--- android/TrainInfoService/src/dk/thoerup/traininfoservice/banedk/DepartureFetcher.java 2010/07/09 22:30:18 974 +++ android/TrainInfoService/src/dk/thoerup/traininfoservice/banedk/DepartureFetcher.java 2010/09/14 05:20:17 1045 @@ -1,16 +1,15 @@ package dk.thoerup.traininfoservice.banedk; -import java.util.ArrayList; + +import java.net.URL; +import java.net.URLEncoder; import java.util.Collections; -import java.util.List; import java.util.Map; import java.util.logging.Logger; -import com.gargoylesoftware.htmlunit.BrowserVersion; -import com.gargoylesoftware.htmlunit.WebClient; -import com.gargoylesoftware.htmlunit.html.DomNodeList; -import com.gargoylesoftware.htmlunit.html.HtmlElement; -import com.gargoylesoftware.htmlunit.html.HtmlPage; +import org.jsoup.nodes.Document; +import org.jsoup.nodes.Element; +import org.jsoup.select.Elements; import dk.thoerup.circuitbreaker.CircuitBreaker; import dk.thoerup.circuitbreaker.CircuitBreakerManager; @@ -27,69 +26,78 @@ Logger logger = Logger.getLogger(DepartureFetcher.class.getName()); - Map> cache; + Map cache; StationDAO stationDao = new StationDAO(); - private boolean useTempSite; + private boolean useAzureSite; + private int replyTimeout; - public DepartureFetcher(boolean tempSite, int cacheTimeout) { - useTempSite = tempSite; - cache = new TimeoutMap>(cacheTimeout); + public DepartureFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) { + this.replyTimeout = replyTimeout; + useAzureSite = azureSite; + cache = new TimeoutMap(cacheTimeout); } - public List cachedLookupDepartures(int stationID, boolean arrival) throws Exception { + public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception { final String key = "" + stationID + ":" + arrival; - List list = cache.get(key); + DepartureBean departureBean = cache.get(key); - if (list == null) { - list = lookupDepartures(stationID,arrival); - cache.put(key, list); + if (departureBean == null) { + departureBean = lookupDepartures(stationID,arrival); + cache.put(key, departureBean); } else { Statistics.getInstance().incrementDepartureCacheHits(); logger.info("Departure: Cache hit " + key); //remove before production } - return list; + return departureBean; } - public List lookupDepartures(int stationID, boolean arrival) throws Exception { - List departureList = new ArrayList(); + public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception { + + DepartureBean departureBean = new DepartureBean(); StationBean station = stationDao.getById(stationID); + departureBean.stationName = station.getName(); + if (station.getRegional() != null) { - List list = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival); - departureList.addAll(list); + DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival); + departureBean.departureEntries.addAll( tempBean.departureEntries ); + departureBean.notifications.addAll(tempBean.notifications); } if (station.getStrain() != null) { - List list = lookupDepartures(station.getStrain(), TrainType.STOG, arrival); - departureList.addAll(list); + DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival); + departureBean.departureEntries.addAll( tempBean.departureEntries ); + departureBean.notifications.addAll(tempBean.notifications); } - Collections.sort( departureList ); + if (departureBean.departureEntries.size() == 0) { + logger.info("No departures found for station " + stationID); + } + + Collections.sort( departureBean.departureEntries ); - return departureList; + return departureBean; } - public List lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception { - if (useTempSite == false) { - return lookupDeparturesNormalSite(stationcode, type, arrival); - } else { - //return lookupDeparturesFromTemporarySite(stationcode, type); - //TODO: find out what to to if they ever put a temp site up on trafikinfo.bane.dk - return null; + public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception { + if (useAzureSite == true) { + return lookupDeparturesAzureSite(stationcode, type, arrival); + } else { + return lookupDeparturesWwwSite(stationcode, type, arrival); } } - private String getTypeString(TrainType type) { + private String getTypeStringAzure(TrainType type) { switch (type) { case STOG: return "S-Tog"; @@ -100,41 +108,62 @@ } } - public List lookupDeparturesNormalSite(String stationcode, TrainType type, boolean arrival) throws Exception { - - List departureList = new ArrayList(); + 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 { - final WebClient webClient = new WebClient( BrowserVersion.FIREFOX_3 ); - webClient.setTimeout(2500); - webClient.setJavaScriptEnabled(false); + DepartureBean departureBean = new DepartureBean(); + - - String typeString = getTypeString(type); + String typeString = getTypeStringAzure(type); String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst"; - - //String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode; - String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning"; + + stationcode = URLEncoder.encode(stationcode,"ISO-8859-1"); - //logger.info("URI: " + uri); - HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri); + String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning"; + + //logger.info("URI: " + uri); + JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout); CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); - HtmlPage page = (HtmlPage) breaker.invoke(wrapper); + Document page = (Document) breaker.invoke(wrapper); String tableName = arrival == false ? "afgangtabel" : "ankomsttabel"; - HtmlElement table = page.getElementById(tableName); + Element table = page.getElementById(tableName); if (table != null) { - DomNodeList tableRows = table.getElementsByTagName("tr"); + Elements tableRows = table.getElementsByTag("tr"); - for (HtmlElement currentRow : tableRows) { - String rowClass = currentRow.getAttribute("class"); + 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") ) { - DomNodeList fields = currentRow.getElementsByTagName("td"); + + Elements fields = currentRow.getElementsByTag("td"); - DepartureBean departure = new DepartureBean(); + DepartureEntry departure = new DepartureEntry(); - String time = fields.get(0).asText(); + String time = fields.get(0).text(); if (time.equals("")) time = "0:00"; //Bane.dk bug work-around departure.setTime(time); @@ -142,21 +171,21 @@ int updated = extractUpdated( fields.get(1) ); departure.setUpdated(updated); - String trainNumber = fields.get(2).asText(); + String trainNumber = fields.get(2).text(); if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber - trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2)); + trainNumber = trainNumber + " " + extractTrainNumberAzure(fields.get(2)); departure.setTrainNumber(trainNumber); - String destination = fields.get(3).asText(); + String destination = fields.get(3).text(); departure.setDestination(destination); - String origin = fields.get(4).asText(); + String origin = fields.get(4).text(); departure.setOrigin(origin); - String location = fields.get(5).asText(); + String location = fields.get(5).text(); departure.setLocation(location); - String status = fields.get(6).asText().trim(); + String status = fields.get(6).text().trim(); departure.setStatus(status); String note = extractNote( fields.get(7) ); @@ -164,91 +193,118 @@ departure.setType(typeString); - departureList.add(departure); + departureBean.departureEntries.add( departure ); } } } else { logger.warning("No departures found for station=" + stationcode + ", type=" + type); } - webClient.closeAllWindows(); - return departureList; + 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; } - /* - @Deprecated - public List lookupDeparturesFromTemporarySite(String stationcode, String type) throws Exception { + + + public static String cleanText(String input) { + //apparently JSoup translates   characters on www.bane.dk to 0xA0 + return input.replace((char) 0xA0, (char)0x20).trim(); + } + + public DepartureBean lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception { - List departureList = new ArrayList(); + DepartureBean departureBean = new DepartureBean(); - final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3); - webClient.setTimeout(2500); - webClient.setJavaScriptEnabled(false); - - - String uri = "http://bane.dk/lite/station.asp?w=" + type + "&s=" + stationcode; - - HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri); + 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.info("URI:" + uri); + JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout); CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); - HtmlPage page = (HtmlPage) breaker.invoke(wrapper); + Element page = (Element) breaker.invoke(wrapper); - HtmlElement table = page.getElementById("traf_afgang"); + String tableName = arrival == false ? "afgangtabel" : "ankomsttabel"; + Element table = page.getElementById(tableName); - if (table != null) { - DomNodeList tableRows = table.getElementsByTagName("tr"); - - boolean isFirst = true; + if (table != null) { + Elements tableRows = table.getElementsByTag("tr"); - for (HtmlElement currentRow : tableRows) { - if (isFirst == true) { //skip table headers - isFirst = false; - continue; - } - - DomNodeList fields = currentRow.getElementsByTagName("td"); - - DepartureBean departure = new DepartureBean(); - - String time = fields.get(0).asText().trim(); - - if (time.equals("")) - time = "0:00"; //Bane.dk bug work-around - departure.setTime(time); - - - String trainNumber = fields.get(1).asText(); - departure.setTrainNumber(trainNumber); - - String destination = fields.get(2).asText(); - departure.setDestination(destination); - - String origin = fields.get(3).asText(); - departure.setOrigin(origin); - - String status = fields.get(4).asText(); - departure.setStatus(status); - - String note = fields.get(5).asText(); - departure.setNote(note); + for (Element currentRow : tableRows) { + String rowClass = currentRow.attr("class"); + if (rowClass != null && rowClass.toLowerCase().contains("station") ) { + Elements fields = currentRow.getElementsByTag("td"); + + DepartureEntry departure = new DepartureEntry(); + - departureList.add(departure); + + 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.departureEntries.add(departure); + + + } } } else { logger.warning("No departures found for station=" + stationcode + ", type=" + type); } - webClient.closeAllWindows(); - return departureList; - }*/ - + return departureBean; + } + - private int extractUpdated(HtmlElement updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" + private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" int updated = -1; - DomNodeList updatedImgs = updatedTd.getElementsByTagName("img"); - String updatedStr = updatedImgs.get(0).getAttribute("src"); + Elements updatedImgs = updatedTd.getElementsByTag("img"); + String updatedStr = updatedImgs.get(0).attr("src"); if (updatedStr != null) { for (int i=0; i elems = noteTd.getElementsByAttribute("span", "class", "bemtype"); + + Elements elems = noteTd.getElementsByClass("bemtype"); if (elems.size() > 0 && note.charAt(note.length()-1) == 'i') note = note.substring(0,note.length() -1 ); - return note; + return note.trim(); } - private String extractTrainNumber(HtmlElement trainTd) { - HtmlElement anchorElement = trainTd.getElementsByTagName("a").get(0); - String href = anchorElement.getAttribute("href"); + 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); @@ -282,6 +339,25 @@ return number; } + private String extractTrainNumberWww(Element trainTd) { + String number = ""; + Element anchorElement = trainTd.getElementsByTag("a").get(0); + String href = anchorElement.attr("href"); + 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 {