--- a/maxious-canberra-transit-feed/01-extracttimes.rb +++ b/maxious-canberra-transit-feed/01-extracttimes.rb @@ -14,21 +14,30 @@ timetable = {"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","").gsub("\r","").gsub("\\"," / ").strip + timing_point = tp.content.squeeze(" ").gsub("Bus Station"," Bus Station ").gsub(" Platform"," (Platform").gsub(" - "," - ").gsub("\n"," ").gsub("\r"," ").gsub("\t"," ").gsub("\\"," / ").gsub("/"," / ").gsub(",",", ").gsub("\302\240","").squeeze(" ").strip + if (tp.content.match('Platform')) + timing_point.concat(")") + end; + timing_point end end time_points.delete(nil) + time_points.delete("WheelchairAccessible") + time_points.delete("Wheelchair Accessible") timetable["time_points"] = time_points.to_a timetable["long_name"] = "To " + time_points.last periodtimes = [] table.css('tr').each do |row| times = row.css('td').map do |cell| time = cell.content.squeeze(" ").strip - time = time.gsub(" AM","a").gsub(" PM","p").gsub("12:08 AM","1208x").gsub(":","").gsub("1.","1").gsub("2.","2") + time = time.gsub(/ *A\S?M/,"a").gsub(/ ?P\S?M/,"p").gsub("12:08 AM","1208x").gsub(":","").gsub("1.","1").gsub("2.","2") time = time.gsub("3.","3").gsub("4.","4") time = time.gsub("5.","5").gsub("6.","6").gsub("7.","7").gsub("8.","8").gsub("9.","9").gsub("10.","10") - time = time.gsub("11.","11").gsub("12.","12").gsub(/\.+/,"-") + time = time.gsub("11.","11").gsub("12.","12").gsub(/\.+/,"-").gsub("\302\240","") + if time == "" then time = nil end + time end + times.delete(nil) if not times.empty? if not (route = times.shift) raise("TODO: account for shifting route numbers eg. intertown/redex 62/162") @@ -41,7 +50,8 @@ end timetable[period] = periodtimes.to_a # pp timetable - filename = timetable["short_name"] + "-" + timetable["long_name"].downcase.gsub(" ","-").gsub("/","") + "." + period + ".yml" + filename = timetable["short_name"] + "-" + timetable["long_name"]+ "." + period + ".yml" + filename = filename.downcase.gsub(" ","-").gsub("/","-").gsub("(","").gsub(")","") puts "Saving " + filename File.open("#{File.dirname(__FILE__)}/output/"+filename, "w") do |f| f.write timetable.to_yaml @@ -49,7 +59,6 @@ timetable end -#TODO fix route 934 Dir.glob("source-html/Route*.htm*") { |file| puts "Opened " + file doc = Nokogiri::HTML(open(file)) @@ -57,7 +66,7 @@ timetables = [] short_name = ""; doc.xpath('//title').each do |title| - short_name = title.content.gsub("Route_","").gsub("Route ","").squeeze(" ").strip + short_name = title.content.gsub("Route_","").gsub("Route ","").gsub(", ","/").gsub("ACTION Buses Timetable for ","").squeeze(" ").strip end if short_name == "" raise "Route number(s) not found in tag" @@ -66,7 +75,15 @@ doc.xpath('//table[preceding::text()="Weekdays"]').each do |table| timetables << makeTimetable(table, "stop_times", short_name) end - + doc.xpath('//table[preceding::text()="This timetable is effective from Monday 15th November 2010."]').each do |table| + timetables << makeTimetable(table, "stop_times", short_name) + end + #all tables are weekdays on some really malformatted timetables + if short_name == "170" + doc.xpath('//table').each do |table| + timetables << makeTimetable(table, "stop_times", short_name) + end + end #weekends doc.xpath('//table[preceding::text()="Saturdays" and following::a]').each do |table| timetables << makeTimetable(table, "stop_times_saturday", short_name)