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.circuitbreaker.CircuitBreaker;
import dk.thoerup.circuitbreaker.CircuitBreakerManager;
import dk.thoerup.traininfoservice.StationDAO;
import dk.thoerup.traininfoservice.Statistics;

public class TimetableFetcher {

                
        Map<String, 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,TimetableBean>(cacheTimeout);
                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 (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;
        }

        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	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	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
16			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
17	torben	836	import dk.thoerup.traininfoservice.StationDAO;
18	torben	711	import dk.thoerup.traininfoservice.Statistics;
19	torben	421
20	torben	350	public class TimetableFetcher {
21	torben	992
22	torben	350
23	torben	1060	Map<String, TimetableBean> cache;
24	torben	836	Map<String, Integer> stationCache;
25
26			StationDAO stationDao = new StationDAO();
27	torben	350
28	torben	387
29	torben	350	Logger logger = Logger.getLogger(TimetableFetcher.class.getName());
30	torben	387
31	torben	1036	private boolean useAzureSite;
32	torben	1026	private int replyTimeout;
33	torben	387
34	torben	1036	public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
35			useAzureSite = azureSite;
36	torben	1026	this.replyTimeout = replyTimeout;
37	torben	584
38	torben	1060	cache = new TimeoutMap<String,TimetableBean>(cacheTimeout);
39	torben	836	stationCache = new TimeoutMap<String,Integer>( 36060*1000 );
40	torben	581	}
41
42
43	torben	1060	TimetableBean cachedLookupTimetable(String trainID, String type) throws Exception {
44	torben	387	String key = trainID+type;
45	torben	1060	TimetableBean list = cache.get(key);
46	torben	387
47			if (list == null) {
48			list = lookupTimetable(trainID,type);
49			cache.put(key, list);
50			} else {
51	torben	711	Statistics.getInstance().incrementTimetableCacheHits();
52	torben	389	logger.info("Timetable: Cache hit " + trainID);
53	torben	387	}
54			return list;
55			}
56	torben	581
57	torben	1060	TimetableBean lookupTimetable(String trainID, String type) throws Exception {
58	torben	1036	if (useAzureSite == true ){
59			return lookupTimetableAzureSite(trainID, type);
60
61	torben	581	} else {
62	torben	1036	return lookupTimetableWwwSite(trainID, type);
63	torben	581	}
64			}
65	torben	836
66			int getStationId(String name) {
67			Integer id = stationCache.get(name);
68
69			if (id == null) {
70			try {
71	torben	842	id = stationDao.getIdByName(name);
72	torben	836	stationCache.put(name, id);
73			} catch (SQLException e) {
74			logger.log(Level.SEVERE, "getStationId failed", e);
75			id = -1;
76			}
77			}
78	torben	350
79	torben	836	return id;
80			}
81
82	torben	1060	TimetableBean lookupTimetableAzureSite(String trainID, String type) throws Exception {
83			TimetableBean timetableBean = new TimetableBean();
84	torben	350
85	torben	1048
86	torben	970	String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID;
87	torben	1048	logger.fine("URL:" + url);
88	torben	350
89	torben	1026	JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout);
90	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
91	torben	350
92	torben	992	Document doc = (Document) breaker.invoke(wrapper);
93	torben	421
94	torben	350
95			boolean currentStation = false;
96			boolean currentStationSaved = false;
97
98	torben	992	Elements tables = doc.getElementsByClass("Rute");
99
100	torben	350	if (tables.size() == 1) {
101	torben	992	Element timetable = tables.get(0);
102			Elements rows = timetable.getElementsByTag("tr");
103	torben	350
104			for (int i=0; i<rows.size(); i++) {
105			if (i==0) //First row is column headers
106			continue;
107
108
109	torben	992	Element row = rows.get(i);
110			Elements fields = row.getElementsByTag("td");
111
112	torben	350
113	torben	992	if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
114	torben	350	currentStation = true;
115			continue;
116			}
117
118	torben	1060	TimetableEntry entry = new TimetableEntry();
119	torben	836
120	torben	992	String station = fields.get(0).text() ;
121	torben	836	if (station.equals("København"))
122			station = "København H"; //correct inconsistency in naming
123
124	torben	1060	entry.setStation( station );
125			entry.setArrival( fields.get(1).text() );
126			entry.setDeparture( fields.get(2).text() );
127	torben	350
128	torben	992	boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst");
129	torben	1060	entry.setCancelled(cancelled);
130	torben	975
131	torben	350	if (currentStation == true && currentStationSaved == false ) {
132	torben	1060	entry.setCurrent(currentStation);
133	torben	350	currentStationSaved = true;
134			}
135
136	torben	1060	entry.setStationId( getStationId( station ));
137	torben	836
138	torben	1060	timetableBean.entries.add(entry);
139	torben	350	}
140
141	torben	977	//TODO: There is an off-by-one error in this cancelled parser thingie
142	torben	975	final String cancelledString = "Aflyst";
143	torben	1060	for (int i=0;i<timetableBean.entries.size(); i++) { //handle cancelled labels
144			final int lastIdx = (timetableBean.entries.size() - 1);
145	torben	975
146	torben	1060	TimetableEntry current = timetableBean.entries.get(i);
147	torben	976	if (current.isCancelled()) {
148			if (i == 0) {
149			current.setDeparture(cancelledString);
150			} else if (i == lastIdx) {
151			current.setArrival(cancelledString);
152			} else if (i>0 && i<lastIdx) {
153	torben	1060	TimetableEntry next = timetableBean.entries.get(i+1);
154			TimetableEntry prev = timetableBean.entries.get(i-1);
155	torben	976
156			if (next.isCancelled())
157			current.setDeparture(cancelledString);
158			if (prev.isCancelled())
159			current.setArrival(cancelledString);
160			}
161	torben	975	}
162			}
163
164	torben	350	} else {
165			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
166			}
167	torben	992
168	torben	350
169	torben	1060	return timetableBean;
170	torben	350	}
171	torben	1036
172	torben	1060	TimetableBean lookupTimetableWwwSite(String trainID, String type) throws Exception {
173			TimetableBean timetableBean = new TimetableBean();
174	torben	1036
175			String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
176	torben	1048	logger.fine("URL:" + url);
177	torben	1036
178
179			JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout);
180			CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
181
182			Document doc = (Document) breaker.invoke(wrapper);
183
184
185			boolean currentStation = false;
186			boolean currentStationSaved = false;
187
188			Elements tables = doc.getElementsByClass("Rute");
189
190			if (tables.size() == 1) {
191			Element timetable = tables.get(0);
192			Elements rows = timetable.getElementsByTag("tr");
193
194			for (int i=0; i<rows.size(); i++) {
195			if (i==0) //First row is column headers
196			continue;
197
198
199			Element row = rows.get(i);
200			Elements fields = row.getElementsByTag("td");
201
202
203			if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
204			currentStation = true;
205			continue;
206			}
207
208	torben	1060	TimetableEntry entry = new TimetableEntry();
209	torben	1036
210	torben	1040	String station = DepartureFetcher.cleanText( fields.get(0).text() ) ;
211	torben	1036	if (station.equals("København"))
212			station = "København H"; //correct inconsistency in naming
213
214	torben	1040	String arrival = DepartureFetcher.cleanText( fields.get(1).text() );
215			String departure = DepartureFetcher.cleanText( fields.get(2).text() );
216
217	torben	1060	entry.setStation( station );
218			entry.setArrival( arrival );
219			entry.setDeparture( departure );
220	torben	1036
221
222			if (currentStation == true && currentStationSaved == false ) {
223	torben	1060	entry.setCurrent(currentStation);
224	torben	1036	currentStationSaved = true;
225			}
226
227	torben	1060	entry.setStationId( getStationId( station ));
228	torben	1036
229	torben	1060	timetableBean.entries.add(entry);
230	torben	1036	}
231
232			} else {
233			logger.warning("No time table found, trainID=" + trainID + " type=" + type);
234			}
235
236
237	torben	1060	return timetableBean;
238	torben	1036	}
239	torben	350
240			}