1 |
package dk.thoerup.traininfoservice.banedk; |
package dk.thoerup.traininfoservice.banedk; |
2 |
|
|
3 |
|
|
4 |
import java.io.IOException; |
|
5 |
import java.net.URL; |
import java.net.URL; |
6 |
import java.sql.SQLException; |
import java.sql.SQLException; |
7 |
import java.util.ArrayList; |
import java.util.HashMap; |
|
import java.util.List; |
|
8 |
import java.util.Map; |
import java.util.Map; |
9 |
import java.util.logging.Level; |
import java.util.logging.Level; |
10 |
import java.util.logging.Logger; |
import java.util.logging.Logger; |
11 |
|
|
12 |
import com.gargoylesoftware.htmlunit.BrowserVersion; |
import net.sf.jsr107cache.Cache; |
13 |
import com.gargoylesoftware.htmlunit.Page; |
import net.sf.jsr107cache.CacheException; |
14 |
import com.gargoylesoftware.htmlunit.RefreshHandler; |
import net.sf.jsr107cache.CacheManager; |
15 |
import com.gargoylesoftware.htmlunit.WebClient; |
|
16 |
import com.gargoylesoftware.htmlunit.html.DomNodeList; |
import org.jsoup.nodes.Document; |
17 |
import com.gargoylesoftware.htmlunit.html.HtmlElement; |
import org.jsoup.nodes.Element; |
18 |
import com.gargoylesoftware.htmlunit.html.HtmlPage; |
import org.jsoup.select.Elements; |
19 |
|
|
20 |
|
import com.google.appengine.api.memcache.jsr107cache.GCacheFactory; |
21 |
|
|
22 |
|
import dk.thoerup.android.traininfo.common.TimetableBean; |
23 |
|
import dk.thoerup.android.traininfo.common.TimetableEntry; |
24 |
import dk.thoerup.circuitbreaker.CircuitBreaker; |
import dk.thoerup.circuitbreaker.CircuitBreaker; |
25 |
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
26 |
import dk.thoerup.traininfoservice.StationDAO; |
import dk.thoerup.traininfoservice.StationDAO; |
27 |
import dk.thoerup.traininfoservice.Statistics; |
import dk.thoerup.traininfoservice.Statistics; |
28 |
|
|
29 |
public class TimetableFetcher { |
public class TimetableFetcher { |
30 |
|
|
|
class NullRefreshHandler implements RefreshHandler { |
|
|
public void handleRefresh(Page arg0, URL arg1, int arg2) throws IOException { |
|
|
} |
|
|
|
|
|
} |
|
31 |
|
|
32 |
Map<String, List<TimetableBean>> cache; |
Cache cache; |
33 |
Map<String, Integer> stationCache; |
Cache stationCache; |
34 |
|
|
35 |
StationDAO stationDao = new StationDAO(); |
StationDAO stationDao = new StationDAO(); |
36 |
|
|
37 |
|
|
38 |
Logger logger = Logger.getLogger(TimetableFetcher.class.getName()); |
Logger logger = Logger.getLogger(TimetableFetcher.class.getName()); |
39 |
|
|
40 |
private boolean useTempSite; |
private boolean useAzureSite; |
41 |
|
private int replyTimeout; |
42 |
|
|
43 |
public TimetableFetcher(boolean tmpSite, int cacheTimeout) { |
@SuppressWarnings("unchecked") |
44 |
useTempSite = tmpSite; |
public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) { |
45 |
|
useAzureSite = azureSite; |
46 |
|
this.replyTimeout = replyTimeout; |
47 |
|
|
48 |
cache = new TimeoutMap<String,List<TimetableBean>>(cacheTimeout); |
Map props = new HashMap(); |
49 |
stationCache = new TimeoutMap<String,Integer>( 3*60*60*1000 ); |
props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, cacheTimeout); |
50 |
|
|
51 |
|
try { |
52 |
|
cache = CacheManager.getInstance().getCacheFactory().createCache(props); |
53 |
|
} catch (CacheException e) { |
54 |
|
logger.log(Level.WARNING, "error creating cache", e); |
55 |
|
} |
56 |
|
|
57 |
|
props = new HashMap(); |
58 |
|
props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, 3*60*60*1000); |
59 |
|
|
60 |
|
try { |
61 |
|
stationCache = CacheManager.getInstance().getCacheFactory().createCache(props); |
62 |
|
} catch (CacheException e) { |
63 |
|
logger.log(Level.WARNING, "error creating cache", e); |
64 |
|
} |
65 |
} |
} |
66 |
|
|
67 |
|
|
68 |
List<TimetableBean> cachedLookupTimetable(String trainID, String type) throws Exception { |
TimetableBean cachedLookupTimetable(String trainID, String type) throws Exception { |
69 |
String key = trainID+type; |
String key = trainID+type; |
70 |
List<TimetableBean> list = cache.get(key); |
TimetableBean list = cache.get(key); |
71 |
|
|
72 |
if (list == null) { |
if (list == null) { |
73 |
list = lookupTimetable(trainID,type); |
list = lookupTimetable(trainID,type); |
79 |
return list; |
return list; |
80 |
} |
} |
81 |
|
|
82 |
List<TimetableBean> lookupTimetable(String trainID, String type) throws Exception { |
TimetableBean lookupTimetable(String trainID, String type) throws Exception { |
83 |
if (useTempSite == false ){ |
if (useAzureSite == true ){ |
84 |
return lookupTimetableRealSite(trainID, type); |
return lookupTimetableAzureSite(trainID, type); |
85 |
|
|
86 |
} else { |
} else { |
87 |
return new ArrayList<TimetableBean>(); // no timetable data on temp site |
return lookupTimetableWwwSite(trainID, type); |
88 |
} |
} |
89 |
} |
} |
90 |
|
|
104 |
return id; |
return id; |
105 |
} |
} |
106 |
|
|
107 |
List<TimetableBean> lookupTimetableRealSite(String trainID, String type) throws Exception { |
TimetableBean lookupTimetableAzureSite(String trainID, String type) throws Exception { |
108 |
List<TimetableBean> timetableList = new ArrayList<TimetableBean>(); |
TimetableBean timetableBean = new TimetableBean(); |
109 |
|
|
|
String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276"; |
|
|
|
|
110 |
|
|
111 |
final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3); |
String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID; |
112 |
webClient.setTimeout(2500); |
logger.fine("URL:" + url); |
|
webClient.setJavaScriptEnabled(false); |
|
|
webClient.setRefreshHandler( new NullRefreshHandler() ); |
|
|
webClient.setCssEnabled(false); |
|
|
|
|
113 |
|
|
114 |
HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, url); |
JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout); |
115 |
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
116 |
|
|
117 |
HtmlPage page = (HtmlPage) breaker.invoke(wrapper); |
Document doc = (Document) breaker.invoke(wrapper); |
118 |
|
|
119 |
|
|
120 |
boolean currentStation = false; |
boolean currentStation = false; |
121 |
boolean currentStationSaved = false; |
boolean currentStationSaved = false; |
122 |
|
|
123 |
List<HtmlElement> tables = page.getDocumentElement().getElementsByAttribute("table", "class", "Rute"); |
Elements tables = doc.getElementsByClass("Rute"); |
124 |
|
|
125 |
if (tables.size() == 1) { |
if (tables.size() == 1) { |
126 |
HtmlElement timetable = tables.get(0); |
Element timetable = tables.get(0); |
127 |
DomNodeList<HtmlElement> rows = timetable.getElementsByTagName("tr"); |
Elements rows = timetable.getElementsByTag("tr"); |
128 |
|
|
129 |
for (int i=0; i<rows.size(); i++) { |
for (int i=0; i<rows.size(); i++) { |
130 |
if (i==0) //First row is column headers |
if (i==0) //First row is column headers |
131 |
continue; |
continue; |
132 |
|
|
133 |
|
|
134 |
HtmlElement row = rows.get(i); |
Element row = rows.get(i); |
135 |
DomNodeList<HtmlElement> fields = row.getElementsByTagName("td"); |
Elements fields = row.getElementsByTag("td"); |
136 |
|
|
137 |
|
|
138 |
if (currentStationSaved == false && fields.get(0).getAttribute("class").equalsIgnoreCase("Tidsstreg")) { |
if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) { |
139 |
currentStation = true; |
currentStation = true; |
140 |
continue; |
continue; |
141 |
} |
} |
142 |
|
|
143 |
TimetableBean bean = new TimetableBean(); |
TimetableEntry entry = new TimetableEntry(); |
144 |
|
|
145 |
String station = fields.get(0).asText() ; |
String station = fields.get(0).text() ; |
146 |
if (station.equals("København")) |
if (station.equals("København")) |
147 |
station = "København H"; //correct inconsistency in naming |
station = "København H"; //correct inconsistency in naming |
148 |
|
|
149 |
bean.setStation( station ); |
entry.setStation( station ); |
150 |
bean.setArrival( fields.get(1).asText() ); |
entry.setArrival( fields.get(1).text() ); |
151 |
bean.setDeparture( fields.get(2).asText() ); |
entry.setDeparture( fields.get(2).text() ); |
152 |
|
|
153 |
|
boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst"); |
154 |
|
entry.setCancelled(cancelled); |
155 |
|
|
156 |
if (currentStation == true && currentStationSaved == false ) { |
if (currentStation == true && currentStationSaved == false ) { |
157 |
bean.setCurrent(currentStation); |
entry.setCurrent(currentStation); |
158 |
currentStationSaved = true; |
currentStationSaved = true; |
159 |
} |
} |
160 |
|
|
161 |
bean.setStationId( getStationId( station )); |
entry.setStationId( getStationId( station )); |
162 |
|
|
163 |
timetableList.add(bean); |
timetableBean.entries.add(entry); |
164 |
|
} |
165 |
|
|
166 |
|
//TODO: There is an off-by-one error in this cancelled parser thingie |
167 |
|
final String cancelledString = "Aflyst"; |
168 |
|
for (int i=0;i<timetableBean.entries.size(); i++) { //handle cancelled labels |
169 |
|
final int lastIdx = (timetableBean.entries.size() - 1); |
170 |
|
|
171 |
|
TimetableEntry current = timetableBean.entries.get(i); |
172 |
|
if (current.isCancelled()) { |
173 |
|
if (i == 0) { |
174 |
|
current.setDeparture(cancelledString); |
175 |
|
} else if (i == lastIdx) { |
176 |
|
current.setArrival(cancelledString); |
177 |
|
} else if (i>0 && i<lastIdx) { |
178 |
|
TimetableEntry next = timetableBean.entries.get(i+1); |
179 |
|
TimetableEntry prev = timetableBean.entries.get(i-1); |
180 |
|
|
181 |
|
if (next.isCancelled()) |
182 |
|
current.setDeparture(cancelledString); |
183 |
|
if (prev.isCancelled()) |
184 |
|
current.setArrival(cancelledString); |
185 |
|
} |
186 |
|
} |
187 |
} |
} |
188 |
|
|
189 |
} else { |
} else { |
190 |
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
191 |
} |
} |
192 |
webClient.closeAllWindows(); |
|
193 |
|
|
194 |
return timetableList; |
return timetableBean; |
195 |
} |
} |
196 |
|
|
197 |
|
TimetableBean lookupTimetableWwwSite(String trainID, String type) throws Exception { |
198 |
|
TimetableBean timetableBean = new TimetableBean(); |
199 |
|
|
200 |
|
String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276"; |
201 |
|
logger.fine("URL:" + url); |
202 |
|
|
203 |
|
|
204 |
|
JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout); |
205 |
|
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
206 |
|
|
207 |
|
Document doc = (Document) breaker.invoke(wrapper); |
208 |
|
|
209 |
|
|
210 |
|
boolean currentStation = false; |
211 |
|
boolean currentStationSaved = false; |
212 |
|
|
213 |
|
Elements tables = doc.getElementsByClass("Rute"); |
214 |
|
|
215 |
|
if (tables.size() == 1) { |
216 |
|
Element timetable = tables.get(0); |
217 |
|
Elements rows = timetable.getElementsByTag("tr"); |
218 |
|
|
219 |
|
for (int i=0; i<rows.size(); i++) { |
220 |
|
if (i==0) //First row is column headers |
221 |
|
continue; |
222 |
|
|
223 |
|
|
224 |
|
Element row = rows.get(i); |
225 |
|
Elements fields = row.getElementsByTag("td"); |
226 |
|
|
227 |
|
|
228 |
|
if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) { |
229 |
|
currentStation = true; |
230 |
|
continue; |
231 |
|
} |
232 |
|
|
233 |
|
TimetableEntry entry = new TimetableEntry(); |
234 |
|
|
235 |
|
String station = DepartureFetcher.cleanText( fields.get(0).text() ) ; |
236 |
|
if (station.equals("København")) |
237 |
|
station = "København H"; //correct inconsistency in naming |
238 |
|
|
239 |
|
String arrival = DepartureFetcher.cleanText( fields.get(1).text() ); |
240 |
|
String departure = DepartureFetcher.cleanText( fields.get(2).text() ); |
241 |
|
|
242 |
|
entry.setStation( station ); |
243 |
|
entry.setArrival( arrival ); |
244 |
|
entry.setDeparture( departure ); |
245 |
|
|
246 |
|
|
247 |
|
if (currentStation == true && currentStationSaved == false ) { |
248 |
|
entry.setCurrent(currentStation); |
249 |
|
currentStationSaved = true; |
250 |
|
} |
251 |
|
|
252 |
|
entry.setStationId( getStationId( station )); |
253 |
|
|
254 |
|
timetableBean.entries.add(entry); |
255 |
|
} |
256 |
|
|
257 |
|
} else { |
258 |
|
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
259 |
|
} |
260 |
|
|
261 |
|
|
262 |
|
return timetableBean; |
263 |
|
} |
264 |
|
|
265 |
} |
} |