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.ArrayList; |
10 |
import java.util.logging.Level; |
import java.util.logging.Level; |
11 |
import java.util.logging.Logger; |
import java.util.logging.Logger; |
12 |
|
|
13 |
import com.gargoylesoftware.htmlunit.BrowserVersion; |
import org.jsoup.nodes.Document; |
14 |
import com.gargoylesoftware.htmlunit.Page; |
import org.jsoup.nodes.Element; |
15 |
import com.gargoylesoftware.htmlunit.RefreshHandler; |
import org.jsoup.select.Elements; |
|
import com.gargoylesoftware.htmlunit.WebClient; |
|
|
import com.gargoylesoftware.htmlunit.html.DomNodeList; |
|
|
import com.gargoylesoftware.htmlunit.html.HtmlElement; |
|
|
import com.gargoylesoftware.htmlunit.html.HtmlPage; |
|
16 |
|
|
17 |
import dk.thoerup.circuitbreaker.CircuitBreaker; |
import dk.thoerup.circuitbreaker.CircuitBreaker; |
18 |
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
20 |
import dk.thoerup.traininfoservice.Statistics; |
import dk.thoerup.traininfoservice.Statistics; |
21 |
|
|
22 |
public class TimetableFetcher { |
public class TimetableFetcher { |
23 |
|
|
|
class NullRefreshHandler implements RefreshHandler { |
|
|
public void handleRefresh(Page arg0, URL arg1, int arg2) throws IOException { |
|
|
} |
|
|
|
|
|
} |
|
24 |
|
|
25 |
Map<String, List<TimetableBean>> cache; |
Map<String, List<TimetableBean>> cache; |
26 |
Map<String, Integer> stationCache; |
Map<String, Integer> stationCache; |
30 |
|
|
31 |
Logger logger = Logger.getLogger(TimetableFetcher.class.getName()); |
Logger logger = Logger.getLogger(TimetableFetcher.class.getName()); |
32 |
|
|
33 |
private boolean useTempSite; |
private boolean useAzureSite; |
34 |
|
private int replyTimeout; |
35 |
|
|
36 |
public TimetableFetcher(boolean tmpSite, int cacheTimeout) { |
public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) { |
37 |
useTempSite = tmpSite; |
useAzureSite = azureSite; |
38 |
|
this.replyTimeout = replyTimeout; |
39 |
|
|
40 |
cache = new TimeoutMap<String,List<TimetableBean>>(cacheTimeout); |
cache = new TimeoutMap<String,List<TimetableBean>>(cacheTimeout); |
41 |
stationCache = new TimeoutMap<String,Integer>( 3*60*60*1000 ); |
stationCache = new TimeoutMap<String,Integer>( 3*60*60*1000 ); |
57 |
} |
} |
58 |
|
|
59 |
List<TimetableBean> lookupTimetable(String trainID, String type) throws Exception { |
List<TimetableBean> lookupTimetable(String trainID, String type) throws Exception { |
60 |
if (useTempSite == false ){ |
if (useAzureSite == true ){ |
61 |
return lookupTimetableRealSite(trainID, type); |
return lookupTimetableAzureSite(trainID, type); |
62 |
|
|
63 |
} else { |
} else { |
64 |
return new ArrayList<TimetableBean>(); // no timetable data on temp site |
return lookupTimetableWwwSite(trainID, type); |
65 |
} |
} |
66 |
} |
} |
67 |
|
|
81 |
return id; |
return id; |
82 |
} |
} |
83 |
|
|
84 |
List<TimetableBean> lookupTimetableRealSite(String trainID, String type) throws Exception { |
List<TimetableBean> lookupTimetableAzureSite(String trainID, String type) throws Exception { |
85 |
List<TimetableBean> timetableList = new ArrayList<TimetableBean>(); |
List<TimetableBean> timetableList = new ArrayList<TimetableBean>(); |
86 |
|
|
87 |
//String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276"; |
//String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276"; |
88 |
String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID; |
String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID; |
89 |
|
|
|
final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3); |
|
|
webClient.setTimeout(2500); |
|
|
webClient.setJavaScriptEnabled(false); |
|
|
webClient.setRefreshHandler( new NullRefreshHandler() ); |
|
|
webClient.setCssEnabled(false); |
|
90 |
|
|
91 |
|
JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout); |
|
HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, url); |
|
92 |
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
93 |
|
|
94 |
HtmlPage page = (HtmlPage) breaker.invoke(wrapper); |
Document doc = (Document) breaker.invoke(wrapper); |
95 |
|
|
96 |
|
|
97 |
boolean currentStation = false; |
boolean currentStation = false; |
98 |
boolean currentStationSaved = false; |
boolean currentStationSaved = false; |
99 |
|
|
100 |
List<HtmlElement> tables = page.getDocumentElement().getElementsByAttribute("table", "class", "Rute"); |
Elements tables = doc.getElementsByClass("Rute"); |
101 |
|
|
102 |
if (tables.size() == 1) { |
if (tables.size() == 1) { |
103 |
HtmlElement timetable = tables.get(0); |
Element timetable = tables.get(0); |
104 |
DomNodeList<HtmlElement> rows = timetable.getElementsByTagName("tr"); |
Elements rows = timetable.getElementsByTag("tr"); |
105 |
|
|
106 |
for (int i=0; i<rows.size(); i++) { |
for (int i=0; i<rows.size(); i++) { |
107 |
if (i==0) //First row is column headers |
if (i==0) //First row is column headers |
108 |
continue; |
continue; |
109 |
|
|
110 |
|
|
111 |
HtmlElement row = rows.get(i); |
Element row = rows.get(i); |
112 |
DomNodeList<HtmlElement> fields = row.getElementsByTagName("td"); |
Elements fields = row.getElementsByTag("td"); |
113 |
|
|
114 |
|
|
115 |
if (currentStationSaved == false && fields.get(0).getAttribute("class").equalsIgnoreCase("Tidsstreg")) { |
if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) { |
116 |
currentStation = true; |
currentStation = true; |
117 |
continue; |
continue; |
118 |
} |
} |
119 |
|
|
120 |
TimetableBean bean = new TimetableBean(); |
TimetableBean bean = new TimetableBean(); |
121 |
|
|
122 |
String station = fields.get(0).asText() ; |
String station = fields.get(0).text() ; |
123 |
if (station.equals("København")) |
if (station.equals("København")) |
124 |
station = "København H"; //correct inconsistency in naming |
station = "København H"; //correct inconsistency in naming |
125 |
|
|
126 |
bean.setStation( station ); |
bean.setStation( station ); |
127 |
bean.setArrival( fields.get(1).asText() ); |
bean.setArrival( fields.get(1).text() ); |
128 |
bean.setDeparture( fields.get(2).asText() ); |
bean.setDeparture( fields.get(2).text() ); |
129 |
|
|
130 |
boolean cancelled = fields.get(3).asText().equalsIgnoreCase("aflyst"); |
boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst"); |
131 |
bean.setCancelled(cancelled); |
bean.setCancelled(cancelled); |
132 |
|
|
133 |
if (currentStation == true && currentStationSaved == false ) { |
if (currentStation == true && currentStationSaved == false ) { |
140 |
timetableList.add(bean); |
timetableList.add(bean); |
141 |
} |
} |
142 |
|
|
143 |
|
//TODO: There is an off-by-one error in this cancelled parser thingie |
144 |
final String cancelledString = "Aflyst"; |
final String cancelledString = "Aflyst"; |
145 |
for (int i=0;i<timetableList.size(); i++) { //handle cancelled labels |
for (int i=0;i<timetableList.size(); i++) { //handle cancelled labels |
146 |
final int lastIdx = (timetableList.size() - 1); |
final int lastIdx = (timetableList.size() - 1); |
166 |
} else { |
} else { |
167 |
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
168 |
} |
} |
169 |
webClient.closeAllWindows(); |
|
170 |
|
|
171 |
return timetableList; |
return timetableList; |
172 |
} |
} |
173 |
|
|
174 |
|
List<TimetableBean> lookupTimetableWwwSite(String trainID, String type) throws Exception { |
175 |
|
List<TimetableBean> timetableList = new ArrayList<TimetableBean>(); |
176 |
|
|
177 |
|
String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276"; |
178 |
|
|
179 |
|
|
180 |
|
|
181 |
|
JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout); |
182 |
|
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
183 |
|
|
184 |
|
Document doc = (Document) breaker.invoke(wrapper); |
185 |
|
|
186 |
|
|
187 |
|
boolean currentStation = false; |
188 |
|
boolean currentStationSaved = false; |
189 |
|
|
190 |
|
Elements tables = doc.getElementsByClass("Rute"); |
191 |
|
|
192 |
|
if (tables.size() == 1) { |
193 |
|
Element timetable = tables.get(0); |
194 |
|
Elements rows = timetable.getElementsByTag("tr"); |
195 |
|
|
196 |
|
for (int i=0; i<rows.size(); i++) { |
197 |
|
if (i==0) //First row is column headers |
198 |
|
continue; |
199 |
|
|
200 |
|
|
201 |
|
Element row = rows.get(i); |
202 |
|
Elements fields = row.getElementsByTag("td"); |
203 |
|
|
204 |
|
|
205 |
|
if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) { |
206 |
|
currentStation = true; |
207 |
|
continue; |
208 |
|
} |
209 |
|
|
210 |
|
TimetableBean bean = new TimetableBean(); |
211 |
|
|
212 |
|
String station = fields.get(0).text() ; |
213 |
|
if (station.equals("København")) |
214 |
|
station = "København H"; //correct inconsistency in naming |
215 |
|
|
216 |
|
bean.setStation( station ); |
217 |
|
bean.setArrival( fields.get(1).text() ); |
218 |
|
bean.setDeparture( fields.get(2).text() ); |
219 |
|
|
220 |
|
|
221 |
|
if (currentStation == true && currentStationSaved == false ) { |
222 |
|
bean.setCurrent(currentStation); |
223 |
|
currentStationSaved = true; |
224 |
|
} |
225 |
|
|
226 |
|
bean.setStationId( getStationId( station )); |
227 |
|
|
228 |
|
timetableList.add(bean); |
229 |
|
} |
230 |
|
|
231 |
|
} else { |
232 |
|
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
233 |
|
} |
234 |
|
|
235 |
|
|
236 |
|
return timetableList; |
237 |
|
} |
238 |
|
|
239 |
} |
} |