traininfoservice/banedk/DepartureFetcher.java

package dk.thoerup.traininfoservice.banedk;


import java.net.URL;
import java.net.URLEncoder;
import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
import java.util.Map;
import java.util.logging.Level;
import java.util.logging.Logger;

import net.sf.jsr107cache.Cache;
import net.sf.jsr107cache.CacheException;
import net.sf.jsr107cache.CacheManager;

import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import com.google.appengine.api.memcache.jsr107cache.GCacheFactory;

import dk.thoerup.android.traininfo.common.DepartureBean;
import dk.thoerup.android.traininfo.common.DepartureEntry;
import dk.thoerup.android.traininfo.common.StationEntry;
import dk.thoerup.circuitbreaker.CircuitBreaker;
import dk.thoerup.circuitbreaker.CircuitBreakerManager;
import dk.thoerup.traininfoservice.StationDAO;
import dk.thoerup.traininfoservice.Statistics;

public class DepartureFetcher {
        
        enum TrainType{
                STOG,
                REGIONAL
        }
        Cache cache;
        
        Logger logger = Logger.getLogger(DepartureFetcher.class.getName());     
        
        StationDAO stationDao = new StationDAO();
        
        private boolean useAzureSite;
        private int replyTimeout;
        
        Comparator<DepartureEntry> departureTimeComparator = new Comparator<DepartureEntry>() {

                @Override
                public int compare(DepartureEntry arg0, DepartureEntry arg1) {          
                                String timeStr1 = arg0.getTime().replace(":","").trim();
                                String timeStr2 = arg1.getTime().replace(":","").trim();
                                
                                int time1 = 0;
                                int time2 = 0;
                                
                                if (timeStr1.length() > 0)
                                        time1 = Integer.parseInt(timeStr1);
                                
                                if (timeStr2.length() > 0)
                                        time2 = Integer.parseInt(timeStr2);
                                
                                //work correctly when clock wraps around at midnight
                                if (Math.abs(time1-time2) < 1200) {
                                        if (time1 > time2)
                                                return 1;
                                        else
                                                return -1;
                                } else {
                                        if (time1 < time2)
                                                return 1;
                                        else
                                                return -1;

                                }               
                }
        };
        
        @SuppressWarnings("unchecked")
        public DepartureFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
                this.replyTimeout = replyTimeout;
                useAzureSite = azureSite;

        Map props = new HashMap();
        props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, cacheTimeout);        
                
                try {
            cache = CacheManager.getInstance().getCacheFactory().createCache(props);            
        } catch (CacheException e) {
                logger.log(Level.WARNING, "error creating cache", e);
        }

        }
        
        
        public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
                final String key = "departure:" + stationID + ":" + arrival;
                
                DepartureBean departureBean = (DepartureBean) cache.get(key);
                
                if (departureBean == null) {
                        departureBean = lookupDepartures(stationID,arrival);
                        cache.put(key, departureBean);
                        logger.info("Departure: Cache miss " + key + " !!! "); //remove before production
                } else {
                        Statistics.getInstance().incrementDepartureCacheHits();
                        logger.info("Departure: Cache hit " + key); 
                }
                
                return departureBean;
        }
                

        public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
                
                DepartureBean departureBean = new DepartureBean();
                
                StationEntry station = stationDao.getById(stationID);

                
                departureBean.stationName = station.getName();
                
                if (station.getRegional() != null) {
                        DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
                        departureBean.entries.addAll( tempBean.entries );
                        departureBean.notifications.addAll(tempBean.notifications);
                }
                
                if (station.getStrain() != null) {
                        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);
                }
                
                Collections.sort( departureBean.entries, departureTimeComparator );

                
                return departureBean;
        }
        
        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 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), replyTimeout);
            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 static String cleanText(String input) {
                //apparently JSoup translates &nbsp; 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 {
                
                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), replyTimeout); 
            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<updatedStr.length(); i++) {
                                char c = updatedStr.charAt(i);
                                if ( Character.isDigit(c))  {
                                        updated = Character.digit(c, 10);
                                        break;
                                }
                        }
                }               
                return updated;
        }
        
        private String extractNote(Element noteTd) {
                String note = noteTd.text().trim();
                
                
                Elements elems = noteTd.getElementsByClass("bemtype");
                if (elems.size() > 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 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 {
                DepartureFetcher f = new DepartureFetcher();
                List<DepartureBean> 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("--------------------------");
        }*/
}
1	torben	305	package dk.thoerup.traininfoservice.banedk;
2
3	torben	978
4	torben	992	import java.net.URL;
5	torben	994	import java.net.URLEncoder;
6	torben	307	import java.util.Collections;
7	torben	1564	import java.util.Comparator;
8	torben	1093	import java.util.HashMap;
9	torben	428	import java.util.Map;
10	torben	1093	import java.util.logging.Level;
11	torben	348	import java.util.logging.Logger;
12	torben	305
13	torben	1093	import net.sf.jsr107cache.Cache;
14			import net.sf.jsr107cache.CacheException;
15			import net.sf.jsr107cache.CacheManager;
16
17	torben	992	import org.jsoup.nodes.Document;
18			import org.jsoup.nodes.Element;
19			import org.jsoup.select.Elements;
20	torben	305
21	torben	1093	import com.google.appengine.api.memcache.jsr107cache.GCacheFactory;
22
23	torben	1061	import dk.thoerup.android.traininfo.common.DepartureBean;
24			import dk.thoerup.android.traininfo.common.DepartureEntry;
25	torben	1415	import dk.thoerup.android.traininfo.common.StationEntry;
26	torben	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
27			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
28	torben	588	import dk.thoerup.traininfoservice.StationDAO;
29	torben	711	import dk.thoerup.traininfoservice.Statistics;
30	torben	307
31	torben	305	public class DepartureFetcher {
32	torben	348
33	torben	972	enum TrainType{
34			STOG,
35			REGIONAL
36			}
37	torben	1093	Cache cache;
38	torben	972
39	torben	1093	Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
40	torben	387
41	torben	588	StationDAO stationDao = new StationDAO();
42
43	torben	1034	private boolean useAzureSite;
44	torben	1026	private int replyTimeout;
45	torben	580
46	torben	1564	Comparator<DepartureEntry> departureTimeComparator = new Comparator<DepartureEntry>() {
47
48			@Override
49			public int compare(DepartureEntry arg0, DepartureEntry arg1) {
50			String timeStr1 = arg0.getTime().replace(":","").trim();
51			String timeStr2 = arg1.getTime().replace(":","").trim();
52
53			int time1 = 0;
54			int time2 = 0;
55
56			if (timeStr1.length() > 0)
57			time1 = Integer.parseInt(timeStr1);
58
59			if (timeStr2.length() > 0)
60			time2 = Integer.parseInt(timeStr2);
61
62			//work correctly when clock wraps around at midnight
63			if (Math.abs(time1-time2) < 1200) {
64			if (time1 > time2)
65			return 1;
66			else
67			return -1;
68			} else {
69			if (time1 < time2)
70			return 1;
71			else
72			return -1;
73
74			}
75			}
76			};
77
78	torben	1093	@SuppressWarnings("unchecked")
79	torben	1034	public DepartureFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
80	torben	1026	this.replyTimeout = replyTimeout;
81	torben	1034	useAzureSite = azureSite;
82	torben	1093
83			Map props = new HashMap();
84			props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, cacheTimeout);
85
86			try {
87			cache = CacheManager.getInstance().getCacheFactory().createCache(props);
88			} catch (CacheException e) {
89			logger.log(Level.WARNING, "error creating cache", e);
90			}
91
92	torben	580	}
93
94
95	torben	307
96	torben	387
97	torben	978	public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
98	torben	1093	final String key = "departure:" + stationID + ":" + arrival;
99	torben	980
100	torben	1093	DepartureBean departureBean = (DepartureBean) cache.get(key);
101	torben	387
102	torben	978	if (departureBean == null) {
103			departureBean = lookupDepartures(stationID,arrival);
104			cache.put(key, departureBean);
105	torben	1093	logger.info("Departure: Cache miss " + key + " !!! "); //remove before production
106	torben	387	} else {
107	torben	711	Statistics.getInstance().incrementDepartureCacheHits();
108	torben	1093	logger.info("Departure: Cache hit " + key);
109	torben	387	}
110	torben	1093
111	torben	980	return departureBean;
112	torben	387	}
113
114
115	torben	978	public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
116	torben	307
117	torben	978	DepartureBean departureBean = new DepartureBean();
118
119	torben	1105	StationEntry station = stationDao.getById(stationID);
120
121	torben	307
122	torben	1021	departureBean.stationName = station.getName();
123
124	torben	588	if (station.getRegional() != null) {
125	torben	978	DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
126	torben	1063	departureBean.entries.addAll( tempBean.entries );
127	torben	978	departureBean.notifications.addAll(tempBean.notifications);
128	torben	307	}
129
130	torben	588	if (station.getStrain() != null) {
131	torben	978	DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
132	torben	1063	departureBean.entries.addAll( tempBean.entries );
133	torben	978	departureBean.notifications.addAll(tempBean.notifications);
134	torben	588	}
135
136	torben	1063	if (departureBean.entries.size() == 0) {
137	torben	1037	logger.info("No departures found for station " + stationID);
138			}
139
140	torben	1564	Collections.sort( departureBean.entries, departureTimeComparator );
141	torben	588
142
143	torben	978	return departureBean;
144	torben	305	}
145
146	torben	978	public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
147	torben	1034	if (useAzureSite == true) {
148			return lookupDeparturesAzureSite(stationcode, type, arrival);
149			} else {
150			return lookupDeparturesWwwSite(stationcode, type, arrival);
151	torben	580	}
152			}
153
154	torben	1034	private String getTypeStringAzure(TrainType type) {
155	torben	972	switch (type) {
156			case STOG:
157			return "S-Tog";
158			case REGIONAL:
159			return "Fjerntog";
160			default:
161			return ""; //Can not happen
162			}
163			}
164
165	torben	1034	private String getTypeStringWww(TrainType type) {
166			switch (type) {
167			case STOG:
168			return "S2";
169			case REGIONAL:
170			return "FJRN";
171			default:
172			return ""; //Can not happen
173			}
174			}
175
176			public DepartureBean lookupDeparturesAzureSite(String stationcode, TrainType type, boolean arrival) throws Exception {
177	torben	305
178	torben	978	DepartureBean departureBean = new DepartureBean();
179	torben	992
180	torben	970
181	torben	1034	String typeString = getTypeStringAzure(type);
182	torben	970	String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
183	torben	994
184			stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
185	torben	970
186	torben	1034	String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
187	torben	994
188	torben	1048	logger.fine("URI: " + uri);
189	torben	1026	JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout);
190	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
191	torben	305
192	torben	992	Document page = (Document) breaker.invoke(wrapper);
193	torben	305
194	torben	829	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
195	torben	992	Element table = page.getElementById(tableName);
196	torben	829
197	torben	342	if (table != null) {
198	torben	992	Elements tableRows = table.getElementsByTag("tr");
199	torben	342
200	torben	1020	boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
201			boolean passedTidsstreg = false;
202
203	torben	992	for (Element currentRow : tableRows) {
204			String rowClass = currentRow.attr("class");
205	torben	1020
206			if (tidsstregExists == true && passedTidsstreg == false) {
207			if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
208			passedTidsstreg = true;
209			} else {
210			continue;
211			}
212			}
213
214	torben	342	if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
215	torben	1020
216	torben	992	Elements fields = currentRow.getElementsByTag("td");
217	torben	342
218	torben	978	DepartureEntry departure = new DepartureEntry();
219	torben	342
220	torben	992	String time = fields.get(0).text();
221	torben	375	if (time.equals(""))
222			time = "0:00"; //Bane.dk bug work-around
223	torben	342	departure.setTime(time);
224
225			int updated = extractUpdated( fields.get(1) );
226			departure.setUpdated(updated);
227
228	torben	992	String trainNumber = fields.get(2).text();
229	torben	972	if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
230	torben	1039	trainNumber = trainNumber + " " + extractTrainNumberAzure(fields.get(2));
231	torben	342	departure.setTrainNumber(trainNumber);
232
233	torben	992	String destination = fields.get(3).text();
234	torben	342	departure.setDestination(destination);
235
236	torben	992	String origin = fields.get(4).text();
237	torben	342	departure.setOrigin(origin);
238
239	torben	992	String location = fields.get(5).text();
240	torben	342	departure.setLocation(location);
241
242	torben	992	String status = fields.get(6).text().trim();
243	torben	342	departure.setStatus(status);
244
245			String note = extractNote( fields.get(7) );
246			departure.setNote(note);
247
248	torben	972	departure.setType(typeString);
249	torben	697
250	torben	1063	departureBean.entries.add( departure );
251	torben	342	}
252			}
253	torben	348	} else {
254			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
255	torben	305	}
256	torben	978
257	torben	992	Element notifDiv = page.getElementById("station_planlagte_text");
258	torben	978	if (notifDiv != null) {
259
260	torben	992	Elements tables = notifDiv.getElementsByTag("table");
261			for (Element tab : tables) {
262	torben	978
263	torben	992	Elements anchors = tab.getElementsByTag("a");
264	torben	978	if (anchors.size() == 2) {
265	torben	992	departureBean.notifications.add( anchors.get(1).text() );
266	torben	978	}
267			}
268
269			}
270
271
272			return departureBean;
273	torben	305	}
274
275	torben	1038
276
277	torben	1040	public static String cleanText(String input) {
278			//apparently JSoup translates   characters on www.bane.dk to 0xA0
279	torben	1038	return input.replace((char) 0xA0, (char)0x20).trim();
280			}
281
282	torben	1034	public DepartureBean lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception {
283	torben	580
284	torben	1034	DepartureBean departureBean = new DepartureBean();
285	torben	580
286	torben	1034	String type = getTypeStringWww(trainType);
287
288	torben	1045	stationcode = URLEncoder.encode(stationcode, "ISO-8859-1");
289
290	torben	1034
291			String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
292	torben	1048	logger.fine("URI:" + uri);
293
294	torben	1047
295	torben	1034	JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout);
296	torben	580	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
297
298	torben	1034	Element page = (Element) breaker.invoke(wrapper);
299	torben	580
300	torben	1034	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
301			Element table = page.getElementById(tableName);
302	torben	580
303	torben	1046
304
305	torben	1034	if (table != null) {
306			Elements tableRows = table.getElementsByTag("tr");
307	torben	580
308	torben	1046	boolean passedTidsstreg = false;
309			boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
310
311	torben	1034	for (Element currentRow : tableRows) {
312			String rowClass = currentRow.attr("class");
313	torben	1046
314			if (tidsstregExists == true && passedTidsstreg == false) {
315			if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
316			passedTidsstreg = true;
317			} else {
318			continue;
319			}
320			}
321
322
323	torben	1034	if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
324			Elements fields = currentRow.getElementsByTag("td");
325
326			DepartureEntry departure = new DepartureEntry();
327
328
329
330	torben	1038	String time = cleanText( fields.get(0).getAllElements().get(2).text() );
331	torben	1034	if (time.equals(""))
332			time = "0:00"; //Bane.dk bug work-around
333			departure.setTime(time);
334
335			int updated = extractUpdated( fields.get(1) );
336			departure.setUpdated(updated);
337
338	torben	1038	String trainNumber = cleanText( fields.get(2).text() );
339	torben	1034	if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber
340	torben	1039	trainNumber = trainNumber + " " + extractTrainNumberWww(fields.get(2));
341	torben	1034	departure.setTrainNumber(trainNumber);
342
343	torben	1038	String destination = cleanText( fields.get(3).text() );
344	torben	1034	departure.setDestination(destination);
345
346	torben	1038	String origin = cleanText( fields.get(4).text() );
347	torben	1034	departure.setOrigin(origin);
348
349	torben	1038	String location = cleanText( fields.get(5).text() );
350	torben	1034	departure.setLocation(location);
351
352	torben	1038	String status = cleanText( fields.get(6).text() );
353	torben	1034	departure.setStatus(status);
354
355	torben	1038	String note = cleanText( extractNote( fields.get(7) ) );
356	torben	1034	departure.setNote(note);
357
358			departure.setType(type);
359
360	torben	1063	departureBean.entries.add(departure);
361	torben	1034
362
363	torben	580	}
364			}
365			} else {
366			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
367			}
368
369	torben	591
370	torben	1034	return departureBean;
371			}
372
373	torben	580
374	torben	992	private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
375	torben	305	int updated = -1;
376
377	torben	992	Elements updatedImgs = updatedTd.getElementsByTag("img");
378			String updatedStr = updatedImgs.get(0).attr("src");
379	torben	305
380			if (updatedStr != null) {
381			for (int i=0; i<updatedStr.length(); i++) {
382			char c = updatedStr.charAt(i);
383			if ( Character.isDigit(c)) {
384			updated = Character.digit(c, 10);
385			break;
386			}
387			}
388			}
389			return updated;
390			}
391
392	torben	992	private String extractNote(Element noteTd) {
393			String note = noteTd.text().trim();
394	torben	313
395	torben	992
396			Elements elems = noteTd.getElementsByClass("bemtype");
397	torben	313	if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
398			note = note.substring(0,note.length() -1 );
399
400	torben	1038	return note.trim();
401	torben	313	}
402
403	torben	1039	private String extractTrainNumberAzure(Element trainTd) {
404	torben	992	Element anchorElement = trainTd.getElementsByTag("a").get(0);
405			String href = anchorElement.attr("href");
406	torben	349
407	torben	973	int pos = href.lastIndexOf('/');
408			String number = href.substring(pos+1);
409	torben	970
410	torben	349	return number;
411			}
412
413	torben	1039	private String extractTrainNumberWww(Element trainTd) {
414			String number = "";
415			Element anchorElement = trainTd.getElementsByTag("a").get(0);
416			String href = anchorElement.attr("href");
417			String argstring = href.substring( href.indexOf('?') + 1);
418
419			String args[] = argstring.split("&");
420			for (String arg : args) {
421			String pair[] = arg.split("="); // Key=pair[0], Value=pair[1]
422
423			if (pair[0].equalsIgnoreCase("TogNr"))
424			number = pair[1];
425			}
426
427
428			return number;
429			}
430
431
432	torben	305	//test
433	torben	580	/*
434	torben	451	public static void main(String args[]) throws Exception {
435	torben	305	DepartureFetcher f = new DepartureFetcher();
436	torben	307	List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
437	torben	305	for(DepartureBean d : deps) {
438			System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
439			d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
440			}
441
442			System.out.println("--------------------------");
443	torben	580	}*/
444	torben	305	}