traininfoservice/banedk/DepartureFetcher.java

package dk.thoerup.traininfoservice.banedk;


import java.util.Collections;
import java.util.List;
import java.util.Map;
import java.util.logging.Logger;

import com.gargoylesoftware.htmlunit.BrowserVersion;
import com.gargoylesoftware.htmlunit.WebClient;
import com.gargoylesoftware.htmlunit.html.DomNodeList;
import com.gargoylesoftware.htmlunit.html.HtmlElement;
import com.gargoylesoftware.htmlunit.html.HtmlPage;

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

public class DepartureFetcher {
        
        enum TrainType{
                STOG,
                REGIONAL
        }
        
        Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
        
        Map<String, DepartureBean> cache;
        
        StationDAO stationDao = new StationDAO();
        
        private boolean useTempSite;
        
        public DepartureFetcher(boolean tempSite, int cacheTimeout) {
                useTempSite = tempSite;
                cache = new TimeoutMap<String,DepartureBean>(cacheTimeout);
        }
        
        
        public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
                final String key = "" + stationID + ":" + arrival;
                
                DepartureBean departureBean = cache.get(key);

                
                if (departureBean == null) {
                        departureBean = lookupDepartures(stationID,arrival);
                        cache.put(key, departureBean);
                } else {
                        Statistics.getInstance().incrementDepartureCacheHits();
                        logger.info("Departure: Cache hit " + key); //remove before production
                }
                return departureBean;
        }
                

        public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
                
                DepartureBean departureBean = new DepartureBean();
                
                StationBean station = stationDao.getById(stationID);
                
                if (station.getRegional() != null) {
                        DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
                        departureBean.departureEntries.addAll( tempBean.departureEntries );
                        departureBean.notifications.addAll(tempBean.notifications);
                }
                
                if (station.getStrain() != null) {
                        DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
                        departureBean.departureEntries.addAll( tempBean.departureEntries );
                        departureBean.notifications.addAll(tempBean.notifications);
                }               
                
                Collections.sort( departureBean.departureEntries );

                
                return departureBean;
        }
        
        public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
                if (useTempSite == false) {
                        return lookupDeparturesNormalSite(stationcode, type, arrival);
                } else {
                        //return lookupDeparturesFromTemporarySite(stationcode, type);
                        //TODO: find out what to to if they ever put a temp site up on trafikinfo.bane.dk 
                        return null;
                }
        }
        
        private String getTypeString(TrainType type) {
                switch (type) {
                case STOG:
                        return "S-Tog";
                case REGIONAL:
                        return "Fjerntog";
                default:
                        return ""; //Can not happen
                }
        }
        
        public DepartureBean lookupDeparturesNormalSite(String stationcode, TrainType type, boolean arrival) throws Exception {
                
                DepartureBean departureBean = new DepartureBean();
                
            final WebClient webClient = new WebClient( BrowserVersion.FIREFOX_3 );
            webClient.setTimeout(2500);
            webClient.setJavaScriptEnabled(false);
            
            
            String typeString = getTypeString(type);
            String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst"; 
                            
            //String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
            String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";

            //logger.info("URI: " + uri);
            HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
            CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
            
            HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
            
            String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
            HtmlElement table = page.getElementById(tableName);
            
            if (table != null) {
                    DomNodeList<HtmlElement> tableRows =  table.getElementsByTagName("tr");
                    
                    for (HtmlElement currentRow : tableRows) {
                        String rowClass = currentRow.getAttribute("class");
                        if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
                                DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
        
                                DepartureEntry departure = new DepartureEntry();
                                
                                String time = fields.get(0).asText();
                                if (time.equals(""))
                                        time = "0:00"; //Bane.dk bug work-around
                                departure.setTime(time);
                                
                                int updated = extractUpdated( fields.get(1) );
                                departure.setUpdated(updated);
                                
                                String trainNumber = fields.get(2).asText();
                                if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber 
                                        trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
                                departure.setTrainNumber(trainNumber);
                                
                                String destination = fields.get(3).asText();
                                departure.setDestination(destination);
                                
                                String origin = fields.get(4).asText();
                                departure.setOrigin(origin);
                                
                                String location = fields.get(5).asText();
                                departure.setLocation(location);
                                
                                String status = fields.get(6).asText().trim();
                                departure.setStatus(status);
                                
                                String note = extractNote( fields.get(7) );
                                departure.setNote(note);
                                
                                departure.setType(typeString);
                                
                                departureBean.departureEntries.add( departure );
                        }
                    }
            } else {
                logger.warning("No departures found for station=" + stationcode + ", type=" + type);
            }
            
            HtmlElement notifDiv = page.getElementById("station_planlagte_text");
            if (notifDiv != null) {

                DomNodeList<HtmlElement> tables = notifDiv.getElementsByTagName("table");
                for (HtmlElement tab : tables) {

                        DomNodeList<HtmlElement> anchors = tab.getElementsByTagName("a");               
                        if (anchors.size() == 2) {
                                departureBean.notifications.add(  anchors.get(1).getTextContent() );
                        }
                }
                
            }
            
            
            webClient.closeAllWindows();
            
            return departureBean;
        }
        
        /*
        @Deprecated
        public List<DepartureBean> lookupDeparturesFromTemporarySite(String stationcode, String type) throws Exception {
                
                List<DepartureBean> departureList = new ArrayList<DepartureBean>();
                
            final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3);
            webClient.setTimeout(2500);
            webClient.setJavaScriptEnabled(false);
            

            String uri = "http://bane.dk/lite/station.asp?w=" + type + "&s=" + stationcode;
            
            HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
            CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
            
            HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
            
            HtmlElement table = page.getElementById("traf_afgang");
            
            if (table != null) {                        
                    DomNodeList<HtmlElement> tableRows =  table.getElementsByTagName("tr");
                    
                    boolean isFirst = true;
                    
                    for (HtmlElement currentRow : tableRows) {
                        if (isFirst == true) { //skip table headers
                                isFirst = false;
                                continue;
                        }
                        
                        DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");

                        DepartureBean departure = new DepartureBean();

                        String time = fields.get(0).asText().trim();

                        if (time.equals(""))
                                time = "0:00"; //Bane.dk bug work-around
                        departure.setTime(time);


                        String trainNumber = fields.get(1).asText();
                        departure.setTrainNumber(trainNumber);

                        String destination = fields.get(2).asText();
                        departure.setDestination(destination);

                        String origin = fields.get(3).asText();
                        departure.setOrigin(origin);

                        String status = fields.get(4).asText();
                        departure.setStatus(status);

                        String note = fields.get(5).asText();
                        departure.setNote(note);

                        departureList.add(departure);
                    }
            } else {
                logger.warning("No departures found for station=" + stationcode + ", type=" + type);
            }
            webClient.closeAllWindows();
            
            
            return departureList;
        }*/

        
        private int extractUpdated(HtmlElement updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" 
                int updated = -1;
                
                DomNodeList<HtmlElement> updatedImgs = updatedTd.getElementsByTagName("img");
                String updatedStr = updatedImgs.get(0).getAttribute("src");
                
                if (updatedStr != null) {
                        for (int i=0; i<updatedStr.length(); i++) {
                                char c = updatedStr.charAt(i);
                                if ( Character.isDigit(c))  {
                                        updated = Character.digit(c, 10);
                                        break;
                                }
                        }
                }               
                return updated;
        }
        
        private String extractNote(HtmlElement noteTd) {
                String note = noteTd.asText().trim();
                
                List<HtmlElement> elems = noteTd.getElementsByAttribute("span", "class", "bemtype");
                if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
                        note = note.substring(0,note.length() -1 );

                return note;
        }
        
        private String extractTrainNumber(HtmlElement trainTd) {
                HtmlElement anchorElement = trainTd.getElementsByTagName("a").get(0);
                String href = anchorElement.getAttribute("href");
                
                int pos = href.lastIndexOf('/');
                String number = href.substring(pos+1);
                
                return number;
        }
        
        //test
        /*
        public static void main(String args[]) throws Exception {
                DepartureFetcher f = new DepartureFetcher();
                List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
                for(DepartureBean d : deps) {
                        System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" + 
                                                d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote()   );
                }
                
                System.out.println("--------------------------");
        }*/
}
1	torben	305	package dk.thoerup.traininfoservice.banedk;
2
3	torben	978
4	torben	307	import java.util.Collections;
5	torben	305	import java.util.List;
6	torben	428	import java.util.Map;
7	torben	348	import java.util.logging.Logger;
8	torben	305
9	torben	591	import com.gargoylesoftware.htmlunit.BrowserVersion;
10	torben	305	import com.gargoylesoftware.htmlunit.WebClient;
11			import com.gargoylesoftware.htmlunit.html.DomNodeList;
12			import com.gargoylesoftware.htmlunit.html.HtmlElement;
13			import com.gargoylesoftware.htmlunit.html.HtmlPage;
14
15	torben	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
16			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
17	torben	588	import dk.thoerup.traininfoservice.StationBean;
18			import dk.thoerup.traininfoservice.StationDAO;
19	torben	711	import dk.thoerup.traininfoservice.Statistics;
20	torben	307
21	torben	305	public class DepartureFetcher {
22	torben	348
23	torben	972	enum TrainType{
24			STOG,
25			REGIONAL
26			}
27
28	torben	348	Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
29	torben	387
30	torben	978	Map<String, DepartureBean> cache;
31	torben	387
32	torben	588	StationDAO stationDao = new StationDAO();
33
34	torben	580	private boolean useTempSite;
35
36	torben	584	public DepartureFetcher(boolean tempSite, int cacheTimeout) {
37	torben	580	useTempSite = tempSite;
38	torben	978	cache = new TimeoutMap<String,DepartureBean>(cacheTimeout);
39	torben	580	}
40
41
42	torben	307
43	torben	387
44	torben	978	public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
45	torben	829	final String key = "" + stationID + ":" + arrival;
46	torben	980
47	torben	978	DepartureBean departureBean = cache.get(key);
48	torben	308
49	torben	387
50	torben	978	if (departureBean == null) {
51			departureBean = lookupDepartures(stationID,arrival);
52			cache.put(key, departureBean);
53	torben	387	} else {
54	torben	711	Statistics.getInstance().incrementDepartureCacheHits();
55	torben	829	logger.info("Departure: Cache hit " + key); //remove before production
56	torben	387	}
57	torben	980	return departureBean;
58	torben	387	}
59
60
61	torben	978	public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
62	torben	307
63	torben	978	DepartureBean departureBean = new DepartureBean();
64
65	torben	588	StationBean station = stationDao.getById(stationID);
66	torben	307
67	torben	588	if (station.getRegional() != null) {
68	torben	978	DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
69			departureBean.departureEntries.addAll( tempBean.departureEntries );
70			departureBean.notifications.addAll(tempBean.notifications);
71	torben	307	}
72
73	torben	588	if (station.getStrain() != null) {
74	torben	978	DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
75			departureBean.departureEntries.addAll( tempBean.departureEntries );
76			departureBean.notifications.addAll(tempBean.notifications);
77	torben	588	}
78
79	torben	978	Collections.sort( departureBean.departureEntries );
80	torben	588
81
82	torben	978	return departureBean;
83	torben	305	}
84
85	torben	978	public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
86	torben	580	if (useTempSite == false) {
87	torben	829	return lookupDeparturesNormalSite(stationcode, type, arrival);
88	torben	580	} else {
89	torben	972	//return lookupDeparturesFromTemporarySite(stationcode, type);
90			//TODO: find out what to to if they ever put a temp site up on trafikinfo.bane.dk
91			return null;
92	torben	580	}
93			}
94
95	torben	972	private String getTypeString(TrainType type) {
96			switch (type) {
97			case STOG:
98			return "S-Tog";
99			case REGIONAL:
100			return "Fjerntog";
101			default:
102			return ""; //Can not happen
103			}
104			}
105
106	torben	978	public DepartureBean lookupDeparturesNormalSite(String stationcode, TrainType type, boolean arrival) throws Exception {
107	torben	305
108	torben	978	DepartureBean departureBean = new DepartureBean();
109	torben	305
110	torben	591	final WebClient webClient = new WebClient( BrowserVersion.FIREFOX_3 );
111	torben	386	webClient.setTimeout(2500);
112	torben	313	webClient.setJavaScriptEnabled(false);
113	torben	970
114	torben	972
115			String typeString = getTypeString(type);
116	torben	970	String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
117	torben	313
118	torben	970	//String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
119	torben	972	String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
120	torben	970
121	torben	971	//logger.info("URI: " + uri);
122	torben	939	HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
123	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
124	torben	305
125	torben	421	HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
126	torben	305
127	torben	829	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
128			HtmlElement table = page.getElementById(tableName);
129
130	torben	342	if (table != null) {
131			DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr");
132
133			for (HtmlElement currentRow : tableRows) {
134			String rowClass = currentRow.getAttribute("class");
135			if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
136			DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
137
138	torben	978	DepartureEntry departure = new DepartureEntry();
139	torben	342
140			String time = fields.get(0).asText();
141	torben	375	if (time.equals(""))
142			time = "0:00"; //Bane.dk bug work-around
143	torben	342	departure.setTime(time);
144
145			int updated = extractUpdated( fields.get(1) );
146			departure.setUpdated(updated);
147
148			String trainNumber = fields.get(2).asText();
149	torben	972	if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
150	torben	349	trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
151	torben	342	departure.setTrainNumber(trainNumber);
152
153			String destination = fields.get(3).asText();
154			departure.setDestination(destination);
155
156			String origin = fields.get(4).asText();
157			departure.setOrigin(origin);
158
159			String location = fields.get(5).asText();
160			departure.setLocation(location);
161
162	torben	860	String status = fields.get(6).asText().trim();
163	torben	342	departure.setStatus(status);
164
165			String note = extractNote( fields.get(7) );
166			departure.setNote(note);
167
168	torben	972	departure.setType(typeString);
169	torben	697
170	torben	978	departureBean.departureEntries.add( departure );
171	torben	342	}
172			}
173	torben	348	} else {
174			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
175	torben	305	}
176	torben	978
177			HtmlElement notifDiv = page.getElementById("station_planlagte_text");
178			if (notifDiv != null) {
179
180			DomNodeList<HtmlElement> tables = notifDiv.getElementsByTagName("table");
181			for (HtmlElement tab : tables) {
182
183			DomNodeList<HtmlElement> anchors = tab.getElementsByTagName("a");
184			if (anchors.size() == 2) {
185			departureBean.notifications.add( anchors.get(1).getTextContent() );
186			}
187			}
188
189			}
190
191
192	torben	591	webClient.closeAllWindows();
193	torben	348
194	torben	978	return departureBean;
195	torben	305	}
196
197	torben	974	/*
198			@Deprecated
199	torben	580	public List<DepartureBean> lookupDeparturesFromTemporarySite(String stationcode, String type) throws Exception {
200
201			List<DepartureBean> departureList = new ArrayList<DepartureBean>();
202
203	torben	591	final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3);
204	torben	580	webClient.setTimeout(2500);
205			webClient.setJavaScriptEnabled(false);
206
207
208			String uri = "http://bane.dk/lite/station.asp?w=" + type + "&s=" + stationcode;
209
210	torben	939	HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
211	torben	580	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
212
213			HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
214
215			HtmlElement table = page.getElementById("traf_afgang");
216
217			if (table != null) {
218			DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr");
219
220			boolean isFirst = true;
221
222			for (HtmlElement currentRow : tableRows) {
223			if (isFirst == true) { //skip table headers
224			isFirst = false;
225			continue;
226			}
227
228			DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
229
230			DepartureBean departure = new DepartureBean();
231
232			String time = fields.get(0).asText().trim();
233	torben	583
234	torben	580	if (time.equals(""))
235			time = "0:00"; //Bane.dk bug work-around
236			departure.setTime(time);
237
238
239			String trainNumber = fields.get(1).asText();
240			departure.setTrainNumber(trainNumber);
241
242			String destination = fields.get(2).asText();
243			departure.setDestination(destination);
244
245			String origin = fields.get(3).asText();
246			departure.setOrigin(origin);
247
248			String status = fields.get(4).asText();
249			departure.setStatus(status);
250
251	torben	582	String note = fields.get(5).asText();
252	torben	580	departure.setNote(note);
253
254			departureList.add(departure);
255			}
256			} else {
257			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
258			}
259	torben	591	webClient.closeAllWindows();
260	torben	580
261	torben	591
262	torben	580	return departureList;
263	torben	974	}*/
264	torben	580
265
266	torben	305	private int extractUpdated(HtmlElement updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
267			int updated = -1;
268
269			DomNodeList<HtmlElement> updatedImgs = updatedTd.getElementsByTagName("img");
270			String updatedStr = updatedImgs.get(0).getAttribute("src");
271
272			if (updatedStr != null) {
273			for (int i=0; i<updatedStr.length(); i++) {
274			char c = updatedStr.charAt(i);
275			if ( Character.isDigit(c)) {
276			updated = Character.digit(c, 10);
277			break;
278			}
279			}
280			}
281			return updated;
282			}
283
284	torben	313	private String extractNote(HtmlElement noteTd) {
285			String note = noteTd.asText().trim();
286
287			List<HtmlElement> elems = noteTd.getElementsByAttribute("span", "class", "bemtype");
288			if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
289			note = note.substring(0,note.length() -1 );
290
291			return note;
292			}
293
294	torben	349	private String extractTrainNumber(HtmlElement trainTd) {
295			HtmlElement anchorElement = trainTd.getElementsByTagName("a").get(0);
296			String href = anchorElement.getAttribute("href");
297
298	torben	973	int pos = href.lastIndexOf('/');
299			String number = href.substring(pos+1);
300	torben	970
301	torben	349	return number;
302			}
303
304	torben	305	//test
305	torben	580	/*
306	torben	451	public static void main(String args[]) throws Exception {
307	torben	305	DepartureFetcher f = new DepartureFetcher();
308	torben	307	List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
309	torben	305	for(DepartureBean d : deps) {
310			System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
311			d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
312			}
313
314			System.out.println("--------------------------");
315	torben	580	}*/
316	torben	305	}