1 |
package dk.thoerup.traininfoservice.banedk; |
2 |
|
3 |
|
4 |
import java.io.IOException; |
5 |
import java.net.URL; |
6 |
import java.sql.SQLException; |
7 |
import java.util.ArrayList; |
8 |
import java.util.List; |
9 |
import java.util.Map; |
10 |
import java.util.logging.Level; |
11 |
import java.util.logging.Logger; |
12 |
|
13 |
import com.gargoylesoftware.htmlunit.BrowserVersion; |
14 |
import com.gargoylesoftware.htmlunit.Page; |
15 |
import com.gargoylesoftware.htmlunit.RefreshHandler; |
16 |
import com.gargoylesoftware.htmlunit.WebClient; |
17 |
import com.gargoylesoftware.htmlunit.html.DomNodeList; |
18 |
import com.gargoylesoftware.htmlunit.html.HtmlElement; |
19 |
import com.gargoylesoftware.htmlunit.html.HtmlPage; |
20 |
|
21 |
import dk.thoerup.circuitbreaker.CircuitBreaker; |
22 |
import dk.thoerup.circuitbreaker.CircuitBreakerManager; |
23 |
import dk.thoerup.traininfoservice.StationDAO; |
24 |
import dk.thoerup.traininfoservice.Statistics; |
25 |
|
26 |
public class TimetableFetcher { |
27 |
|
28 |
class NullRefreshHandler implements RefreshHandler { |
29 |
public void handleRefresh(Page arg0, URL arg1, int arg2) throws IOException { |
30 |
} |
31 |
|
32 |
} |
33 |
|
34 |
Map<String, List<TimetableBean>> cache; |
35 |
Map<String, Integer> stationCache; |
36 |
|
37 |
StationDAO stationDao = new StationDAO(); |
38 |
|
39 |
|
40 |
Logger logger = Logger.getLogger(TimetableFetcher.class.getName()); |
41 |
|
42 |
private boolean useTempSite; |
43 |
|
44 |
public TimetableFetcher(boolean tmpSite, int cacheTimeout) { |
45 |
useTempSite = tmpSite; |
46 |
|
47 |
cache = new TimeoutMap<String,List<TimetableBean>>(cacheTimeout); |
48 |
stationCache = new TimeoutMap<String,Integer>( 3*60*60*1000 ); |
49 |
} |
50 |
|
51 |
|
52 |
List<TimetableBean> cachedLookupTimetable(String trainID, String type) throws Exception { |
53 |
String key = trainID+type; |
54 |
List<TimetableBean> list = null;//cache.get(key); |
55 |
|
56 |
if (list == null) { |
57 |
list = lookupTimetable(trainID,type); |
58 |
cache.put(key, list); |
59 |
} else { |
60 |
Statistics.getInstance().incrementTimetableCacheHits(); |
61 |
logger.info("Timetable: Cache hit " + trainID); |
62 |
} |
63 |
return list; |
64 |
} |
65 |
|
66 |
List<TimetableBean> lookupTimetable(String trainID, String type) throws Exception { |
67 |
if (useTempSite == false ){ |
68 |
return lookupTimetableRealSite(trainID, type); |
69 |
} else { |
70 |
return new ArrayList<TimetableBean>(); // no timetable data on temp site |
71 |
} |
72 |
} |
73 |
|
74 |
int getStationId(String name) { |
75 |
Integer id = stationCache.get(name); |
76 |
|
77 |
if (id == null) { |
78 |
try { |
79 |
id = stationDao.getBySpecificName(name); |
80 |
stationCache.put(name, id); |
81 |
} catch (SQLException e) { |
82 |
logger.log(Level.SEVERE, "getStationId failed", e); |
83 |
id = -1; |
84 |
} |
85 |
} |
86 |
|
87 |
return id; |
88 |
} |
89 |
|
90 |
List<TimetableBean> lookupTimetableRealSite(String trainID, String type) throws Exception { |
91 |
List<TimetableBean> timetableList = new ArrayList<TimetableBean>(); |
92 |
|
93 |
String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276"; |
94 |
|
95 |
|
96 |
final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3); |
97 |
webClient.setTimeout(2500); |
98 |
webClient.setJavaScriptEnabled(false); |
99 |
webClient.setRefreshHandler( new NullRefreshHandler() ); |
100 |
webClient.setCssEnabled(false); |
101 |
|
102 |
|
103 |
BanedkInvocation wrapper = new BanedkInvocation(webClient, url); |
104 |
CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk"); |
105 |
|
106 |
HtmlPage page = (HtmlPage) breaker.invoke(wrapper); |
107 |
|
108 |
|
109 |
boolean currentStation = false; |
110 |
boolean currentStationSaved = false; |
111 |
|
112 |
List<HtmlElement> tables = page.getDocumentElement().getElementsByAttribute("table", "class", "Rute"); |
113 |
if (tables.size() == 1) { |
114 |
HtmlElement timetable = tables.get(0); |
115 |
DomNodeList<HtmlElement> rows = timetable.getElementsByTagName("tr"); |
116 |
|
117 |
for (int i=0; i<rows.size(); i++) { |
118 |
if (i==0) //First row is column headers |
119 |
continue; |
120 |
|
121 |
|
122 |
HtmlElement row = rows.get(i); |
123 |
DomNodeList<HtmlElement> fields = row.getElementsByTagName("td"); |
124 |
|
125 |
if (currentStationSaved == false && fields.get(0).getAttribute("class").equalsIgnoreCase("Tidsstreg")) { |
126 |
currentStation = true; |
127 |
continue; |
128 |
} |
129 |
|
130 |
TimetableBean bean = new TimetableBean(); |
131 |
|
132 |
String station = fields.get(0).asText() ; |
133 |
if (station.equals("København")) |
134 |
station = "København H"; //correct inconsistency in naming |
135 |
|
136 |
bean.setStation( station ); |
137 |
bean.setArrival( fields.get(1).asText() ); |
138 |
bean.setDeparture( fields.get(2).asText() ); |
139 |
|
140 |
if (currentStation == true && currentStationSaved == false ) { |
141 |
bean.setCurrent(currentStation); |
142 |
currentStationSaved = true; |
143 |
} |
144 |
|
145 |
bean.setStationId( getStationId( station )); |
146 |
|
147 |
timetableList.add(bean); |
148 |
|
149 |
} |
150 |
|
151 |
} else { |
152 |
logger.warning("No time table found, trainID=" + trainID + " type=" + type); |
153 |
} |
154 |
webClient.closeAllWindows(); |
155 |
|
156 |
return timetableList; |
157 |
} |
158 |
|
159 |
} |