/[projects]/android/TrainInfoService/src/dk/thoerup/traininfoservice/banedk/DepartureFetcher.java
ViewVC logotype

Contents of /android/TrainInfoService/src/dk/thoerup/traininfoservice/banedk/DepartureFetcher.java

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1333 - (show annotations) (download)
Wed Apr 20 06:26:19 2011 UTC (13 years ago) by torben
File size: 16567 byte(s)
make temp mobile site parser work with stog
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.android.traininfo.common.DepartureBean;
15 import dk.thoerup.android.traininfo.common.DepartureEntry;
16 import dk.thoerup.android.traininfo.common.StationBean.StationEntry;
17 import dk.thoerup.circuitbreaker.CircuitBreaker;
18 import dk.thoerup.circuitbreaker.CircuitBreakerManager;
19 import dk.thoerup.traininfoservice.Statistics;
20 import dk.thoerup.traininfoservice.TraininfoSettings;
21 import dk.thoerup.traininfoservice.db.StationDAO;
22
23 public class DepartureFetcher {
24
25 enum TrainType{
26 STOG,
27 REGIONAL
28 }
29
30 enum FetchTrainType {
31 STOG,
32 REGIONAL,
33 BOTH
34 }
35
36 Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
37
38 Map<String, DepartureBean> cache;
39
40 StationDAO stationDao = new StationDAO();
41
42
43 private TraininfoSettings settings;
44
45 public DepartureFetcher(TraininfoSettings settings) {
46 this.settings = settings;
47 cache = new TimeoutMap<String,DepartureBean>( settings.getCacheTimeout() );
48 }
49
50
51
52
53 public DepartureBean cachedLookupDepartures(int stationID, boolean arrival, FetchTrainType type) throws Exception {
54
55 final String key = "" + stationID + ":" + arrival + ":" + type.toString();
56
57 DepartureBean departureBean = cache.get(key);
58
59
60 if (departureBean == null) {
61 departureBean = lookupDepartures(stationID, arrival, type);
62 cache.put(key, departureBean);
63 } else {
64 Statistics.getInstance().incrementDepartureCacheHits();
65 logger.info("Departure: Cache hit " + key); //remove before production
66 }
67 return departureBean;
68 }
69
70
71 public DepartureBean lookupDepartures(int stationID, boolean arrival, FetchTrainType type) throws Exception {
72
73 DepartureBean departureBean = new DepartureBean();
74
75 StationEntry station = stationDao.getById(stationID);
76
77 departureBean.stationName = station.getName();
78
79 if (station.getRegional() != null && (type == FetchTrainType.REGIONAL||type == FetchTrainType.BOTH) ) {
80 DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
81 departureBean.entries.addAll( tempBean.entries );
82 departureBean.notifications.addAll(tempBean.notifications);
83 }
84
85 if (station.getStrain() != null && (type == FetchTrainType.STOG||type == FetchTrainType.BOTH)) {
86 DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
87 departureBean.entries.addAll( tempBean.entries );
88 departureBean.notifications.addAll(tempBean.notifications);
89 }
90
91 if (departureBean.entries.size() == 0) {
92 logger.info("No departures found for station " + stationID);
93 }
94
95 if (type == FetchTrainType.BOTH) { //if we have both S-tog and regional order by departure/arrival time
96 Collections.sort( departureBean.entries );
97 }
98
99
100 return departureBean;
101 }
102
103 public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
104 if ( settings.getUseAzureSite() == true) {
105 return lookupDeparturesAzureSite(stationcode, type, arrival);
106 } else {
107 return lookupDeparturesMobileSite(stationcode, type, arrival);
108 }
109 }
110
111 private String getTypeStringAzure(TrainType type) {
112 switch (type) {
113 case STOG:
114 return "S-Tog";
115 case REGIONAL:
116 return "Fjerntog";
117 default:
118 return ""; //Can not happen
119 }
120 }
121
122 private String getTypeStringWww(TrainType type) {
123 switch (type) {
124 case STOG:
125 return "S2";
126 case REGIONAL:
127 return "FJRN";
128 default:
129 return ""; //Can not happen
130 }
131 }
132
133 public DepartureBean lookupDeparturesAzureSite(String stationcode, TrainType type, boolean arrival) throws Exception {
134
135 DepartureBean departureBean = new DepartureBean();
136
137
138 String typeString = getTypeStringAzure(type);
139 String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
140
141 stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
142
143 String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
144
145 logger.fine("URI: " + uri);
146 JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), settings.getReplyTimeout() );
147 CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
148
149 Document page = (Document) breaker.invoke(wrapper);
150
151 String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
152 Element table = page.getElementById(tableName);
153
154 if (table != null) {
155 Elements tableRows = table.getElementsByTag("tr");
156
157 //boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
158 //boolean passedTidsstreg = false;
159
160 for (Element currentRow : tableRows) {
161 String rowClass = currentRow.attr("class");
162 /*
163 if (tidsstregExists == true && passedTidsstreg == false) {
164 if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
165 passedTidsstreg = true;
166 } else {
167 continue;
168 }
169 }*/
170
171 if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
172
173 Elements fields = currentRow.getElementsByTag("td");
174
175 DepartureEntry departure = new DepartureEntry();
176
177 String time = fields.get(0).text();
178 if (time.equals(""))
179 time = "0:00"; //Bane.dk bug work-around
180 departure.setTime(time);
181
182 int updated = extractUpdated( fields.get(1) );
183 departure.setUpdated(updated);
184
185 String trainNumber = fields.get(2).text();
186 if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
187 trainNumber = trainNumber + " " + extractTrainNumberAzure(fields.get(2));
188 departure.setTrainNumber(trainNumber);
189
190 String destination = fields.get(3).text();
191 departure.setDestination(destination);
192
193 String origin = fields.get(4).text();
194 departure.setOrigin(origin);
195
196 String location = fields.get(5).text();
197 departure.setLocation(location);
198
199 String status = fields.get(6).text().trim();
200 departure.setStatus(status);
201
202 String note = extractNote( fields.get(7) );
203 departure.setNote(note);
204
205 departure.setType(typeString);
206
207 departureBean.entries.add( departure );
208 }
209 }
210 } else {
211 logger.warning("No departures found for station=" + stationcode + ", type=" + type);
212 }
213
214 Element notifDiv = page.getElementById("station_planlagte_text");
215 if (notifDiv != null) {
216
217 Elements tables = notifDiv.getElementsByTag("table");
218 for (Element tab : tables) {
219
220 Elements anchors = tab.getElementsByTag("a");
221 if (anchors.size() == 2) {
222 departureBean.notifications.add( anchors.get(1).text() );
223 }
224 }
225
226 }
227
228
229 return departureBean;
230 }
231
232 public DepartureBean lookupDeparturesMobileSite(String stationcode, TrainType traintype, boolean arrival) throws Exception {
233
234 DepartureBean departureBean = new DepartureBean();
235
236
237 String typeString = getTypeStringWww(traintype);
238 String arrivalDeparture = (arrival==false) ? "afgang" : "ankomst";
239
240 stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
241
242 //String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
243 String uri = "http://mobil.bane.dk/mobilStation.asp?artikelID=5332&stat_kode=" + stationcode + "&webprofil=" + typeString +"&beskrivelse=&mode=ankomstafgang&ankomstafgang=" + arrivalDeparture + "&gemstation=&fuldvisning=1";
244 logger.fine("URI: " + uri);
245 JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), settings.getReplyTimeout() );
246 CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
247
248 Document page = (Document) breaker.invoke(wrapper);
249
250
251 Element content = page.getElementsByClass("contentDiv").get(0);
252
253
254 if (content != null) {
255 Elements tableRows = content.child(0).children();
256
257
258
259 for (Element currentRow : tableRows) {
260 if (currentRow.tagName().equals("br") ) {
261 break;
262 }
263
264
265 Element link = currentRow.child(0);
266 System.out.println( currentRow.text() );;
267
268
269 String parts[] = currentRow.text().split(",");
270
271
272 DepartureEntry departure = new DepartureEntry();
273
274 /*
275 http://mobil.bane.dk/mobilStation.asp?artikelID=5332&tognummer=111&webprofil=FJRN&mode=rute&strBemaerkning=Afg%E5r+fra+%C5rhus+H+kl%2E07%3A21++&strRefURL=%2FmobilStation%2Easp%3FartikelID%3D5332%26stat%5Fkode%3DAR%26webprofil%3DFJRN%26beskrivelse%3D%25C5rhus%2BH%26mode%3Dankomstafgang%26ankomstafgang%3Dafgang%26gemstation%3D
276 */
277 int offset = 0;
278
279 String time = parts[offset++];
280 if (time.equals(""))
281 time = "0:00"; //Bane.dk bug work-around
282 departure.setTime(time);
283
284 int updated = 4; //does not exist on mobile
285 departure.setUpdated(updated);
286
287 String trainNumber = "-"; //extractTrainNumberAzure(fields.get(2));
288 /*if (traintype == TrainType.STOG) //If it is S-train we need to extract the trainNumber
289 trainNumber = trainNumber + " " + extractTrainNumberAzure(fields.get(2));*/
290 departure.setTrainNumber(trainNumber);
291
292 if (trainType == TrainType.STOG) { //if it is stog the next vield is the "Line" code - this should be used somewhere, but skippint ahead for now
293 offset++;
294 }
295
296 String destination = parts[offset++];
297 departure.setDestination(destination);
298
299 String origin = "-"; // fields.get(4).text(); does not exist on mobile
300 departure.setOrigin(origin);
301
302 String location = ""; // fields.get(5).text(); does not exist on mobile
303 departure.setLocation(location);
304
305 String status = ""; //fields.get(6).text().trim(); - extract from url
306 departure.setStatus(status);
307
308 String note = ""; //extractNote( fields.get(7) ); - extract from url
309 departure.setNote(note);
310
311 departure.setType(typeString);
312
313 departureBean.entries.add( departure );
314
315 }
316 } else {
317 logger.warning("No departures found for station=" + stationcode + ", type=" + traintype);
318 }
319
320 return departureBean;
321 }
322
323
324
325 public static String cleanText(String input) {
326 //apparently JSoup translates &nbsp; characters on www.bane.dk to 0xA0
327 return input.replace((char) 0xA0, (char)0x20).trim();
328 }
329
330
331 // old www site is not available any more
332 @Deprecated
333 public DepartureBean lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception {
334
335 DepartureBean departureBean = new DepartureBean();
336
337 String type = getTypeStringWww(trainType);
338
339 stationcode = URLEncoder.encode(stationcode, "ISO-8859-1");
340
341
342 String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
343 logger.fine("URI:" + uri);
344
345
346 JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), settings.getReplyTimeout() );
347 CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
348
349 Element page = (Element) breaker.invoke(wrapper);
350
351 String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
352 Element table = page.getElementById(tableName);
353
354
355
356 if (table != null) {
357 Elements tableRows = table.getElementsByTag("tr");
358
359 //boolean passedTidsstreg = false;
360 //boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
361
362 for (Element currentRow : tableRows) {
363 String rowClass = currentRow.attr("class");
364 /*
365 if (tidsstregExists == true && passedTidsstreg == false) {
366 if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
367 passedTidsstreg = true;
368 } else {
369 continue;
370 }
371 }*/
372
373
374 if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
375 Elements fields = currentRow.getElementsByTag("td");
376
377 DepartureEntry departure = new DepartureEntry();
378
379
380
381 String time = cleanText( fields.get(0).getAllElements().get(2).text() );
382 if (time.equals(""))
383 time = "0:00"; //Bane.dk bug work-around
384 departure.setTime(time);
385
386 int updated = extractUpdated( fields.get(1) );
387 departure.setUpdated(updated);
388
389 String trainNumber = cleanText( fields.get(2).text() );
390 if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber
391 trainNumber = trainNumber + " " + extractTrainNumberWww(fields.get(2));
392 departure.setTrainNumber(trainNumber);
393
394 String destination = cleanText( fields.get(3).text() );
395 departure.setDestination(destination);
396
397 String origin = cleanText( fields.get(4).text() );
398 departure.setOrigin(origin);
399
400 String location = cleanText( fields.get(5).text() );
401 departure.setLocation(location);
402
403 String status = cleanText( fields.get(6).text() );
404 departure.setStatus(status);
405
406 String note = cleanText( extractNote( fields.get(7) ) );
407 departure.setNote(note);
408
409 departure.setType(type);
410
411 departureBean.entries.add(departure);
412
413
414 }
415 }
416 } else {
417 logger.warning("No departures found for station=" + stationcode + ", type=" + type);
418 }
419
420
421 return departureBean;
422 }
423
424
425 private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
426 int updated = -1;
427
428 Elements updatedImgs = updatedTd.getElementsByTag("img");
429 String updatedStr = updatedImgs.get(0).attr("src");
430
431 if (updatedStr != null) {
432 for (int i=0; i<updatedStr.length(); i++) {
433 char c = updatedStr.charAt(i);
434 if ( Character.isDigit(c)) {
435 updated = Character.digit(c, 10);
436 break;
437 }
438 }
439 }
440 return updated;
441 }
442
443 private String extractNote(Element noteTd) {
444 String note = noteTd.text().trim();
445
446
447 Elements elems = noteTd.getElementsByClass("bemtype");
448 if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
449 note = note.substring(0,note.length() -1 );
450
451 return note.trim();
452 }
453
454 private String extractTrainNumberAzure(Element trainTd) {
455 Element anchorElement = trainTd.getElementsByTag("a").get(0);
456 String href = anchorElement.attr("href");
457
458 int pos = href.lastIndexOf('/');
459 String number = href.substring(pos+1);
460
461 return number;
462 }
463
464 private String extractTrainNumberWww(Element trainTd) {
465 String number = "";
466 Element anchorElement = trainTd.getElementsByTag("a").get(0);
467 String href = anchorElement.attr("href");
468 String argstring = href.substring( href.indexOf('?') + 1);
469
470 String args[] = argstring.split("&");
471 for (String arg : args) {
472 String pair[] = arg.split("="); // Key=pair[0], Value=pair[1]
473
474 if (pair[0].equalsIgnoreCase("TogNr"))
475 number = pair[1];
476 }
477
478
479 return number;
480 }
481
482
483 //test
484 /*
485 public static void main(String args[]) throws Exception {
486 DepartureFetcher f = new DepartureFetcher();
487 List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
488 for(DepartureBean d : deps) {
489 System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
490 d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
491 }
492
493 System.out.println("--------------------------");
494 }*/
495 }

  ViewVC Help
Powered by ViewVC 1.1.20