traininfoservice/banedk/TimetableFetcher.java

package dk.thoerup.traininfoservice.banedk;


import java.net.URL;
import java.sql.SQLException;
import java.util.Map;
import java.util.logging.Level;
import java.util.logging.Logger;

import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import dk.thoerup.android.traininfo.common.TimetableBean;
import dk.thoerup.android.traininfo.common.TimetableEntry;
import dk.thoerup.circuitbreaker.CircuitBreaker;
import dk.thoerup.circuitbreaker.CircuitBreakerManager;
import dk.thoerup.genericjavautils.TimeoutMap;
import dk.thoerup.traininfoservice.Statistics;
import dk.thoerup.traininfoservice.TraininfoSettings;
import dk.thoerup.traininfoservice.db.StationDAO;

public class TimetableFetcher {

                
        Map<String, TimetableBean> cache; 
        Map<String, Integer> stationCache;

        StationDAO stationDao = new StationDAO();
        
        
        Logger logger = Logger.getLogger(TimetableFetcher.class.getName());

        TraininfoSettings settings;     
        
        public TimetableFetcher(TraininfoSettings settings) {
                this.settings = settings;
                
                cache = new TimeoutMap<String,TimetableBean>( settings.getCacheTimeout() );
                stationCache = new TimeoutMap<String,Integer>( 3*60*60*1000 );
        }
        
        
        TimetableBean cachedLookupTimetable(String trainID, String type) throws Exception {
                String key = trainID+type;
                TimetableBean list = cache.get(key);
                
                if (list == null) {
                        list = lookupTimetable(trainID,type);
                        cache.put(key, list);
                } else {
                        Statistics.getInstance().incrementTimetableCacheHits();
                        logger.info("Timetable: Cache hit " + trainID); 
                }
                return list;
        }
        
        TimetableBean lookupTimetable(String trainID, String type) throws Exception {
                if (settings.getBackend() == TraininfoSettings.Backend.Azure ){
                        return lookupTimetableAzureSite(trainID, type);
                        
                } else {
                        return lookupTimetableMobileSite(trainID, type);
                }
        }
        
        int getStationId(String name) {
                Integer id = stationCache.get(name);
                
                if (id == null) {
                        try {
                                id = stationDao.getIdByName(name);
                                stationCache.put(name, id);
                        } catch (SQLException e) {
                                logger.log(Level.SEVERE, "getStationId failed", e);
                                id = -1;
                        }
                } 

                return id;
        }
        
        String correctStationName(String name) {
                if (name.equals("København"))
                        name = "København H"; //correct inconsistency in naming
                
                return name;            
        }

        TimetableBean lookupTimetableAzureSite(String trainID, String type) throws Exception {          
                TimetableBean timetableBean = new TimetableBean();
                

                String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID;                 
                logger.fine("URL:" + url);
            
            JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , settings.getReplyTimeout() );
            CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
            
            Document doc = (Document) breaker.invoke(wrapper);
                    
            
            boolean currentStation = false;
            boolean currentStationSaved = false;
            
            Elements tables = doc.getElementsByClass("Rute");
            
            if (tables.size() == 1) {
                Element timetable = tables.get(0);
                Elements rows = timetable.getElementsByTag("tr");
                
                for (int i=0; i<rows.size(); i++) {
                        if (i==0) //First row is column headers
                                continue;
                        
                        
                        Element row = rows.get(i);
                        Elements fields = row.getElementsByTag("td");

                        
                        if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
                                currentStation = true;
                                continue;
                        }
                        
                        TimetableEntry entry = new TimetableEntry();
                        
                        String station = correctStationName( fields.get(0).text() );
                        
                        entry.setStation( station );
                        entry.setArrival( fields.get(1).text() );
                        entry.setDeparture( fields.get(2).text() );
                        
                        boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst");
                        entry.setCancelled(cancelled);
                        
                        if (currentStation == true && currentStationSaved == false ) {
                                entry.setCurrent(currentStation);
                                currentStationSaved = true;
                        }
                        
                        entry.setStationId( getStationId( station ));
                        
                        timetableBean.entries.add(entry);
                }
                
                //TODO: There is an off-by-one error in this cancelled parser thingie
                final String cancelledString = "Aflyst";
                for (int i=0;i<timetableBean.entries.size(); i++) { //handle cancelled labels
                        final int lastIdx = (timetableBean.entries.size() - 1);
                        
                        TimetableEntry current = timetableBean.entries.get(i);
                        if (current.isCancelled()) {
                                if (i == 0) {
                                        current.setDeparture(cancelledString);
                                } else if (i == lastIdx) {
                                        current.setArrival(cancelledString);
                                } else if (i>0 && i<lastIdx) {
                                        TimetableEntry next = timetableBean.entries.get(i+1);
                                        TimetableEntry prev = timetableBean.entries.get(i-1);
                                        
                                        if (next.isCancelled())
                                                current.setDeparture(cancelledString);
                                        if (prev.isCancelled())
                                                current.setArrival(cancelledString);
                                }
                        }
                }
                
            } else {
                logger.warning("No time table found, trainID=" + trainID + " type=" + type);
            }
            
                
                return timetableBean;
        }

        TimetableBean lookupTimetableMobileSite(String trainID, String type) throws Exception {
                TimetableBean timetableBean = new TimetableBean();
                
                String url = "http://mobil.bane.dk/mobilStation.asp?artikelID=5332&tognummer=" + trainID + "&webprofil=" + type + "&mode=rute";
                logger.fine("URL:" + url);

            
            JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , settings.getReplyTimeout() );
            CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
            
            Document doc = (Document) breaker.invoke(wrapper);

            Element content = doc.getElementsByClass("contentDiv").get(1);
            Element dlist = content.child(0);

            
            Elements rows = dlist.getElementsByTag("dt");

            for (int i=0; i<rows.size(); i++) {

                Element row = rows.get(i);
                
                logger.fine( row.text() );
                
                String parts[] = row.text().split(",");
                
                TimetableEntry entry = new TimetableEntry();

                String station = DepartureFetcher.cleanText( parts[0] ) ;
                station = correctStationName(station);


                String arrival = DepartureFetcher.cleanText( parts[1] );
                String departure = DepartureFetcher.cleanText( "" );

                entry.setStation( station );
                entry.setArrival( arrival );
                entry.setDeparture( departure );


                entry.setStationId( getStationId( station ));

                timetableBean.entries.add(entry);
            }           


            return timetableBean;

        }
        
        @Deprecated
        TimetableBean lookupTimetableWwwSite(String trainID, String type) throws Exception {            
                TimetableBean timetableBean = new TimetableBean();
                
                String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
                logger.fine("URL:" + url);

            
            JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , settings.getReplyTimeout() );
            CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
            
            Document doc = (Document) breaker.invoke(wrapper);
                    
            
            boolean currentStation = false;
            boolean currentStationSaved = false;
            
            Elements tables = doc.getElementsByClass("Rute");
            
            if (tables.size() == 1) {
                Element timetable = tables.get(0);
                Elements rows = timetable.getElementsByTag("tr");
                
                for (int i=0; i<rows.size(); i++) {
                        if (i==0) //First row is column headers
                                continue;
                        
                        
                        Element row = rows.get(i);
                        Elements fields = row.getElementsByTag("td");

                        
                        if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
                                currentStation = true;
                                continue;
                        }
                        
                        TimetableEntry entry = new TimetableEntry();
                        
                        String station = DepartureFetcher.cleanText( fields.get(0).text() ) ;
                        station = correctStationName(station);

                        
                        String arrival = DepartureFetcher.cleanText( fields.get(1).text() );
                        String departure = DepartureFetcher.cleanText( fields.get(2).text() );
                        
                        entry.setStation( station );
                        entry.setArrival( arrival );
                        entry.setDeparture( departure );
                
                        
                        if (currentStation == true && currentStationSaved == false ) {
                                entry.setCurrent(currentStation);
                                currentStationSaved = true;
                        }
                        
                        entry.setStationId( getStationId( station ));
                        
                        timetableBean.entries.add(entry);
                }               
                
            } else {
                logger.warning("No time table found, trainID=" + trainID + " type=" + type);
            }
            
                
                return timetableBean;
        }       
        
}
1	torben	350	package dk.thoerup.traininfoservice.banedk;
2
3
4	torben	992
5	torben	350	import java.net.URL;
6	torben	836	import java.sql.SQLException;
7	torben	428	import java.util.Map;
8	torben	836	import java.util.logging.Level;
9	torben	350	import java.util.logging.Logger;
10
11	torben	992	import org.jsoup.nodes.Document;
12			import org.jsoup.nodes.Element;
13			import org.jsoup.select.Elements;
14	torben	350
15	torben	1061	import dk.thoerup.android.traininfo.common.TimetableBean;
16			import dk.thoerup.android.traininfo.common.TimetableEntry;
17	torben	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
18			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
19	torben	1355	import dk.thoerup.genericjavautils.TimeoutMap;
20	torben	711	import dk.thoerup.traininfoservice.Statistics;
21	torben	1303	import dk.thoerup.traininfoservice.TraininfoSettings;
22	torben	1255	import dk.thoerup.traininfoservice.db.StationDAO;
23	torben	421
24	torben	350	public class TimetableFetcher {
25	torben	992
26	torben	350
27	torben	1060	Map<String, TimetableBean> cache;
28	torben	836	Map<String, Integer> stationCache;
29
30			StationDAO stationDao = new StationDAO();
31	torben	350
32	torben	387
33	torben	350	Logger logger = Logger.getLogger(TimetableFetcher.class.getName());
34	torben	1303
35			TraininfoSettings settings;
36	torben	387
37	torben	1303	public TimetableFetcher(TraininfoSettings settings) {
38			this.settings = settings;
39	torben	584
40	torben	1303	cache = new TimeoutMap<String,TimetableBean>( settings.getCacheTimeout() );
41	torben	836	stationCache = new TimeoutMap<String,Integer>( 36060*1000 );
42	torben	581	}
43
44
45	torben	1060	TimetableBean cachedLookupTimetable(String trainID, String type) throws Exception {
46	torben	387	String key = trainID+type;
47	torben	1060	TimetableBean list = cache.get(key);
48	torben	387
49			if (list == null) {
50			list = lookupTimetable(trainID,type);
51			cache.put(key, list);
52			} else {
53	torben	711	Statistics.getInstance().incrementTimetableCacheHits();
54	torben	389	logger.info("Timetable: Cache hit " + trainID);
55	torben	387	}
56			return list;
57			}
58	torben	581
59	torben	1060	TimetableBean lookupTimetable(String trainID, String type) throws Exception {
60	torben	1372	if (settings.getBackend() == TraininfoSettings.Backend.Azure ){
61	torben	1036	return lookupTimetableAzureSite(trainID, type);
62
63	torben	581	} else {
64	torben	1331	return lookupTimetableMobileSite(trainID, type);
65	torben	581	}
66			}
67	torben	836
68			int getStationId(String name) {
69			Integer id = stationCache.get(name);
70
71			if (id == null) {
72			try {
73	torben	842	id = stationDao.getIdByName(name);
74	torben	836	stationCache.put(name, id);
75			} catch (SQLException e) {
76			logger.log(Level.SEVERE, "getStationId failed", e);
77			id = -1;
78			}
79			}
80	torben	350
81	torben	836	return id;
82			}
83	torben	1406
84			String correctStationName(String name) {
85			if (name.equals("København"))
86			name = "København H"; //correct inconsistency in naming
87
88			return name;
89			}
90	torben	836
91	torben	1060	TimetableBean lookupTimetableAzureSite(String trainID, String type) throws Exception {
92			TimetableBean timetableBean = new TimetableBean();
93	torben	350
94	torben	1048
95	torben	970	String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID;
96	torben	1048	logger.fine("URL:" + url);
97	torben	350
98	torben	1303	JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , settings.getReplyTimeout() );
99	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
100	torben	350
101	torben	992	Document doc = (Document) breaker.invoke(wrapper);
102	torben	421
103	torben	350
104			boolean currentStation = false;
105			boolean currentStationSaved = false;
106
107	torben	992	Elements tables = doc.getElementsByClass("Rute");
108
109	torben	350	if (tables.size() == 1) {
110	torben	992	Element timetable = tables.get(0);
111			Elements rows = timetable.getElementsByTag("tr");
112	torben	350
113			for (int i=0; i<rows.size(); i++) {
114			if (i==0) //First row is column headers
115			continue;
116
117
118	torben	992	Element row = rows.get(i);
119			Elements fields = row.getElementsByTag("td");
120
121	torben	350
122	torben	992	if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
123	torben	350	currentStation = true;
124			continue;
125			}
126
127	torben	1060	TimetableEntry entry = new TimetableEntry();
128	torben	836
129	torben	1406	String station = correctStationName( fields.get(0).text() );
130	torben	836
131	torben	1060	entry.setStation( station );
132			entry.setArrival( fields.get(1).text() );
133			entry.setDeparture( fields.get(2).text() );
134	torben	350
135	torben	992	boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst");
136	torben	1060	entry.setCancelled(cancelled);
137	torben	975
138	torben	350	if (currentStation == true && currentStationSaved == false ) {
139	torben	1060	entry.setCurrent(currentStation);
140	torben	350	currentStationSaved = true;
141			}
142
143	torben	1060	entry.setStationId( getStationId( station ));
144	torben	836
145	torben	1060	timetableBean.entries.add(entry);
146	torben	350	}
147
148	torben	977	//TODO: There is an off-by-one error in this cancelled parser thingie
149	torben	975	final String cancelledString = "Aflyst";
150	torben	1060	for (int i=0;i<timetableBean.entries.size(); i++) { //handle cancelled labels
151			final int lastIdx = (timetableBean.entries.size() - 1);
152	torben	975
153	torben	1060	TimetableEntry current = timetableBean.entries.get(i);
154	torben	976	if (current.isCancelled()) {
155			if (i == 0) {
156			current.setDeparture(cancelledString);
157			} else if (i == lastIdx) {
158			current.setArrival(cancelledString);
159			} else if (i>0 && i<lastIdx) {
160	torben	1060	TimetableEntry next = timetableBean.entries.get(i+1);
161			TimetableEntry prev = timetableBean.entries.get(i-1);
162	torben	976
163			if (next.isCancelled())
164			current.setDeparture(cancelledString);
165			if (prev.isCancelled())
166			current.setArrival(cancelledString);
167			}
168	torben	975	}
169			}
170
171	torben	350	} else {
172			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
173			}
174	torben	992
175	torben	350
176	torben	1060	return timetableBean;
177	torben	350	}
178	torben	1036
179	torben	1331	TimetableBean lookupTimetableMobileSite(String trainID, String type) throws Exception {
180	torben	1367	TimetableBean timetableBean = new TimetableBean();
181
182			String url = "http://mobil.bane.dk/mobilStation.asp?artikelID=5332&tognummer=" + trainID + "&webprofil=" + type + "&mode=rute";
183			logger.fine("URL:" + url);
184
185
186			JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , settings.getReplyTimeout() );
187			CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
188
189			Document doc = (Document) breaker.invoke(wrapper);
190
191			Element content = doc.getElementsByClass("contentDiv").get(1);
192			Element dlist = content.child(0);
193
194
195			Elements rows = dlist.getElementsByTag("dt");
196
197			for (int i=0; i<rows.size(); i++) {
198
199			Element row = rows.get(i);
200
201	torben	1368	logger.fine( row.text() );
202	torben	1367
203			String parts[] = row.text().split(",");
204
205			TimetableEntry entry = new TimetableEntry();
206
207			String station = DepartureFetcher.cleanText( parts[0] ) ;
208	torben	1406	station = correctStationName(station);
209	torben	1367
210	torben	1406
211	torben	1367	String arrival = DepartureFetcher.cleanText( parts[1] );
212			String departure = DepartureFetcher.cleanText( "" );
213
214			entry.setStation( station );
215			entry.setArrival( arrival );
216			entry.setDeparture( departure );
217
218
219			entry.setStationId( getStationId( station ));
220
221			timetableBean.entries.add(entry);
222			}
223
224
225			return timetableBean;
226
227	torben	1331	}
228
229			@Deprecated
230	torben	1060	TimetableBean lookupTimetableWwwSite(String trainID, String type) throws Exception {
231			TimetableBean timetableBean = new TimetableBean();
232	torben	1036
233			String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
234	torben	1048	logger.fine("URL:" + url);
235	torben	1036
236
237	torben	1303	JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , settings.getReplyTimeout() );
238	torben	1036	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
239
240			Document doc = (Document) breaker.invoke(wrapper);
241
242
243			boolean currentStation = false;
244			boolean currentStationSaved = false;
245
246			Elements tables = doc.getElementsByClass("Rute");
247
248			if (tables.size() == 1) {
249			Element timetable = tables.get(0);
250			Elements rows = timetable.getElementsByTag("tr");
251
252			for (int i=0; i<rows.size(); i++) {
253			if (i==0) //First row is column headers
254			continue;
255
256
257			Element row = rows.get(i);
258			Elements fields = row.getElementsByTag("td");
259
260
261			if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
262			currentStation = true;
263			continue;
264			}
265
266	torben	1060	TimetableEntry entry = new TimetableEntry();
267	torben	1036
268	torben	1040	String station = DepartureFetcher.cleanText( fields.get(0).text() ) ;
269	torben	1406	station = correctStationName(station);
270
271	torben	1036
272	torben	1040	String arrival = DepartureFetcher.cleanText( fields.get(1).text() );
273			String departure = DepartureFetcher.cleanText( fields.get(2).text() );
274
275	torben	1060	entry.setStation( station );
276			entry.setArrival( arrival );
277			entry.setDeparture( departure );
278	torben	1036
279
280			if (currentStation == true && currentStationSaved == false ) {
281	torben	1060	entry.setCurrent(currentStation);
282	torben	1036	currentStationSaved = true;
283			}
284
285	torben	1060	entry.setStationId( getStationId( station ));
286	torben	1036
287	torben	1060	timetableBean.entries.add(entry);
288	torben	1036	}
289
290			} else {
291			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
292			}
293
294
295	torben	1060	return timetableBean;
296	torben	1036	}
297	torben	350
298			}