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