package dk.thoerup.traininfoservice.banedk; import java.net.URL; import java.net.URLEncoder; import java.util.Collections; import java.util.Map; import java.util.logging.Logger; import org.jsoup.nodes.Document; import org.jsoup.nodes.Element; import org.jsoup.select.Elements; import dk.thoerup.android.traininfo.common.DepartureBean; import dk.thoerup.android.traininfo.common.DepartureEntry; import dk.thoerup.android.traininfo.common.StationBean.StationEntry; import dk.thoerup.circuitbreaker.CircuitBreaker; import dk.thoerup.circuitbreaker.CircuitBreakerManager; import dk.thoerup.genericjavautils.HttpUtil; import dk.thoerup.genericjavautils.TimeoutMap; import dk.thoerup.traininfoservice.Statistics; import dk.thoerup.traininfoservice.TraininfoSettings; import dk.thoerup.traininfoservice.db.StationDAO; public class DepartureFetcher { enum TrainType{ STOG, REGIONAL } enum FetchTrainType { STOG, REGIONAL, BOTH } Logger logger = Logger.getLogger(DepartureFetcher.class.getName()); Map cache; StationDAO stationDao = new StationDAO(); private TraininfoSettings settings; public DepartureFetcher(TraininfoSettings settings) { this.settings = settings; cache = new TimeoutMap( settings.getCacheTimeout() ); } public DepartureBean cachedLookupDepartures(int stationID, boolean arrival, FetchTrainType type) throws Exception { final String key = "" + stationID + ":" + arrival + ":" + type.toString(); DepartureBean departureBean = cache.get(key); if (departureBean == null) { departureBean = lookupDepartures(stationID, arrival, type); cache.put(key, departureBean); } else { Statistics.getInstance().incrementDepartureCacheHits(); logger.info("Departure: Cache hit " + key); //remove before production } return departureBean; } public DepartureBean lookupDepartures(int stationID, boolean arrival, FetchTrainType type) throws Exception { DepartureBean departureBean = new DepartureBean(); StationEntry station = stationDao.getById(stationID); departureBean.stationName = station.getName(); if (station.getRegional() != null && (type == FetchTrainType.REGIONAL||type == FetchTrainType.BOTH) ) { DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival); departureBean.entries.addAll( tempBean.entries ); departureBean.notifications.addAll(tempBean.notifications); } if (station.getStrain() != null && (type == FetchTrainType.STOG||type == FetchTrainType.BOTH)) { DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival); departureBean.entries.addAll( tempBean.entries ); departureBean.notifications.addAll(tempBean.notifications); } if (departureBean.entries.size() == 0) { logger.info("No departures found for station " + stationID); } if (type == FetchTrainType.BOTH) { //if we have both S-tog and regional order by departure/arrival time Collections.sort( departureBean.entries ); } return departureBean; } public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception { if ( settings.getUseAzureSite() == true) { return lookupDeparturesAzureSite(stationcode, type, arrival); } else { return lookupDeparturesMobileSite(stationcode, type, arrival); } } private String getTypeStringAzure(TrainType type) { switch (type) { case STOG: return "S-Tog"; case REGIONAL: return "Fjerntog"; default: return ""; //Can not happen } } private String getTypeStringWww(TrainType type) { switch (type) { case STOG: return "S2"; case REGIONAL: return "FJRN"; default: return ""; //Can not happen } } public DepartureBean lookupDeparturesAzureSite(String stationcode, TrainType type, boolean arrival) throws Exception { DepartureBean departureBean = new DepartureBean(); String typeString = getTypeStringAzure(type); String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst"; stationcode = URLEncoder.encode(stationcode,"ISO-8859-1"); String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning"; logger.fine("URI: " + uri); JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), settings.getReplyTimeout() ); CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); Document page = (Document) breaker.invoke(wrapper); String tableName = arrival == false ? "afgangtabel" : "ankomsttabel"; Element table = page.getElementById(tableName); if (table != null) { Elements tableRows = table.getElementsByTag("tr"); //boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0); //boolean passedTidsstreg = false; for (Element currentRow : tableRows) { String rowClass = currentRow.attr("class"); /* if (tidsstregExists == true && passedTidsstreg == false) { if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) { passedTidsstreg = true; } else { continue; } }*/ if (rowClass != null && rowClass.toLowerCase().contains("station") ) { Elements fields = currentRow.getElementsByTag("td"); DepartureEntry departure = new DepartureEntry(); String time = fields.get(0).text(); if (time.equals("")) time = "0:00"; //Bane.dk bug work-around departure.setTime(time); int updated = extractUpdated( fields.get(1) ); departure.setUpdated(updated); String trainNumber = fields.get(2).text(); if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber trainNumber = trainNumber + " " + extractTrainNumberAzure(fields.get(2)); departure.setTrainNumber(trainNumber); String destination = fields.get(3).text(); departure.setDestination(destination); String origin = fields.get(4).text(); departure.setOrigin(origin); String location = fields.get(5).text(); departure.setLocation(location); String status = fields.get(6).text().trim(); departure.setStatus(status); String note = extractNote( fields.get(7) ); departure.setNote(note); departure.setType(typeString); departureBean.entries.add( departure ); } } } else { logger.warning("No departures found for station=" + stationcode + ", type=" + type); } Element notifDiv = page.getElementById("station_planlagte_text"); if (notifDiv != null) { Elements tables = notifDiv.getElementsByTag("table"); for (Element tab : tables) { Elements anchors = tab.getElementsByTag("a"); if (anchors.size() == 2) { departureBean.notifications.add( anchors.get(1).text() ); } } } return departureBean; } public DepartureBean lookupDeparturesMobileSite(String stationcode, TrainType traintype, boolean arrival) throws Exception { DepartureBean departureBean = new DepartureBean(); String typeString = getTypeStringWww(traintype); String arrivalDeparture = (arrival==false) ? "afgang" : "ankomst"; stationcode = URLEncoder.encode(stationcode,"ISO-8859-1"); //String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning"; String uri = "http://mobil.bane.dk/mobilStation.asp?artikelID=5332&stat_kode=" + stationcode + "&webprofil=" + typeString +"&beskrivelse=&mode=ankomstafgang&ankomstafgang=" + arrivalDeparture + "&gemstation=&fuldvisning=1"; logger.fine("URI: " + uri); JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), settings.getReplyTimeout() ); CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); Document page = (Document) breaker.invoke(wrapper); Element content = page.getElementsByClass("contentDiv").get(0); if (content != null) { Elements tableRows = content.child(0).children(); for (Element currentRow : tableRows) { if (currentRow.tagName().equals("br") ) { break; } String link = currentRow.child(0).attr("href"); logger.fine( currentRow.text() ); logger.fine("Href: " + link); String parts[] = currentRow.text().split(","); DepartureEntry departure = new DepartureEntry(); //if we do these things upfront, then we are allowed to use continue statement when row contains no more data departure.setType(typeString); departureBean.entries.add( departure ); /* http://mobil.bane.dk/mobilStation.asp?artikelID=5332&tognummer=111&webprofil=FJRN&mode=rute&strBemaerkning=Afg%E5r+fra+%C5rhus+H+kl%2E07%3A21++&strRefURL=%2FmobilStation%2Easp%3FartikelID%3D5332%26stat%5Fkode%3DAR%26webprofil%3DFJRN%26beskrivelse%3D%25C5rhus%2BH%26mode%3Dankomstafgang%26ankomstafgang%3Dafgang%26gemstation%3D */ int offset = 0; String time = parts[offset++]; if (time.equals("")) time = "0:00"; //Bane.dk bug work-around departure.setTime(time); int updated = 4; //does not exist on mobile departure.setUpdated(updated); String trainNumber = extractTrainNumberMobile(link); /*if (traintype == TrainType.STOG) //If it is S-train we need to extract the trainNumber trainNumber = trainNumber + " " + extractTrainNumberAzure(fields.get(2));*/ departure.setTrainNumber(trainNumber); if (traintype == TrainType.STOG) { //if it is stog the next vield is the "Line" code - this should be used somewhere, but skippint ahead for now String stogLine = parts[offset++].trim(); departure.setTrainNumber(stogLine + " " + trainNumber); } String destination = parts[offset++].trim();; departure.setDestination(destination); String origin = "-"; // fields.get(4).text(); does not exist on mobile departure.setOrigin(origin); String location = ""; // fields.get(5).text(); does not exist on mobile departure.setLocation(location); if (offset == parts.length) { continue; } if (parts[offset].trim().equalsIgnoreCase("NB!")) { offset++; } if (offset == parts.length) { continue; } String status = parts[offset++].trim();; //fields.get(6).text().trim(); - extract from url departure.setStatus(status); String note = ""; //extractNote( fields.get(7) ); - extract from url departure.setNote(note); } } else { logger.warning("No departures found for station=" + stationcode + ", type=" + traintype); } return departureBean; } public static String cleanText(String input) { //apparently JSoup translates   characters on www.bane.dk to 0xA0 return input.replace((char) 0xA0, (char)0x20).trim(); } // old www site is not available any more @Deprecated public DepartureBean lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception { DepartureBean departureBean = new DepartureBean(); String type = getTypeStringWww(trainType); stationcode = URLEncoder.encode(stationcode, "ISO-8859-1"); String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode; logger.fine("URI:" + uri); JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), settings.getReplyTimeout() ); CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); Element page = (Element) breaker.invoke(wrapper); String tableName = arrival == false ? "afgangtabel" : "ankomsttabel"; Element table = page.getElementById(tableName); if (table != null) { Elements tableRows = table.getElementsByTag("tr"); //boolean passedTidsstreg = false; //boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0); for (Element currentRow : tableRows) { String rowClass = currentRow.attr("class"); /* if (tidsstregExists == true && passedTidsstreg == false) { if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) { passedTidsstreg = true; } else { continue; } }*/ if (rowClass != null && rowClass.toLowerCase().contains("station") ) { Elements fields = currentRow.getElementsByTag("td"); DepartureEntry departure = new DepartureEntry(); String time = cleanText( fields.get(0).getAllElements().get(2).text() ); if (time.equals("")) time = "0:00"; //Bane.dk bug work-around departure.setTime(time); int updated = extractUpdated( fields.get(1) ); departure.setUpdated(updated); String trainNumber = cleanText( fields.get(2).text() ); if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber trainNumber = trainNumber + " " + extractTrainNumberWww(fields.get(2)); departure.setTrainNumber(trainNumber); String destination = cleanText( fields.get(3).text() ); departure.setDestination(destination); String origin = cleanText( fields.get(4).text() ); departure.setOrigin(origin); String location = cleanText( fields.get(5).text() ); departure.setLocation(location); String status = cleanText( fields.get(6).text() ); departure.setStatus(status); String note = cleanText( extractNote( fields.get(7) ) ); departure.setNote(note); departure.setType(type); departureBean.entries.add(departure); } } } else { logger.warning("No departures found for station=" + stationcode + ", type=" + type); } return departureBean; } private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" int updated = -1; Elements updatedImgs = updatedTd.getElementsByTag("img"); String updatedStr = updatedImgs.get(0).attr("src"); if (updatedStr != null) { for (int i=0; i 0 && note.charAt(note.length()-1) == 'i') note = note.substring(0,note.length() -1 ); return note.trim(); } private String extractTrainNumberAzure(Element trainTd) { Element anchorElement = trainTd.getElementsByTag("a").get(0); String href = anchorElement.attr("href"); int pos = href.lastIndexOf('/'); String number = href.substring(pos+1); return number; } private String extractTrainNumberMobile(String link) { Map elements = HttpUtil.decodeParams(link); return elements.get("tognummer"); } private String extractTrainNumberWww(Element trainTd) { String number = ""; Element anchorElement = trainTd.getElementsByTag("a").get(0); String href = anchorElement.attr("href"); String argstring = href.split("?")[1]; Map elements = HttpUtil.decodeParams(argstring); number = elements.get("TogNr"); /*String argstring = href.substring( href.indexOf('?') + 1); String args[] = argstring.split("&"); for (String arg : args) { String pair[] = arg.split("="); // Key=pair[0], Value=pair[1] if (pair[0].equalsIgnoreCase("TogNr")) number = pair[1]; }*/ return number; } //test /* public static void main(String args[]) throws Exception { DepartureFetcher f = new DepartureFetcher(); List deps = f.lookupDepartures("AR", "FJRN"); for(DepartureBean d : deps) { System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" + d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() ); } System.out.println("--------------------------"); }*/ }