1 |
package dk.thoerup.traininfoservice.banedk; |
package dk.thoerup.traininfoservice.banedk; |
2 |
|
|
3 |
import java.sql.Connection; |
|
4 |
import java.sql.ResultSet; |
import java.net.URL; |
5 |
import java.sql.Statement; |
import java.net.URLEncoder; |
|
import java.util.ArrayList; |
|
6 |
import java.util.Collections; |
import java.util.Collections; |
7 |
import java.util.List; |
import java.util.Map; |
8 |
import java.util.logging.Logger; |
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 com.gargoylesoftware.htmlunit.WebClient; |
import dk.thoerup.android.traininfo.common.DepartureBean; |
15 |
import com.gargoylesoftware.htmlunit.html.DomNodeList; |
import dk.thoerup.android.traininfo.common.DepartureEntry; |
16 |
import com.gargoylesoftware.htmlunit.html.HtmlElement; |
import dk.thoerup.android.traininfo.common.StationBean.StationEntry; |
17 |
import com.gargoylesoftware.htmlunit.html.HtmlPage; |
import dk.thoerup.circuitbreaker.CircuitBreaker; |
18 |
|
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
19 |
import dk.thoerup.traininfoservice.DBConnection; |
import dk.thoerup.traininfoservice.Statistics; |
20 |
|
import dk.thoerup.traininfoservice.TraininfoSettings;; |
21 |
|
import dk.thoerup.traininfoservice.db.StationDAO; |
22 |
|
|
23 |
public class DepartureFetcher { |
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()); |
Logger logger = Logger.getLogger(DepartureFetcher.class.getName()); |
37 |
|
|
38 |
TimeoutCache<Integer, List<DepartureBean>> cache = new TimeoutCache<Integer,List<DepartureBean>>(120 * 1000); |
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 List<DepartureBean> cachedLookupDepartures(int stationID) throws Exception { |
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 |
|
|
|
List<DepartureBean> list = cache.get(stationID); |
|
59 |
|
|
60 |
if (list == null) { |
if (departureBean == null) { |
61 |
list = lookupDepartures(stationID); |
departureBean = lookupDepartures(stationID, arrival, type); |
62 |
cache.put(stationID, list); |
cache.put(key, departureBean); |
63 |
} else { |
} else { |
64 |
logger.info("Departure: Cache hit " + stationID); //remove before production |
Statistics.getInstance().incrementDepartureCacheHits(); |
65 |
|
logger.info("Departure: Cache hit " + key); //remove before production |
66 |
} |
} |
67 |
return list; |
return departureBean; |
68 |
} |
} |
69 |
|
|
70 |
|
|
71 |
public List<DepartureBean> lookupDepartures(int stationID) throws Exception { |
public DepartureBean lookupDepartures(int stationID, boolean arrival, FetchTrainType type) throws Exception { |
|
List<DepartureBean> departureList = new ArrayList<DepartureBean>(); |
|
72 |
|
|
73 |
Connection conn = null; |
DepartureBean departureBean = new DepartureBean(); |
|
try |
|
|
{ |
|
|
conn = DBConnection.getConnection(); |
|
74 |
|
|
75 |
String SQL = "SELECT stationcode_fjrn, stationcode_stog FROM trainstations WHERE id=" + stationID; |
StationEntry station = stationDao.getById(stationID); |
76 |
Statement stmt = conn.createStatement(); |
|
77 |
ResultSet rs = stmt.executeQuery(SQL); |
departureBean.stationName = station.getName(); |
78 |
|
|
79 |
if (rs.next()) { |
if (station.getRegional() != null && (type == FetchTrainType.REGIONAL||type == FetchTrainType.BOTH) ) { |
80 |
String code = rs.getString( 1 ); |
DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival); |
81 |
if (! rs.wasNull() ) { |
departureBean.entries.addAll( tempBean.entries ); |
82 |
List<DepartureBean> list = lookupDepartures(code, "FJRN"); |
departureBean.notifications.addAll(tempBean.notifications); |
83 |
departureList.addAll(list); |
} |
84 |
} |
|
85 |
|
if (station.getStrain() != null && (type == FetchTrainType.STOG||type == FetchTrainType.BOTH)) { |
86 |
code = rs.getString(2); |
DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival); |
87 |
if (! rs.wasNull() ) { |
departureBean.entries.addAll( tempBean.entries ); |
88 |
List<DepartureBean> list = lookupDepartures(code, "S2"); |
departureBean.notifications.addAll(tempBean.notifications); |
89 |
departureList.addAll(list); |
} |
90 |
} |
|
91 |
Collections.sort( departureList ); |
if (departureBean.entries.size() == 0) { |
92 |
|
logger.info("No departures found for station " + stationID); |
93 |
} |
} |
94 |
|
|
95 |
} finally { |
if (type == FetchTrainType.BOTH) { //if we have both S-tog and regional order by departure/arrival time |
96 |
if (conn != null && !conn.isClosed() ) { |
Collections.sort( departureBean.entries ); |
|
conn.close(); |
|
|
} |
|
97 |
} |
} |
98 |
|
|
99 |
|
|
100 |
return departureList; |
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 lookupDeparturesWwwSite(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 |
|
|
233 |
|
|
234 |
|
public static String cleanText(String input) { |
235 |
|
//apparently JSoup translates characters on www.bane.dk to 0xA0 |
236 |
|
return input.replace((char) 0xA0, (char)0x20).trim(); |
237 |
} |
} |
238 |
|
|
239 |
public List<DepartureBean> lookupDepartures(String stationcode, String type) throws Exception { |
public DepartureBean lookupDeparturesWwwSite(String stationcode, TrainType trainType, boolean arrival) throws Exception { |
240 |
|
|
241 |
|
DepartureBean departureBean = new DepartureBean(); |
242 |
|
|
243 |
List<DepartureBean> departureList = new ArrayList<DepartureBean>(); |
String type = getTypeStringWww(trainType); |
244 |
|
|
245 |
final WebClient webClient = new WebClient(); |
stationcode = URLEncoder.encode(stationcode, "ISO-8859-1"); |
246 |
webClient.setTimeout(2500); |
|
247 |
webClient.setJavaScriptEnabled(false); |
|
248 |
|
String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode; |
249 |
|
logger.fine("URI:" + uri); |
250 |
|
|
251 |
final HtmlPage page = webClient.getPage("http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode); |
|
252 |
|
JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), settings.getReplyTimeout() ); |
253 |
|
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
254 |
|
|
255 |
HtmlElement table = page.getElementById("afgangtabel"); |
Element page = (Element) breaker.invoke(wrapper); |
256 |
|
|
257 |
|
String tableName = arrival == false ? "afgangtabel" : "ankomsttabel"; |
258 |
|
Element table = page.getElementById(tableName); |
259 |
|
|
260 |
|
|
261 |
|
|
262 |
if (table != null) { |
if (table != null) { |
263 |
DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr"); |
Elements tableRows = table.getElementsByTag("tr"); |
264 |
|
|
265 |
|
//boolean passedTidsstreg = false; |
266 |
|
//boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0); |
267 |
|
|
268 |
for (HtmlElement currentRow : tableRows) { |
for (Element currentRow : tableRows) { |
269 |
String rowClass = currentRow.getAttribute("class"); |
String rowClass = currentRow.attr("class"); |
270 |
|
/* |
271 |
|
if (tidsstregExists == true && passedTidsstreg == false) { |
272 |
|
if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) { |
273 |
|
passedTidsstreg = true; |
274 |
|
} else { |
275 |
|
continue; |
276 |
|
} |
277 |
|
}*/ |
278 |
|
|
279 |
|
|
280 |
if (rowClass != null && rowClass.toLowerCase().contains("station") ) { |
if (rowClass != null && rowClass.toLowerCase().contains("station") ) { |
281 |
DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td"); |
Elements fields = currentRow.getElementsByTag("td"); |
282 |
|
|
283 |
DepartureBean departure = new DepartureBean(); |
DepartureEntry departure = new DepartureEntry(); |
284 |
|
|
285 |
String time = fields.get(0).asText(); |
|
286 |
|
|
287 |
|
String time = cleanText( fields.get(0).getAllElements().get(2).text() ); |
288 |
if (time.equals("")) |
if (time.equals("")) |
289 |
time = "0:00"; //Bane.dk bug work-around |
time = "0:00"; //Bane.dk bug work-around |
290 |
departure.setTime(time); |
departure.setTime(time); |
292 |
int updated = extractUpdated( fields.get(1) ); |
int updated = extractUpdated( fields.get(1) ); |
293 |
departure.setUpdated(updated); |
departure.setUpdated(updated); |
294 |
|
|
295 |
String trainNumber = fields.get(2).asText(); |
String trainNumber = cleanText( fields.get(2).text() ); |
296 |
if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber |
if (type.equalsIgnoreCase("S2")) //If it is S-train we need to extract the trainNumber |
297 |
trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2)); |
trainNumber = trainNumber + " " + extractTrainNumberWww(fields.get(2)); |
298 |
departure.setTrainNumber(trainNumber); |
departure.setTrainNumber(trainNumber); |
299 |
|
|
300 |
String destination = fields.get(3).asText(); |
String destination = cleanText( fields.get(3).text() ); |
301 |
departure.setDestination(destination); |
departure.setDestination(destination); |
302 |
|
|
303 |
String origin = fields.get(4).asText(); |
String origin = cleanText( fields.get(4).text() ); |
304 |
departure.setOrigin(origin); |
departure.setOrigin(origin); |
305 |
|
|
306 |
String location = fields.get(5).asText(); |
String location = cleanText( fields.get(5).text() ); |
307 |
departure.setLocation(location); |
departure.setLocation(location); |
308 |
|
|
309 |
String status = fields.get(6).asText(); |
String status = cleanText( fields.get(6).text() ); |
310 |
departure.setStatus(status); |
departure.setStatus(status); |
311 |
|
|
312 |
String note = extractNote( fields.get(7) ); |
String note = cleanText( extractNote( fields.get(7) ) ); |
313 |
departure.setNote(note); |
departure.setNote(note); |
314 |
|
|
315 |
departureList.add(departure); |
departure.setType(type); |
316 |
|
|
317 |
|
departureBean.entries.add(departure); |
318 |
|
|
319 |
|
|
320 |
} |
} |
321 |
} |
} |
322 |
} else { |
} else { |
323 |
logger.warning("No departures found for station=" + stationcode + ", type=" + type); |
logger.warning("No departures found for station=" + stationcode + ", type=" + type); |
324 |
} |
} |
325 |
|
|
326 |
return departureList; |
|
327 |
|
return departureBean; |
328 |
} |
} |
329 |
|
|
330 |
|
|
331 |
private int extractUpdated(HtmlElement updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" |
private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" |
332 |
int updated = -1; |
int updated = -1; |
333 |
|
|
334 |
DomNodeList<HtmlElement> updatedImgs = updatedTd.getElementsByTagName("img"); |
Elements updatedImgs = updatedTd.getElementsByTag("img"); |
335 |
String updatedStr = updatedImgs.get(0).getAttribute("src"); |
String updatedStr = updatedImgs.get(0).attr("src"); |
336 |
|
|
337 |
if (updatedStr != null) { |
if (updatedStr != null) { |
338 |
for (int i=0; i<updatedStr.length(); i++) { |
for (int i=0; i<updatedStr.length(); i++) { |
346 |
return updated; |
return updated; |
347 |
} |
} |
348 |
|
|
349 |
private String extractNote(HtmlElement noteTd) { |
private String extractNote(Element noteTd) { |
350 |
String note = noteTd.asText().trim(); |
String note = noteTd.text().trim(); |
351 |
|
|
352 |
|
|
353 |
List<HtmlElement> elems = noteTd.getElementsByAttribute("span", "class", "bemtype"); |
Elements elems = noteTd.getElementsByClass("bemtype"); |
354 |
if (elems.size() > 0 && note.charAt(note.length()-1) == 'i') |
if (elems.size() > 0 && note.charAt(note.length()-1) == 'i') |
355 |
note = note.substring(0,note.length() -1 ); |
note = note.substring(0,note.length() -1 ); |
356 |
|
|
357 |
return note; |
return note.trim(); |
358 |
|
} |
359 |
|
|
360 |
|
private String extractTrainNumberAzure(Element trainTd) { |
361 |
|
Element anchorElement = trainTd.getElementsByTag("a").get(0); |
362 |
|
String href = anchorElement.attr("href"); |
363 |
|
|
364 |
|
int pos = href.lastIndexOf('/'); |
365 |
|
String number = href.substring(pos+1); |
366 |
|
|
367 |
|
return number; |
368 |
} |
} |
369 |
|
|
370 |
private String extractTrainNumber(HtmlElement trainTd) { |
private String extractTrainNumberWww(Element trainTd) { |
371 |
String number = ""; |
String number = ""; |
372 |
HtmlElement anchorElement = trainTd.getElementsByTagName("a").get(0); |
Element anchorElement = trainTd.getElementsByTag("a").get(0); |
373 |
String href = anchorElement.getAttribute("href"); |
String href = anchorElement.attr("href"); |
374 |
String argstring = href.substring( href.indexOf('?') + 1); |
String argstring = href.substring( href.indexOf('?') + 1); |
375 |
|
|
376 |
String args[] = argstring.split("&"); |
String args[] = argstring.split("&"); |
382 |
} |
} |
383 |
|
|
384 |
|
|
|
|
|
385 |
return number; |
return number; |
386 |
} |
} |
387 |
|
|
388 |
|
|
389 |
//test |
//test |
390 |
public static void main(String args[]) throws Exception{ |
/* |
391 |
|
public static void main(String args[]) throws Exception { |
392 |
DepartureFetcher f = new DepartureFetcher(); |
DepartureFetcher f = new DepartureFetcher(); |
393 |
List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN"); |
List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN"); |
394 |
for(DepartureBean d : deps) { |
for(DepartureBean d : deps) { |
397 |
} |
} |
398 |
|
|
399 |
System.out.println("--------------------------"); |
System.out.println("--------------------------"); |
400 |
} |
}*/ |
401 |
} |
} |