traininfoservice/banedk/TimetableFetcher.java

package dk.thoerup.traininfoservice.banedk;


import java.net.URL;
import java.sql.SQLException;
import java.util.ArrayList;
import java.util.List;
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.circuitbreaker.CircuitBreaker;
import dk.thoerup.circuitbreaker.CircuitBreakerManager;
import dk.thoerup.traininfoservice.StationDAO;
import dk.thoerup.traininfoservice.Statistics;

public class TimetableFetcher {

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

        StationDAO stationDao = new StationDAO();
        
        
        Logger logger = Logger.getLogger(TimetableFetcher.class.getName());
        
        private boolean useAzureSite;
        private int replyTimeout;
        
        public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
                useAzureSite = azureSite;
                this.replyTimeout = replyTimeout;
                
                cache = new TimeoutMap<String,List<TimetableBean>>(cacheTimeout);
                stationCache = new TimeoutMap<String,Integer>( 3*60*60*1000 );
        }
        
        
        List<TimetableBean> cachedLookupTimetable(String trainID, String type) throws Exception {
                String key = trainID+type;
                List<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;
        }
        
        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 = 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;
        }

        List<TimetableBean> lookupTimetableAzureSite(String trainID, String type) throws Exception {            
                List<TimetableBean> timetableList = new ArrayList<TimetableBean>();
                
                //String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
                String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID;                 

            
            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;
                        }
                        
                        TimetableBean bean = new TimetableBean();
                        
                        String station = fields.get(0).text() ;
                        if (station.equals("København"))
                                station = "København H"; //correct inconsistency in naming
                        
                        bean.setStation( station );
                        bean.setArrival( fields.get(1).text() );
                        bean.setDeparture( fields.get(2).text() );
                        
                        boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst");
                        bean.setCancelled(cancelled);
                        
                        if (currentStation == true && currentStationSaved == false ) {
                                bean.setCurrent(currentStation);
                                currentStationSaved = true;
                        }
                        
                        bean.setStationId( getStationId( station ));
                        
                        timetableList.add(bean);
                }
                
                //TODO: There is an off-by-one error in this cancelled parser thingie
                final String cancelledString = "Aflyst";
                for (int i=0;i<timetableList.size(); i++) { //handle cancelled labels
                        final int lastIdx = (timetableList.size() - 1);
                        
                        TimetableBean current = timetableList.get(i);
                        if (current.isCancelled()) {
                                if (i == 0) {
                                        current.setDeparture(cancelledString);
                                } else if (i == lastIdx) {
                                        current.setArrival(cancelledString);
                                } else if (i>0 && i<lastIdx) {
                                        TimetableBean next = timetableList.get(i+1);
                                        TimetableBean prev = timetableList.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 timetableList;
        }

        List<TimetableBean> lookupTimetableWwwSite(String trainID, String type) throws Exception {              
                List<TimetableBean> timetableList = new ArrayList<TimetableBean>();
                
                String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";


            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;
                        }
                        
                        TimetableBean bean = new TimetableBean();
                        
                        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() );
                        
                        bean.setStation( station );
                        bean.setArrival( arrival );
                        bean.setDeparture( departure );
                
                        
                        if (currentStation == true && currentStationSaved == false ) {
                                bean.setCurrent(currentStation);
                                currentStationSaved = true;
                        }
                        
                        bean.setStationId( getStationId( station ));
                        
                        timetableList.add(bean);
                }               
                
            } else {
                logger.warning("No time table found, trainID=" + trainID + " type=" + type);
            }
            
                
                return timetableList;
        }       
        
}
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	350	import java.util.ArrayList;
8			import java.util.List;
9	torben	428	import java.util.Map;
10	torben	836	import java.util.logging.Level;
11	torben	350	import java.util.logging.Logger;
12
13	torben	992	import org.jsoup.nodes.Document;
14			import org.jsoup.nodes.Element;
15			import org.jsoup.select.Elements;
16	torben	350
17	torben	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
18			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
19	torben	836	import dk.thoerup.traininfoservice.StationDAO;
20	torben	711	import dk.thoerup.traininfoservice.Statistics;
21	torben	421
22	torben	350	public class TimetableFetcher {
23	torben	992
24	torben	350
25	torben	584	Map<String, List<TimetableBean>> cache;
26	torben	836	Map<String, Integer> stationCache;
27
28			StationDAO stationDao = new StationDAO();
29	torben	350
30	torben	387
31	torben	350	Logger logger = Logger.getLogger(TimetableFetcher.class.getName());
32	torben	387
33	torben	1036	private boolean useAzureSite;
34	torben	1026	private int replyTimeout;
35	torben	387
36	torben	1036	public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
37			useAzureSite = azureSite;
38	torben	1026	this.replyTimeout = replyTimeout;
39	torben	584
40			cache = new TimeoutMap<String,List<TimetableBean>>(cacheTimeout);
41	torben	836	stationCache = new TimeoutMap<String,Integer>( 36060*1000 );
42	torben	581	}
43
44
45	torben	451	List<TimetableBean> cachedLookupTimetable(String trainID, String type) throws Exception {
46	torben	387	String key = trainID+type;
47	torben	837	List<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			List<TimetableBean> lookupTimetable(String trainID, String type) throws Exception {
60	torben	1036	if (useAzureSite == true ){
61			return lookupTimetableAzureSite(trainID, type);
62
63	torben	581	} else {
64	torben	1036	return lookupTimetableWwwSite(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
84	torben	1036	List<TimetableBean> lookupTimetableAzureSite(String trainID, String type) throws Exception {
85	torben	350	List<TimetableBean> timetableList = new ArrayList<TimetableBean>();
86
87	torben	970	//String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
88			String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID;
89	torben	350
90
91	torben	1026	JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout);
92	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
93	torben	350
94	torben	992	Document doc = (Document) breaker.invoke(wrapper);
95	torben	421
96	torben	350
97			boolean currentStation = false;
98			boolean currentStationSaved = false;
99
100	torben	992	Elements tables = doc.getElementsByClass("Rute");
101
102	torben	350	if (tables.size() == 1) {
103	torben	992	Element timetable = tables.get(0);
104			Elements rows = timetable.getElementsByTag("tr");
105	torben	350
106			for (int i=0; i<rows.size(); i++) {
107			if (i==0) //First row is column headers
108			continue;
109
110
111	torben	992	Element row = rows.get(i);
112			Elements fields = row.getElementsByTag("td");
113
114	torben	350
115	torben	992	if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
116	torben	350	currentStation = true;
117			continue;
118			}
119
120			TimetableBean bean = new TimetableBean();
121	torben	836
122	torben	992	String station = fields.get(0).text() ;
123	torben	836	if (station.equals("København"))
124			station = "København H"; //correct inconsistency in naming
125
126			bean.setStation( station );
127	torben	992	bean.setArrival( fields.get(1).text() );
128			bean.setDeparture( fields.get(2).text() );
129	torben	350
130	torben	992	boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst");
131	torben	975	bean.setCancelled(cancelled);
132
133	torben	350	if (currentStation == true && currentStationSaved == false ) {
134			bean.setCurrent(currentStation);
135			currentStationSaved = true;
136			}
137
138	torben	836	bean.setStationId( getStationId( station ));
139
140	torben	350	timetableList.add(bean);
141			}
142
143	torben	977	//TODO: There is an off-by-one error in this cancelled parser thingie
144	torben	975	final String cancelledString = "Aflyst";
145			for (int i=0;i<timetableList.size(); i++) { //handle cancelled labels
146			final int lastIdx = (timetableList.size() - 1);
147
148			TimetableBean current = timetableList.get(i);
149	torben	976	if (current.isCancelled()) {
150			if (i == 0) {
151			current.setDeparture(cancelledString);
152			} else if (i == lastIdx) {
153			current.setArrival(cancelledString);
154			} else if (i>0 && i<lastIdx) {
155			TimetableBean next = timetableList.get(i+1);
156			TimetableBean prev = timetableList.get(i-1);
157
158			if (next.isCancelled())
159			current.setDeparture(cancelledString);
160			if (prev.isCancelled())
161			current.setArrival(cancelledString);
162			}
163	torben	975	}
164			}
165
166	torben	350	} else {
167			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
168			}
169	torben	992
170	torben	350
171			return timetableList;
172			}
173	torben	1036
174			List<TimetableBean> lookupTimetableWwwSite(String trainID, String type) throws Exception {
175			List<TimetableBean> timetableList = new ArrayList<TimetableBean>();
176
177			String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
178
179
180
181			JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout);
182			CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
183
184			Document doc = (Document) breaker.invoke(wrapper);
185
186
187			boolean currentStation = false;
188			boolean currentStationSaved = false;
189
190			Elements tables = doc.getElementsByClass("Rute");
191
192			if (tables.size() == 1) {
193			Element timetable = tables.get(0);
194			Elements rows = timetable.getElementsByTag("tr");
195
196			for (int i=0; i<rows.size(); i++) {
197			if (i==0) //First row is column headers
198			continue;
199
200
201			Element row = rows.get(i);
202			Elements fields = row.getElementsByTag("td");
203
204
205			if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
206			currentStation = true;
207			continue;
208			}
209
210			TimetableBean bean = new TimetableBean();
211
212	torben	1040	String station = DepartureFetcher.cleanText( fields.get(0).text() ) ;
213	torben	1036	if (station.equals("København"))
214			station = "København H"; //correct inconsistency in naming
215
216	torben	1040	String arrival = DepartureFetcher.cleanText( fields.get(1).text() );
217			String departure = DepartureFetcher.cleanText( fields.get(2).text() );
218
219	torben	1036	bean.setStation( station );
220	torben	1040	bean.setArrival( arrival );
221			bean.setDeparture( departure );
222	torben	1036
223
224			if (currentStation == true && currentStationSaved == false ) {
225			bean.setCurrent(currentStation);
226			currentStationSaved = true;
227			}
228
229			bean.setStationId( getStationId( station ));
230
231			timetableList.add(bean);
232			}
233
234			} else {
235			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
236			}
237
238
239			return timetableList;
240			}
241	torben	350
242			}