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);
                list = null; //TODO: DEBUG
                
                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;
        }

        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 = fields.get(0).text() ;
                        if (station.equals("København"))
                                station = "København H"; //correct inconsistency in naming
                        
                        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] ) ;
                if (station.equals("København"))
                        station = "København H"; //correct inconsistency in naming

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