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	package dk.thoerup.traininfoservice.banedk;
2
3
4
5	import java.net.URL;
6	import java.sql.SQLException;
7	import java.util.Map;
8	import java.util.logging.Level;
9	import java.util.logging.Logger;
10
11	import org.jsoup.nodes.Document;
12	import org.jsoup.nodes.Element;
13	import org.jsoup.select.Elements;
14
15	import dk.thoerup.circuitbreaker.CircuitBreaker;
16	import dk.thoerup.circuitbreaker.CircuitBreakerManager;
17	import dk.thoerup.traininfoservice.StationDAO;
18	import dk.thoerup.traininfoservice.Statistics;
19
20	public class TimetableFetcher {
21
22
23	Map<String, TimetableBean> cache;
24	Map<String, Integer> stationCache;
25
26	StationDAO stationDao = new StationDAO();
27
28
29	Logger logger = Logger.getLogger(TimetableFetcher.class.getName());
30
31	private boolean useAzureSite;
32	private int replyTimeout;
33
34	public TimetableFetcher(boolean azureSite, int cacheTimeout, int replyTimeout) {
35	useAzureSite = azureSite;
36	this.replyTimeout = replyTimeout;
37
38	cache = new TimeoutMap<String,TimetableBean>(cacheTimeout);
39	stationCache = new TimeoutMap<String,Integer>( 36060*1000 );
40	}
41
42
43	TimetableBean cachedLookupTimetable(String trainID, String type) throws Exception {
44	String key = trainID+type;
45	TimetableBean list = cache.get(key);
46
47	if (list == null) {
48	list = lookupTimetable(trainID,type);
49	cache.put(key, list);
50	} else {
51	Statistics.getInstance().incrementTimetableCacheHits();
52	logger.info("Timetable: Cache hit " + trainID);
53	}
54	return list;
55	}
56
57	TimetableBean lookupTimetable(String trainID, String type) throws Exception {
58	if (useAzureSite == true ){
59	return lookupTimetableAzureSite(trainID, type);
60
61	} else {
62	return lookupTimetableWwwSite(trainID, type);
63	}
64	}
65
66	int getStationId(String name) {
67	Integer id = stationCache.get(name);
68
69	if (id == null) {
70	try {
71	id = stationDao.getIdByName(name);
72	stationCache.put(name, id);
73	} catch (SQLException e) {
74	logger.log(Level.SEVERE, "getStationId failed", e);
75	id = -1;
76	}
77	}
78
79	return id;
80	}
81
82	TimetableBean lookupTimetableAzureSite(String trainID, String type) throws Exception {
83	TimetableBean timetableBean = new TimetableBean();
84
85
86	String url = "http://trafikinfo.bane.dk/TrafikInformation/Ruteplan/" + trainID;
87	logger.fine("URL:" + url);
88
89	JsoupInvocation wrapper = new JsoupInvocation( new URL(url) , replyTimeout);
90	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
91
92	Document doc = (Document) breaker.invoke(wrapper);
93
94
95	boolean currentStation = false;
96	boolean currentStationSaved = false;
97
98	Elements tables = doc.getElementsByClass("Rute");
99
100	if (tables.size() == 1) {
101	Element timetable = tables.get(0);
102	Elements rows = timetable.getElementsByTag("tr");
103
104	for (int i=0; i<rows.size(); i++) {
105	if (i==0) //First row is column headers
106	continue;
107
108
109	Element row = rows.get(i);
110	Elements fields = row.getElementsByTag("td");
111
112
113	if (currentStationSaved == false && fields.get(0).attr("class").equalsIgnoreCase("Tidsstreg")) {
114	currentStation = true;
115	continue;
116	}
117
118	TimetableEntry entry = new TimetableEntry();
119
120	String station = fields.get(0).text() ;
121	if (station.equals("København"))
122	station = "København H"; //correct inconsistency in naming
123
124	entry.setStation( station );
125	entry.setArrival( fields.get(1).text() );
126	entry.setDeparture( fields.get(2).text() );
127
128	boolean cancelled = fields.get(3).text().equalsIgnoreCase("aflyst");
129	entry.setCancelled(cancelled);
130
131	if (currentStation == true && currentStationSaved == false ) {
132	entry.setCurrent(currentStation);
133	currentStationSaved = true;
134	}
135
136	entry.setStationId( getStationId( station ));
137
138	timetableBean.entries.add(entry);
139	}
140
141	//TODO: There is an off-by-one error in this cancelled parser thingie
142	final String cancelledString = "Aflyst";
143	for (int i=0;i<timetableBean.entries.size(); i++) { //handle cancelled labels
144	final int lastIdx = (timetableBean.entries.size() - 1);
145
146	TimetableEntry current = timetableBean.entries.get(i);
147	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	TimetableEntry next = timetableBean.entries.get(i+1);
154	TimetableEntry prev = timetableBean.entries.get(i-1);
155
156	if (next.isCancelled())
157	current.setDeparture(cancelledString);
158	if (prev.isCancelled())
159	current.setArrival(cancelledString);
160	}
161	}
162	}
163
164	} else {
165	logger.warning("No time table found, trainID=" + trainID + " type=" + type);
166	}
167
168
169	return timetableBean;
170	}
171
172	TimetableBean lookupTimetableWwwSite(String trainID, String type) throws Exception {
173	TimetableBean timetableBean = new TimetableBean();
174
175	String url = "http://www.bane.dk/visRute.asp?W=" + type + "&TogNr=" + trainID + "&artikelId=4276";
176	logger.fine("URL:" + url);
177
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	TimetableEntry entry = new TimetableEntry();
209
210	String station = DepartureFetcher.cleanText( fields.get(0).text() ) ;
211	if (station.equals("København"))
212	station = "København H"; //correct inconsistency in naming
213
214	String arrival = DepartureFetcher.cleanText( fields.get(1).text() );
215	String departure = DepartureFetcher.cleanText( fields.get(2).text() );
216
217	entry.setStation( station );
218	entry.setArrival( arrival );
219	entry.setDeparture( departure );
220
221
222	if (currentStation == true && currentStationSaved == false ) {
223	entry.setCurrent(currentStation);
224	currentStationSaved = true;
225	}
226
227	entry.setStationId( getStationId( station ));
228
229	timetableBean.entries.add(entry);
230	}
231
232	} else {
233	logger.warning("No time table found, trainID=" + trainID + " type=" + type);
234	}
235
236
237	return timetableBean;
238	}
239
240	}