traininfoservice/banedk/DepartureFetcher.java

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.circuitbreaker.CircuitBreaker;
import dk.thoerup.circuitbreaker.CircuitBreakerManager;
import dk.thoerup.traininfoservice.StationBean;
import dk.thoerup.traininfoservice.StationDAO;
import dk.thoerup.traininfoservice.Statistics;

public class DepartureFetcher {
        
        enum TrainType{
                STOG,
                REGIONAL
        }
        
        Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
        
        Map<String, DepartureBean> cache;
        
        StationDAO stationDao = new StationDAO();
        
        private boolean useAzureSite;
        private int replyTimeout;
        
        public DepartureFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
                this.replyTimeout = replyTimeout;
                useAzureSite = azureSite;
                cache = new TimeoutMap<String,DepartureBean>(cacheTimeout);
        }
        
        
        public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
                final String key = "" + stationID + ":" + arrival;
                
                DepartureBean departureBean = cache.get(key);

                
                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 departureBean;
        }
                

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

                
                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.info("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 + " " + extractTrainNumber(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.departureEntries.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 lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception {
                
                DepartureBean departureBean = new DepartureBean();
                
                String type = getTypeStringWww(trainType);
                
                                            
            String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
            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");
                    
                    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();
                                

                                String time = fields.get(0).getAllElements().get(2).text();
                                if (time.equals(""))
                                        time = "0:00"; //Bane.dk bug work-around
                                departure.setTime(time);
                                logger.info("Time: " + time);// TODO: remove
                                
                                int updated = extractUpdated( fields.get(1) );
                                departure.setUpdated(updated);
                                logger.info("Updated: " + updated);// TODO: remove
                                
                                String trainNumber = fields.get(2).text();
                                if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber 
                                        trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
                                departure.setTrainNumber(trainNumber);
                                logger.info("TrainNumber: " + trainNumber);// TODO: remove
                                
                                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(type);
                                
                                departureBean.departureEntries.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;
        }
        
        private String extractTrainNumber(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;
        }
        
        //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	428	import java.util.Map;
8	torben	348	import java.util.logging.Logger;
9	torben	305
10	torben	992	import org.jsoup.nodes.Document;
11			import org.jsoup.nodes.Element;
12			import org.jsoup.select.Elements;
13	torben	305
14	torben	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
15			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
16	torben	588	import dk.thoerup.traininfoservice.StationBean;
17			import dk.thoerup.traininfoservice.StationDAO;
18	torben	711	import dk.thoerup.traininfoservice.Statistics;
19	torben	307
20	torben	305	public class DepartureFetcher {
21	torben	348
22	torben	972	enum TrainType{
23			STOG,
24			REGIONAL
25			}
26
27	torben	348	Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
28	torben	387
29	torben	978	Map<String, DepartureBean> cache;
30	torben	387
31	torben	588	StationDAO stationDao = new StationDAO();
32
33	torben	1034	private boolean useAzureSite;
34	torben	1026	private int replyTimeout;
35	torben	580
36	torben	1034	public DepartureFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
37	torben	1026	this.replyTimeout = replyTimeout;
38	torben	1034	useAzureSite = azureSite;
39	torben	978	cache = new TimeoutMap<String,DepartureBean>(cacheTimeout);
40	torben	580	}
41
42
43	torben	307
44	torben	387
45	torben	978	public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
46	torben	829	final String key = "" + stationID + ":" + arrival;
47	torben	980
48	torben	978	DepartureBean departureBean = cache.get(key);
49	torben	308
50	torben	387
51	torben	978	if (departureBean == null) {
52			departureBean = lookupDepartures(stationID,arrival);
53			cache.put(key, departureBean);
54	torben	387	} else {
55	torben	711	Statistics.getInstance().incrementDepartureCacheHits();
56	torben	829	logger.info("Departure: Cache hit " + key); //remove before production
57	torben	387	}
58	torben	980	return departureBean;
59	torben	387	}
60
61
62	torben	978	public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
63	torben	307
64	torben	978	DepartureBean departureBean = new DepartureBean();
65
66	torben	588	StationBean station = stationDao.getById(stationID);
67	torben	307
68	torben	1021	departureBean.stationName = station.getName();
69
70	torben	588	if (station.getRegional() != null) {
71	torben	978	DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
72			departureBean.departureEntries.addAll( tempBean.departureEntries );
73			departureBean.notifications.addAll(tempBean.notifications);
74	torben	307	}
75
76	torben	588	if (station.getStrain() != null) {
77	torben	978	DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
78			departureBean.departureEntries.addAll( tempBean.departureEntries );
79			departureBean.notifications.addAll(tempBean.notifications);
80	torben	588	}
81
82	torben	978	Collections.sort( departureBean.departureEntries );
83	torben	588
84
85	torben	978	return departureBean;
86	torben	305	}
87
88	torben	978	public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
89	torben	1034	if (useAzureSite == true) {
90			return lookupDeparturesAzureSite(stationcode, type, arrival);
91			} else {
92			return lookupDeparturesWwwSite(stationcode, type, arrival);
93	torben	580	}
94			}
95
96	torben	1034	private String getTypeStringAzure(TrainType type) {
97	torben	972	switch (type) {
98			case STOG:
99			return "S-Tog";
100			case REGIONAL:
101			return "Fjerntog";
102			default:
103			return ""; //Can not happen
104			}
105			}
106
107	torben	1034	private String getTypeStringWww(TrainType type) {
108			switch (type) {
109			case STOG:
110			return "S2";
111			case REGIONAL:
112			return "FJRN";
113			default:
114			return ""; //Can not happen
115			}
116			}
117
118			public DepartureBean lookupDeparturesAzureSite(String stationcode, TrainType type, boolean arrival) throws Exception {
119	torben	305
120	torben	978	DepartureBean departureBean = new DepartureBean();
121	torben	992
122	torben	970
123	torben	1034	String typeString = getTypeStringAzure(type);
124	torben	970	String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
125	torben	994
126			stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
127	torben	970
128	torben	1034	String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
129	torben	994
130	torben	992	//logger.info("URI: " + uri);
131	torben	1026	JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout);
132	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
133	torben	305
134	torben	992	Document page = (Document) breaker.invoke(wrapper);
135	torben	305
136	torben	829	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
137	torben	992	Element table = page.getElementById(tableName);
138	torben	829
139	torben	342	if (table != null) {
140	torben	992	Elements tableRows = table.getElementsByTag("tr");
141	torben	342
142	torben	1020	boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
143			boolean passedTidsstreg = false;
144
145	torben	992	for (Element currentRow : tableRows) {
146			String rowClass = currentRow.attr("class");
147	torben	1020
148			if (tidsstregExists == true && passedTidsstreg == false) {
149			if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
150			passedTidsstreg = true;
151			} else {
152			continue;
153			}
154			}
155
156	torben	342	if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
157	torben	1020
158	torben	992	Elements fields = currentRow.getElementsByTag("td");
159	torben	342
160	torben	978	DepartureEntry departure = new DepartureEntry();
161	torben	342
162	torben	992	String time = fields.get(0).text();
163	torben	375	if (time.equals(""))
164			time = "0:00"; //Bane.dk bug work-around
165	torben	342	departure.setTime(time);
166
167			int updated = extractUpdated( fields.get(1) );
168			departure.setUpdated(updated);
169
170	torben	992	String trainNumber = fields.get(2).text();
171	torben	972	if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
172	torben	349	trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
173	torben	342	departure.setTrainNumber(trainNumber);
174
175	torben	992	String destination = fields.get(3).text();
176	torben	342	departure.setDestination(destination);
177
178	torben	992	String origin = fields.get(4).text();
179	torben	342	departure.setOrigin(origin);
180
181	torben	992	String location = fields.get(5).text();
182	torben	342	departure.setLocation(location);
183
184	torben	992	String status = fields.get(6).text().trim();
185	torben	342	departure.setStatus(status);
186
187			String note = extractNote( fields.get(7) );
188			departure.setNote(note);
189
190	torben	972	departure.setType(typeString);
191	torben	697
192	torben	978	departureBean.departureEntries.add( departure );
193	torben	342	}
194			}
195	torben	348	} else {
196			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
197	torben	305	}
198	torben	978
199	torben	992	Element notifDiv = page.getElementById("station_planlagte_text");
200	torben	978	if (notifDiv != null) {
201
202	torben	992	Elements tables = notifDiv.getElementsByTag("table");
203			for (Element tab : tables) {
204	torben	978
205	torben	992	Elements anchors = tab.getElementsByTag("a");
206	torben	978	if (anchors.size() == 2) {
207	torben	992	departureBean.notifications.add( anchors.get(1).text() );
208	torben	978	}
209			}
210
211			}
212
213
214			return departureBean;
215	torben	305	}
216
217	torben	1034	public DepartureBean lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception {
218	torben	580
219	torben	1034	DepartureBean departureBean = new DepartureBean();
220	torben	580
221	torben	1034	String type = getTypeStringWww(trainType);
222
223
224			String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
225			JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout);
226	torben	580	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
227
228	torben	1034	Element page = (Element) breaker.invoke(wrapper);
229	torben	580
230	torben	1034	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
231			Element table = page.getElementById(tableName);
232	torben	580
233	torben	1034	if (table != null) {
234			Elements tableRows = table.getElementsByTag("tr");
235	torben	580
236	torben	1034	for (Element currentRow : tableRows) {
237			String rowClass = currentRow.attr("class");
238			if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
239			Elements fields = currentRow.getElementsByTag("td");
240
241			DepartureEntry departure = new DepartureEntry();
242
243
244
245			String time = fields.get(0).getAllElements().get(2).text();
246			if (time.equals(""))
247			time = "0:00"; //Bane.dk bug work-around
248			departure.setTime(time);
249			logger.info("Time: " + time);// TODO: remove
250
251			int updated = extractUpdated( fields.get(1) );
252			departure.setUpdated(updated);
253			logger.info("Updated: " + updated);// TODO: remove
254
255			String trainNumber = fields.get(2).text();
256			if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber
257			trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
258			departure.setTrainNumber(trainNumber);
259			logger.info("TrainNumber: " + trainNumber);// TODO: remove
260
261			String destination = fields.get(3).text();
262			departure.setDestination(destination);
263
264			String origin = fields.get(4).text();
265			departure.setOrigin(origin);
266
267			String location = fields.get(5).text();
268			departure.setLocation(location);
269
270			String status = fields.get(6).text().trim();
271			departure.setStatus(status);
272
273			String note = extractNote( fields.get(7) );
274			departure.setNote(note);
275
276			departure.setType(type);
277
278			departureBean.departureEntries.add(departure);
279
280
281	torben	580	}
282			}
283			} else {
284			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
285			}
286
287	torben	591
288	torben	1034	return departureBean;
289			}
290
291	torben	580
292	torben	992	private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
293	torben	305	int updated = -1;
294
295	torben	992	Elements updatedImgs = updatedTd.getElementsByTag("img");
296			String updatedStr = updatedImgs.get(0).attr("src");
297	torben	305
298			if (updatedStr != null) {
299			for (int i=0; i<updatedStr.length(); i++) {
300			char c = updatedStr.charAt(i);
301			if ( Character.isDigit(c)) {
302			updated = Character.digit(c, 10);
303			break;
304			}
305			}
306			}
307			return updated;
308			}
309
310	torben	992	private String extractNote(Element noteTd) {
311			String note = noteTd.text().trim();
312	torben	313
313	torben	992
314			Elements elems = noteTd.getElementsByClass("bemtype");
315	torben	313	if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
316			note = note.substring(0,note.length() -1 );
317
318			return note;
319			}
320
321	torben	992	private String extractTrainNumber(Element trainTd) {
322			Element anchorElement = trainTd.getElementsByTag("a").get(0);
323			String href = anchorElement.attr("href");
324	torben	349
325	torben	973	int pos = href.lastIndexOf('/');
326			String number = href.substring(pos+1);
327	torben	970
328	torben	349	return number;
329			}
330
331	torben	305	//test
332	torben	580	/*
333	torben	451	public static void main(String args[]) throws Exception {
334	torben	305	DepartureFetcher f = new DepartureFetcher();
335	torben	307	List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
336	torben	305	for(DepartureBean d : deps) {
337			System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
338			d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
339			}
340
341			System.out.println("--------------------------");
342	torben	580	}*/
343	torben	305	}