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; |
7 |
import java.util.ArrayList; |
import java.util.ArrayList; |
8 |
import java.util.List; |
import java.util.List; |
9 |
|
import java.util.Map; |
10 |
|
import java.util.logging.Level; |
11 |
import java.util.logging.Logger; |
import java.util.logging.Logger; |
12 |
|
|
13 |
import com.gargoylesoftware.htmlunit.Page; |
import org.jsoup.nodes.Document; |
14 |
import com.gargoylesoftware.htmlunit.RefreshHandler; |
import org.jsoup.nodes.Element; |
15 |
import com.gargoylesoftware.htmlunit.WebClient; |
import org.jsoup.select.Elements; |
16 |
import com.gargoylesoftware.htmlunit.html.DomNodeList; |
|
17 |
import com.gargoylesoftware.htmlunit.html.HtmlElement; |
import dk.thoerup.circuitbreaker.CircuitBreaker; |
18 |
import com.gargoylesoftware.htmlunit.html.HtmlPage; |
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
19 |
|
import dk.thoerup.traininfoservice.StationDAO; |
20 |
|
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 |
TimeoutCache<String, List<TimetableBean>> cache = new TimeoutCache<String,List<TimetableBean>>(120 * 1000); |
Map<String, List<TimetableBean>> cache; |
26 |
|
Map<String, Integer> stationCache; |
27 |
|
|
28 |
|
StationDAO stationDao = new StationDAO(); |
29 |
|
|
30 |
|
|
31 |
Logger logger = Logger.getLogger(TimetableFetcher.class.getName()); |
Logger logger = Logger.getLogger(TimetableFetcher.class.getName()); |
32 |
|
|
33 |
|
private boolean useTempSite; |
34 |
|
private int replyTimeout; |
35 |
|
|
36 |
|
public TimetableFetcher(boolean tmpSite, int cacheTimeout, int replyTimeout) { |
37 |
|
useTempSite = tmpSite; |
38 |
|
this.replyTimeout = replyTimeout; |
39 |
|
|
40 |
|
cache = new TimeoutMap<String,List<TimetableBean>>(cacheTimeout); |
41 |
|
stationCache = new TimeoutMap<String,Integer>( 3*60*60*1000 ); |
42 |
|
} |
43 |
|
|
44 |
|
|
45 |
List<TimetableBean> cachedLookupTimetable(String trainID, String type) throws Exception { |
List<TimetableBean> cachedLookupTimetable(String trainID, String type) throws Exception { |
46 |
String key = trainID+type; |
String key = trainID+type; |
50 |
list = lookupTimetable(trainID,type); |
list = lookupTimetable(trainID,type); |
51 |
cache.put(key, list); |
cache.put(key, list); |
52 |
} else { |
} else { |
53 |
|
Statistics.getInstance().incrementTimetableCacheHits(); |
54 |
logger.info("Timetable: Cache hit " + trainID); |
logger.info("Timetable: Cache hit " + trainID); |
55 |
} |
} |
56 |
return list; |
return list; |
57 |
} |
} |
58 |
|
|
59 |
|
List<TimetableBean> lookupTimetable(String trainID, String type) throws Exception { |
60 |
|
if (useTempSite == false ){ |
61 |
|
return lookupTimetableRealSite(trainID, type); |
62 |
|
} else { |
63 |
|
return new ArrayList<TimetableBean>(); // no timetable data on temp site |
64 |
|
} |
65 |
|
} |
66 |
|
|
67 |
|
int getStationId(String name) { |
68 |
|
Integer id = stationCache.get(name); |
69 |
|
|
70 |
|
if (id == null) { |
71 |
|
try { |
72 |
|
id = stationDao.getIdByName(name); |
73 |
|
stationCache.put(name, id); |
74 |
|
} catch (SQLException e) { |
75 |
|
logger.log(Level.SEVERE, "getStationId failed", e); |
76 |
|
id = -1; |
77 |
|
} |
78 |
|
} |
79 |
|
|
80 |
|
return id; |
81 |
|
} |
82 |
|
|
83 |
List<TimetableBean> lookupTimetable(String trainID, String type) throws Exception { |
List<TimetableBean> lookupTimetableRealSite(String trainID, String type) throws Exception { |
84 |
List<TimetableBean> timetableList = new ArrayList<TimetableBean>(); |
List<TimetableBean> timetableList = new ArrayList<TimetableBean>(); |
85 |
|
|
86 |
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"; |
87 |
|
String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID; |
88 |
|
|
|
final WebClient webClient = new WebClient(); |
|
|
webClient.setTimeout(2500); |
|
|
webClient.setJavaScriptEnabled(false); |
|
|
webClient.setRefreshHandler( new NullRefreshHandler() ); |
|
|
webClient.setCssEnabled(false); |
|
89 |
|
|
90 |
|
JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout); |
91 |
|
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
92 |
|
|
93 |
final HtmlPage page = webClient.getPage(url); |
Document doc = (Document) breaker.invoke(wrapper); |
94 |
|
|
95 |
|
|
96 |
boolean currentStation = false; |
boolean currentStation = false; |
97 |
boolean currentStationSaved = false; |
boolean currentStationSaved = false; |
98 |
|
|
99 |
List<HtmlElement> tables = page.getDocumentElement().getElementsByAttribute("table", "class", "Rute"); |
Elements tables = doc.getElementsByClass("Rute"); |
100 |
|
|
101 |
if (tables.size() == 1) { |
if (tables.size() == 1) { |
102 |
HtmlElement timetable = tables.get(0); |
Element timetable = tables.get(0); |
103 |
DomNodeList<HtmlElement> rows = timetable.getElementsByTagName("tr"); |
Elements rows = timetable.getElementsByTag("tr"); |
104 |
|
|
105 |
for (int i=0; i<rows.size(); i++) { |
for (int i=0; i<rows.size(); i++) { |
106 |
if (i==0) //First row is column headers |
if (i==0) //First row is column headers |
107 |
continue; |
continue; |
108 |
|
|
109 |
|
|
110 |
HtmlElement row = rows.get(i); |
Element row = rows.get(i); |
111 |
DomNodeList<HtmlElement> fields = row.getElementsByTagName("td"); |
Elements fields = row.getElementsByTag("td"); |
112 |
|
|
113 |
|
|
114 |
if (currentStationSaved == false && fields.get(0).getAttribute("class").equalsIgnoreCase("Tidsstreg")) { |
if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) { |
115 |
currentStation = true; |
currentStation = true; |
116 |
continue; |
continue; |
117 |
} |
} |
118 |
|
|
119 |
TimetableBean bean = new TimetableBean(); |
TimetableBean bean = new TimetableBean(); |
120 |
bean.setStation( fields.get(0).asText() ); |
|
121 |
bean.setArrival( fields.get(1).asText() ); |
String station = fields.get(0).text() ; |
122 |
bean.setDeparture( fields.get(2).asText() ); |
if (station.equals("København")) |
123 |
|
station = "København H"; //correct inconsistency in naming |
124 |
|
|
125 |
|
bean.setStation( station ); |
126 |
|
bean.setArrival( fields.get(1).text() ); |
127 |
|
bean.setDeparture( fields.get(2).text() ); |
128 |
|
|
129 |
|
boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst"); |
130 |
|
bean.setCancelled(cancelled); |
131 |
|
|
132 |
if (currentStation == true && currentStationSaved == false ) { |
if (currentStation == true && currentStationSaved == false ) { |
133 |
bean.setCurrent(currentStation); |
bean.setCurrent(currentStation); |
134 |
currentStationSaved = true; |
currentStationSaved = true; |
135 |
} |
} |
136 |
|
|
137 |
|
bean.setStationId( getStationId( station )); |
138 |
|
|
139 |
timetableList.add(bean); |
timetableList.add(bean); |
140 |
|
} |
141 |
|
|
142 |
|
//TODO: There is an off-by-one error in this cancelled parser thingie |
143 |
|
final String cancelledString = "Aflyst"; |
144 |
|
for (int i=0;i<timetableList.size(); i++) { //handle cancelled labels |
145 |
|
final int lastIdx = (timetableList.size() - 1); |
146 |
|
|
147 |
|
TimetableBean current = timetableList.get(i); |
148 |
|
if (current.isCancelled()) { |
149 |
|
if (i == 0) { |
150 |
|
current.setDeparture(cancelledString); |
151 |
|
} else if (i == lastIdx) { |
152 |
|
current.setArrival(cancelledString); |
153 |
|
} else if (i>0 && i<lastIdx) { |
154 |
|
TimetableBean next = timetableList.get(i+1); |
155 |
|
TimetableBean prev = timetableList.get(i-1); |
156 |
|
|
157 |
|
if (next.isCancelled()) |
158 |
|
current.setDeparture(cancelledString); |
159 |
|
if (prev.isCancelled()) |
160 |
|
current.setArrival(cancelledString); |
161 |
|
} |
162 |
|
} |
163 |
} |
} |
164 |
|
|
165 |
} else { |
} else { |
166 |
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
167 |
} |
} |
168 |
|
|
169 |
|
|
170 |
return timetableList; |
return timetableList; |
171 |
} |
} |