traininfoservice/banedk/TimetableFetcher.java

package dk.thoerup.traininfoservice.banedk;


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

import net.sf.jsr107cache.Cache;
import net.sf.jsr107cache.CacheException;
import net.sf.jsr107cache.CacheManager;

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

import com.google.appengine.api.memcache.jsr107cache.GCacheFactory;

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.traininfoservice.StationDAO;
import dk.thoerup.traininfoservice.Statistics;

public class TimetableFetcher {

                
        Cache cache; 
        Cache stationCache;

        StationDAO stationDao = new StationDAO();
        
        
        Logger logger = Logger.getLogger(TimetableFetcher.class.getName());
        
        private boolean useAzureSite;
        private int replyTimeout;
        
        @SuppressWarnings("unchecked")
        public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
                useAzureSite = azureSite;
                this.replyTimeout = replyTimeout;               
                
        Map props = new HashMap();
        props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, cacheTimeout);        
                
                try {
            cache = CacheManager.getInstance().getCacheFactory().createCache(props);            
        } catch (CacheException e) {
                logger.log(Level.WARNING, "error creating cache", e);
        }
        
        props = new HashMap();
        props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, 3*60*60*1000);        
                
                try {
                        stationCache = CacheManager.getInstance().getCacheFactory().createCache(props);            
        } catch (CacheException e) {
                logger.log(Level.WARNING, "error creating cache", e);
        }               
        }
        
        
        TimetableBean cachedLookupTimetable(String trainID, String type) throws Exception {
                String key = trainID+type;
                TimetableBean list = (TimetableBean) 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 (useAzureSite == true ){
                        return lookupTimetableAzureSite(trainID, type);
                        
                } else {
                        return lookupTimetableWwwSite(trainID, type);
                }
        }
        
        int getStationId(String name) {
                Integer id = (Integer) stationCache.get(name);
                
                if (id == null) {

                        id = stationDao.getIdByName(name);
                        stationCache.put(name, id);
                } 

                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) , replyTimeout);
            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 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) , replyTimeout);
            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	1093	import java.util.HashMap;
8	torben	428	import java.util.Map;
9	torben	836	import java.util.logging.Level;
10	torben	350	import java.util.logging.Logger;
11
12	torben	1093	import net.sf.jsr107cache.Cache;
13			import net.sf.jsr107cache.CacheException;
14			import net.sf.jsr107cache.CacheManager;
15
16	torben	992	import org.jsoup.nodes.Document;
17			import org.jsoup.nodes.Element;
18			import org.jsoup.select.Elements;
19	torben	350
20	torben	1093	import com.google.appengine.api.memcache.jsr107cache.GCacheFactory;
21
22	torben	1061	import dk.thoerup.android.traininfo.common.TimetableBean;
23			import dk.thoerup.android.traininfo.common.TimetableEntry;
24	torben	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
25			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
26	torben	836	import dk.thoerup.traininfoservice.StationDAO;
27	torben	711	import dk.thoerup.traininfoservice.Statistics;
28	torben	421
29	torben	350	public class TimetableFetcher {
30	torben	992
31	torben	350
32	torben	1093	Cache cache;
33			Cache stationCache;
34	torben	836
35			StationDAO stationDao = new StationDAO();
36	torben	350
37	torben	387
38	torben	350	Logger logger = Logger.getLogger(TimetableFetcher.class.getName());
39	torben	387
40	torben	1036	private boolean useAzureSite;
41	torben	1026	private int replyTimeout;
42	torben	387
43	torben	1093	@SuppressWarnings("unchecked")
44	torben	1036	public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
45			useAzureSite = azureSite;
46	torben	1093	this.replyTimeout = replyTimeout;
47	torben	584
48	torben	1093	Map props = new HashMap();
49			props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, cacheTimeout);
50
51			try {
52			cache = CacheManager.getInstance().getCacheFactory().createCache(props);
53			} catch (CacheException e) {
54			logger.log(Level.WARNING, "error creating cache", e);
55			}
56
57			props = new HashMap();
58			props.put(GCacheFactory.EXPIRATION_DELTA_MILLIS, 36060*1000);
59
60			try {
61			stationCache = CacheManager.getInstance().getCacheFactory().createCache(props);
62			} catch (CacheException e) {
63			logger.log(Level.WARNING, "error creating cache", e);
64			}
65	torben	581	}
66
67
68	torben	1060	TimetableBean cachedLookupTimetable(String trainID, String type) throws Exception {
69	torben	387	String key = trainID+type;
70	torben	1095	TimetableBean list = (TimetableBean) cache.get(key);
71	torben	387
72			if (list == null) {
73			list = lookupTimetable(trainID,type);
74			cache.put(key, list);
75			} else {
76	torben	711	Statistics.getInstance().incrementTimetableCacheHits();
77	torben	389	logger.info("Timetable: Cache hit " + trainID);
78	torben	387	}
79			return list;
80			}
81	torben	581
82	torben	1060	TimetableBean lookupTimetable(String trainID, String type) throws Exception {
83	torben	1036	if (useAzureSite == true ){
84			return lookupTimetableAzureSite(trainID, type);
85
86	torben	581	} else {
87	torben	1036	return lookupTimetableWwwSite(trainID, type);
88	torben	581	}
89			}
90	torben	836
91			int getStationId(String name) {
92	torben	1095	Integer id = (Integer) stationCache.get(name);
93	torben	836
94			if (id == null) {
95	torben	1105
96			id = stationDao.getIdByName(name);
97			stationCache.put(name, id);
98	torben	836	}
99	torben	350
100	torben	836	return id;
101			}
102
103	torben	1060	TimetableBean lookupTimetableAzureSite(String trainID, String type) throws Exception {
104			TimetableBean timetableBean = new TimetableBean();
105	torben	350
106	torben	1048
107	torben	970	String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID;
108	torben	1048	logger.fine("URL:" + url);
109	torben	350
110	torben	1026	JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout);
111	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
112	torben	350
113	torben	992	Document doc = (Document) breaker.invoke(wrapper);
114	torben	421
115	torben	350
116			boolean currentStation = false;
117			boolean currentStationSaved = false;
118
119	torben	992	Elements tables = doc.getElementsByClass("Rute");
120
121	torben	350	if (tables.size() == 1) {
122	torben	992	Element timetable = tables.get(0);
123			Elements rows = timetable.getElementsByTag("tr");
124	torben	350
125			for (int i=0; i<rows.size(); i++) {
126			if (i==0) //First row is column headers
127			continue;
128
129
130	torben	992	Element row = rows.get(i);
131			Elements fields = row.getElementsByTag("td");
132
133	torben	350
134	torben	992	if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
135	torben	350	currentStation = true;
136			continue;
137			}
138
139	torben	1060	TimetableEntry entry = new TimetableEntry();
140	torben	836
141	torben	992	String station = fields.get(0).text() ;
142	torben	836	if (station.equals("København"))
143			station = "København H"; //correct inconsistency in naming
144
145	torben	1060	entry.setStation( station );
146			entry.setArrival( fields.get(1).text() );
147			entry.setDeparture( fields.get(2).text() );
148	torben	350
149	torben	992	boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst");
150	torben	1060	entry.setCancelled(cancelled);
151	torben	975
152	torben	350	if (currentStation == true && currentStationSaved == false ) {
153	torben	1060	entry.setCurrent(currentStation);
154	torben	350	currentStationSaved = true;
155			}
156
157	torben	1060	entry.setStationId( getStationId( station ));
158	torben	836
159	torben	1060	timetableBean.entries.add(entry);
160	torben	350	}
161
162	torben	977	//TODO: There is an off-by-one error in this cancelled parser thingie
163	torben	975	final String cancelledString = "Aflyst";
164	torben	1060	for (int i=0;i<timetableBean.entries.size(); i++) { //handle cancelled labels
165			final int lastIdx = (timetableBean.entries.size() - 1);
166	torben	975
167	torben	1060	TimetableEntry current = timetableBean.entries.get(i);
168	torben	976	if (current.isCancelled()) {
169			if (i == 0) {
170			current.setDeparture(cancelledString);
171			} else if (i == lastIdx) {
172			current.setArrival(cancelledString);
173			} else if (i>0 && i<lastIdx) {
174	torben	1060	TimetableEntry next = timetableBean.entries.get(i+1);
175			TimetableEntry prev = timetableBean.entries.get(i-1);
176	torben	976
177			if (next.isCancelled())
178			current.setDeparture(cancelledString);
179			if (prev.isCancelled())
180			current.setArrival(cancelledString);
181			}
182	torben	975	}
183			}
184
185	torben	350	} else {
186			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
187			}
188	torben	992
189	torben	350
190	torben	1060	return timetableBean;
191	torben	350	}
192	torben	1036
193	torben	1060	TimetableBean lookupTimetableWwwSite(String trainID, String type) throws Exception {
194			TimetableBean timetableBean = new TimetableBean();
195	torben	1036
196			String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
197	torben	1048	logger.fine("URL:" + url);
198	torben	1036
199
200			JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout);
201			CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
202
203			Document doc = (Document) breaker.invoke(wrapper);
204
205
206			boolean currentStation = false;
207			boolean currentStationSaved = false;
208
209			Elements tables = doc.getElementsByClass("Rute");
210
211			if (tables.size() == 1) {
212			Element timetable = tables.get(0);
213			Elements rows = timetable.getElementsByTag("tr");
214
215			for (int i=0; i<rows.size(); i++) {
216			if (i==0) //First row is column headers
217			continue;
218
219
220			Element row = rows.get(i);
221			Elements fields = row.getElementsByTag("td");
222
223
224			if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
225			currentStation = true;
226			continue;
227			}
228
229	torben	1060	TimetableEntry entry = new TimetableEntry();
230	torben	1036
231	torben	1040	String station = DepartureFetcher.cleanText( fields.get(0).text() ) ;
232	torben	1036	if (station.equals("København"))
233			station = "København H"; //correct inconsistency in naming
234
235	torben	1040	String arrival = DepartureFetcher.cleanText( fields.get(1).text() );
236			String departure = DepartureFetcher.cleanText( fields.get(2).text() );
237
238	torben	1060	entry.setStation( station );
239			entry.setArrival( arrival );
240			entry.setDeparture( departure );
241	torben	1036
242
243			if (currentStation == true && currentStationSaved == false ) {
244	torben	1060	entry.setCurrent(currentStation);
245	torben	1036	currentStationSaved = true;
246			}
247
248	torben	1060	entry.setStationId( getStationId( station ));
249	torben	1036
250	torben	1060	timetableBean.entries.add(entry);
251	torben	1036	}
252
253			} else {
254			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
255			}
256
257
258	torben	1060	return timetableBean;
259	torben	1036	}
260	torben	350
261			}