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;*/
                
                return lookupDepartures(stationID,arrival);
        }
                

        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	package dk.thoerup.traininfoservice.banedk;
2
3
4	import java.util.Collections;
5	import java.util.List;
6	import java.util.Map;
7	import java.util.logging.Logger;
8
9	import com.gargoylesoftware.htmlunit.BrowserVersion;
10	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	import dk.thoerup.circuitbreaker.CircuitBreaker;
16	import dk.thoerup.circuitbreaker.CircuitBreakerManager;
17	import dk.thoerup.traininfoservice.StationBean;
18	import dk.thoerup.traininfoservice.StationDAO;
19	import dk.thoerup.traininfoservice.Statistics;
20
21	public class DepartureFetcher {
22
23	enum TrainType{
24	STOG,
25	REGIONAL
26	}
27
28	Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
29
30	Map<String, DepartureBean> cache;
31
32	StationDAO stationDao = new StationDAO();
33
34	private boolean useTempSite;
35
36	public DepartureFetcher(boolean tempSite, int cacheTimeout) {
37	useTempSite = tempSite;
38	cache = new TimeoutMap<String,DepartureBean>(cacheTimeout);
39	}
40
41
42
43
44	public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
45	final String key = "" + stationID + ":" + arrival;
46	/*
47	DepartureBean departureBean = cache.get(key);
48
49
50	if (departureBean == null) {
51	departureBean = lookupDepartures(stationID,arrival);
52	cache.put(key, departureBean);
53	} else {
54	Statistics.getInstance().incrementDepartureCacheHits();
55	logger.info("Departure: Cache hit " + key); //remove before production
56	}
57	return departureBean;*/
58
59	return lookupDepartures(stationID,arrival);
60	}
61
62
63	public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
64
65	DepartureBean departureBean = new DepartureBean();
66
67	StationBean station = stationDao.getById(stationID);
68
69	if (station.getRegional() != null) {
70	DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
71	departureBean.departureEntries.addAll( tempBean.departureEntries );
72	departureBean.notifications.addAll(tempBean.notifications);
73	}
74
75	if (station.getStrain() != null) {
76	DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
77	departureBean.departureEntries.addAll( tempBean.departureEntries );
78	departureBean.notifications.addAll(tempBean.notifications);
79	}
80
81	Collections.sort( departureBean.departureEntries );
82
83
84	return departureBean;
85	}
86
87	public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
88	if (useTempSite == false) {
89	return lookupDeparturesNormalSite(stationcode, type, arrival);
90	} else {
91	//return lookupDeparturesFromTemporarySite(stationcode, type);
92	//TODO: find out what to to if they ever put a temp site up on trafikinfo.bane.dk
93	return null;
94	}
95	}
96
97	private String getTypeString(TrainType type) {
98	switch (type) {
99	case STOG:
100	return "S-Tog";
101	case REGIONAL:
102	return "Fjerntog";
103	default:
104	return ""; //Can not happen
105	}
106	}
107
108	public DepartureBean lookupDeparturesNormalSite(String stationcode, TrainType type, boolean arrival) throws Exception {
109
110	DepartureBean departureBean = new DepartureBean();
111
112	final WebClient webClient = new WebClient( BrowserVersion.FIREFOX_3 );
113	webClient.setTimeout(2500);
114	webClient.setJavaScriptEnabled(false);
115
116
117	String typeString = getTypeString(type);
118	String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
119
120	//String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
121	String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
122
123	//logger.info("URI: " + uri);
124	HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
125	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
126
127	HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
128
129	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
130	HtmlElement table = page.getElementById(tableName);
131
132	if (table != null) {
133	DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr");
134
135	for (HtmlElement currentRow : tableRows) {
136	String rowClass = currentRow.getAttribute("class");
137	if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
138	DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
139
140	DepartureEntry departure = new DepartureEntry();
141
142	String time = fields.get(0).asText();
143	if (time.equals(""))
144	time = "0:00"; //Bane.dk bug work-around
145	departure.setTime(time);
146
147	int updated = extractUpdated( fields.get(1) );
148	departure.setUpdated(updated);
149
150	String trainNumber = fields.get(2).asText();
151	if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
152	trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
153	departure.setTrainNumber(trainNumber);
154
155	String destination = fields.get(3).asText();
156	departure.setDestination(destination);
157
158	String origin = fields.get(4).asText();
159	departure.setOrigin(origin);
160
161	String location = fields.get(5).asText();
162	departure.setLocation(location);
163
164	String status = fields.get(6).asText().trim();
165	departure.setStatus(status);
166
167	String note = extractNote( fields.get(7) );
168	departure.setNote(note);
169
170	departure.setType(typeString);
171
172	departureBean.departureEntries.add( departure );
173	}
174	}
175	} else {
176	logger.warning("No departures found for station=" + stationcode + ", type=" + type);
177	}
178
179	HtmlElement notifDiv = page.getElementById("station_planlagte_text");
180	if (notifDiv != null) {
181
182	DomNodeList<HtmlElement> tables = notifDiv.getElementsByTagName("table");
183	for (HtmlElement tab : tables) {
184
185	DomNodeList<HtmlElement> anchors = tab.getElementsByTagName("a");
186	if (anchors.size() == 2) {
187	departureBean.notifications.add( anchors.get(1).getTextContent() );
188	}
189	}
190
191	}
192
193
194	webClient.closeAllWindows();
195
196	return departureBean;
197	}
198
199	/*
200	@Deprecated
201	public List<DepartureBean> lookupDeparturesFromTemporarySite(String stationcode, String type) throws Exception {
202
203	List<DepartureBean> departureList = new ArrayList<DepartureBean>();
204
205	final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3);
206	webClient.setTimeout(2500);
207	webClient.setJavaScriptEnabled(false);
208
209
210	String uri = "http://bane.dk/lite/station.asp?w=" + type + "&s=" + stationcode;
211
212	HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
213	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
214
215	HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
216
217	HtmlElement table = page.getElementById("traf_afgang");
218
219	if (table != null) {
220	DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr");
221
222	boolean isFirst = true;
223
224	for (HtmlElement currentRow : tableRows) {
225	if (isFirst == true) { //skip table headers
226	isFirst = false;
227	continue;
228	}
229
230	DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
231
232	DepartureBean departure = new DepartureBean();
233
234	String time = fields.get(0).asText().trim();
235
236	if (time.equals(""))
237	time = "0:00"; //Bane.dk bug work-around
238	departure.setTime(time);
239
240
241	String trainNumber = fields.get(1).asText();
242	departure.setTrainNumber(trainNumber);
243
244	String destination = fields.get(2).asText();
245	departure.setDestination(destination);
246
247	String origin = fields.get(3).asText();
248	departure.setOrigin(origin);
249
250	String status = fields.get(4).asText();
251	departure.setStatus(status);
252
253	String note = fields.get(5).asText();
254	departure.setNote(note);
255
256	departureList.add(departure);
257	}
258	} else {
259	logger.warning("No departures found for station=" + stationcode + ", type=" + type);
260	}
261	webClient.closeAllWindows();
262
263
264	return departureList;
265	}*/
266
267
268	private int extractUpdated(HtmlElement updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
269	int updated = -1;
270
271	DomNodeList<HtmlElement> updatedImgs = updatedTd.getElementsByTagName("img");
272	String updatedStr = updatedImgs.get(0).getAttribute("src");
273
274	if (updatedStr != null) {
275	for (int i=0; i<updatedStr.length(); i++) {
276	char c = updatedStr.charAt(i);
277	if ( Character.isDigit(c)) {
278	updated = Character.digit(c, 10);
279	break;
280	}
281	}
282	}
283	return updated;
284	}
285
286	private String extractNote(HtmlElement noteTd) {
287	String note = noteTd.asText().trim();
288
289	List<HtmlElement> elems = noteTd.getElementsByAttribute("span", "class", "bemtype");
290	if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
291	note = note.substring(0,note.length() -1 );
292
293	return note;
294	}
295
296	private String extractTrainNumber(HtmlElement trainTd) {
297	HtmlElement anchorElement = trainTd.getElementsByTagName("a").get(0);
298	String href = anchorElement.getAttribute("href");
299
300	int pos = href.lastIndexOf('/');
301	String number = href.substring(pos+1);
302
303	return number;
304	}
305
306	//test
307	/*
308	public static void main(String args[]) throws Exception {
309	DepartureFetcher f = new DepartureFetcher();
310	List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
311	for(DepartureBean d : deps) {
312	System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
313	d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
314	}
315
316	System.out.println("--------------------------");
317	}*/
318	}