Canberra Ave" => "National Circuit / Canberra Ave", - } -time_point_corrections.each do |wrong, right| - time_points_sources[wrong].each do |wrongfile| - badtimetable = YAML::load_file(wrongfile) - badentrynumber = badtimetable["time_points"].index wrong - badtimetable["time_points"][badentrynumber] = right - puts "Corrected '" + wrong + "' to '" + right + "' in " + wrongfile - File.open(wrongfile, "w") do |f| - f.write badtimetable.to_yaml - end - end -end -