traininfoservice/banedk/DepartureFetcher.java

package dk.thoerup.traininfoservice.banedk;


import java.net.URL;
import java.net.URLEncoder;
import java.util.Collections;
import java.util.Map;
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.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;
        private int replyTimeout;
        
        public DepartureFetcher(boolean tempSite, int cacheTimeout, int replyTimeout) {
                this.replyTimeout = replyTimeout;
                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);
                
                departureBean.stationName = station.getName();
                
                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();
                 
            
            String typeString = getTypeString(type);
            String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst"; 
            
            stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
            //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);           
            JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout);
            CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
            
            Document page = (Document) breaker.invoke(wrapper);
            
            String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
            Element table = page.getElementById(tableName);
            
            if (table != null) {
                    Elements tableRows =  table.getElementsByTag("tr");
                    
                    boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0); 
                    boolean passedTidsstreg = false;
                    
                    for (Element currentRow : tableRows) {
                        String rowClass = currentRow.attr("class");
                        
                        if (tidsstregExists == true && passedTidsstreg == false) {
                                if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
                                        passedTidsstreg = true;
                                } else {
                                        continue;
                                }
                        }
                        
                        if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
                                
                                Elements fields = currentRow.getElementsByTag("td");
        
                                DepartureEntry departure = new DepartureEntry();
                                
                                String time = fields.get(0).text();
                                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).text();
                                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).text();
                                departure.setDestination(destination);
                                
                                String origin = fields.get(4).text();
                                departure.setOrigin(origin);
                                
                                String location = fields.get(5).text();
                                departure.setLocation(location);
                                
                                String status = fields.get(6).text().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);
            }
            
            Element notifDiv = page.getElementById("station_planlagte_text");
            if (notifDiv != null) {

                Elements tables = notifDiv.getElementsByTag("table");
                for (Element tab : tables) {

                        Elements anchors = tab.getElementsByTag("a");           
                        if (anchors.size() == 2) {
                                departureBean.notifications.add(  anchors.get(1).text() );
                        }
                }
                
            }
            
            
            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(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif" 
                int updated = -1;
                
                Elements updatedImgs = updatedTd.getElementsByTag("img");
                String updatedStr = updatedImgs.get(0).attr("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(Element noteTd) {
                String note = noteTd.text().trim();
                
                
                Elements elems = noteTd.getElementsByClass("bemtype");
                if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
                        note = note.substring(0,note.length() -1 );

                return note;
        }
        
        private String extractTrainNumber(Element trainTd) {
                Element anchorElement = trainTd.getElementsByTag("a").get(0);
                String href = anchorElement.attr("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.net.URL;
5	import java.net.URLEncoder;
6	import java.util.Collections;
7	import java.util.Map;
8	import java.util.logging.Logger;
9
10	import org.jsoup.nodes.Document;
11	import org.jsoup.nodes.Element;
12	import org.jsoup.select.Elements;
13
14	import dk.thoerup.circuitbreaker.CircuitBreaker;
15	import dk.thoerup.circuitbreaker.CircuitBreakerManager;
16	import dk.thoerup.traininfoservice.StationBean;
17	import dk.thoerup.traininfoservice.StationDAO;
18	import dk.thoerup.traininfoservice.Statistics;
19
20	public class DepartureFetcher {
21
22	enum TrainType{
23	STOG,
24	REGIONAL
25	}
26
27	Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
28
29	Map<String, DepartureBean> cache;
30
31	StationDAO stationDao = new StationDAO();
32
33	private boolean useTempSite;
34	private int replyTimeout;
35
36	public DepartureFetcher(boolean tempSite, int cacheTimeout, int replyTimeout) {
37	this.replyTimeout = replyTimeout;
38	useTempSite = tempSite;
39	cache = new TimeoutMap<String,DepartureBean>(cacheTimeout);
40	}
41
42
43
44
45	public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
46	final String key = "" + stationID + ":" + arrival;
47
48	DepartureBean departureBean = cache.get(key);
49
50
51	if (departureBean == null) {
52	departureBean = lookupDepartures(stationID,arrival);
53	cache.put(key, departureBean);
54	} else {
55	Statistics.getInstance().incrementDepartureCacheHits();
56	logger.info("Departure: Cache hit " + key); //remove before production
57	}
58	return departureBean;
59	}
60
61
62	public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
63
64	DepartureBean departureBean = new DepartureBean();
65
66	StationBean station = stationDao.getById(stationID);
67
68	departureBean.stationName = station.getName();
69
70	if (station.getRegional() != null) {
71	DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
72	departureBean.departureEntries.addAll( tempBean.departureEntries );
73	departureBean.notifications.addAll(tempBean.notifications);
74	}
75
76	if (station.getStrain() != null) {
77	DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
78	departureBean.departureEntries.addAll( tempBean.departureEntries );
79	departureBean.notifications.addAll(tempBean.notifications);
80	}
81
82	Collections.sort( departureBean.departureEntries );
83
84
85	return departureBean;
86	}
87
88	public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
89	if (useTempSite == false) {
90	return lookupDeparturesNormalSite(stationcode, type, arrival);
91	} else {
92	//return lookupDeparturesFromTemporarySite(stationcode, type);
93	//TODO: find out what to to if they ever put a temp site up on trafikinfo.bane.dk
94	return null;
95	}
96	}
97
98	private String getTypeString(TrainType type) {
99	switch (type) {
100	case STOG:
101	return "S-Tog";
102	case REGIONAL:
103	return "Fjerntog";
104	default:
105	return ""; //Can not happen
106	}
107	}
108
109	public DepartureBean lookupDeparturesNormalSite(String stationcode, TrainType type, boolean arrival) throws Exception {
110
111	DepartureBean departureBean = new DepartureBean();
112
113
114	String typeString = getTypeString(type);
115	String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
116
117	stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
118	//String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
119	String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
120
121
122
123	//logger.info("URI: " + uri);
124	JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), replyTimeout);
125	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
126
127	Document page = (Document) breaker.invoke(wrapper);
128
129	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
130	Element table = page.getElementById(tableName);
131
132	if (table != null) {
133	Elements tableRows = table.getElementsByTag("tr");
134
135	boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
136	boolean passedTidsstreg = false;
137
138	for (Element currentRow : tableRows) {
139	String rowClass = currentRow.attr("class");
140
141	if (tidsstregExists == true && passedTidsstreg == false) {
142	if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
143	passedTidsstreg = true;
144	} else {
145	continue;
146	}
147	}
148
149	if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
150
151	Elements fields = currentRow.getElementsByTag("td");
152
153	DepartureEntry departure = new DepartureEntry();
154
155	String time = fields.get(0).text();
156	if (time.equals(""))
157	time = "0:00"; //Bane.dk bug work-around
158	departure.setTime(time);
159
160	int updated = extractUpdated( fields.get(1) );
161	departure.setUpdated(updated);
162
163	String trainNumber = fields.get(2).text();
164	if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
165	trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
166	departure.setTrainNumber(trainNumber);
167
168	String destination = fields.get(3).text();
169	departure.setDestination(destination);
170
171	String origin = fields.get(4).text();
172	departure.setOrigin(origin);
173
174	String location = fields.get(5).text();
175	departure.setLocation(location);
176
177	String status = fields.get(6).text().trim();
178	departure.setStatus(status);
179
180	String note = extractNote( fields.get(7) );
181	departure.setNote(note);
182
183	departure.setType(typeString);
184
185	departureBean.departureEntries.add( departure );
186	}
187	}
188	} else {
189	logger.warning("No departures found for station=" + stationcode + ", type=" + type);
190	}
191
192	Element notifDiv = page.getElementById("station_planlagte_text");
193	if (notifDiv != null) {
194
195	Elements tables = notifDiv.getElementsByTag("table");
196	for (Element tab : tables) {
197
198	Elements anchors = tab.getElementsByTag("a");
199	if (anchors.size() == 2) {
200	departureBean.notifications.add( anchors.get(1).text() );
201	}
202	}
203
204	}
205
206
207	return departureBean;
208	}
209
210	/*
211	@Deprecated
212	public List<DepartureBean> lookupDeparturesFromTemporarySite(String stationcode, String type) throws Exception {
213
214	List<DepartureBean> departureList = new ArrayList<DepartureBean>();
215
216	final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3);
217	webClient.setTimeout(2500);
218	webClient.setJavaScriptEnabled(false);
219
220
221	String uri = "http://bane.dk/lite/station.asp?w=" + type + "&s=" + stationcode;
222
223	HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
224	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
225
226	HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
227
228	HtmlElement table = page.getElementById("traf_afgang");
229
230	if (table != null) {
231	DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr");
232
233	boolean isFirst = true;
234
235	for (HtmlElement currentRow : tableRows) {
236	if (isFirst == true) { //skip table headers
237	isFirst = false;
238	continue;
239	}
240
241	DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
242
243	DepartureBean departure = new DepartureBean();
244
245	String time = fields.get(0).asText().trim();
246
247	if (time.equals(""))
248	time = "0:00"; //Bane.dk bug work-around
249	departure.setTime(time);
250
251
252	String trainNumber = fields.get(1).asText();
253	departure.setTrainNumber(trainNumber);
254
255	String destination = fields.get(2).asText();
256	departure.setDestination(destination);
257
258	String origin = fields.get(3).asText();
259	departure.setOrigin(origin);
260
261	String status = fields.get(4).asText();
262	departure.setStatus(status);
263
264	String note = fields.get(5).asText();
265	departure.setNote(note);
266
267	departureList.add(departure);
268	}
269	} else {
270	logger.warning("No departures found for station=" + stationcode + ", type=" + type);
271	}
272	webClient.closeAllWindows();
273
274
275	return departureList;
276	}*/
277
278
279	private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
280	int updated = -1;
281
282	Elements updatedImgs = updatedTd.getElementsByTag("img");
283	String updatedStr = updatedImgs.get(0).attr("src");
284
285	if (updatedStr != null) {
286	for (int i=0; i<updatedStr.length(); i++) {
287	char c = updatedStr.charAt(i);
288	if ( Character.isDigit(c)) {
289	updated = Character.digit(c, 10);
290	break;
291	}
292	}
293	}
294	return updated;
295	}
296
297	private String extractNote(Element noteTd) {
298	String note = noteTd.text().trim();
299
300
301	Elements elems = noteTd.getElementsByClass("bemtype");
302	if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
303	note = note.substring(0,note.length() -1 );
304
305	return note;
306	}
307
308	private String extractTrainNumber(Element trainTd) {
309	Element anchorElement = trainTd.getElementsByTag("a").get(0);
310	String href = anchorElement.attr("href");
311
312	int pos = href.lastIndexOf('/');
313	String number = href.substring(pos+1);
314
315	return number;
316	}
317
318	//test
319	/*
320	public static void main(String args[]) throws Exception {
321	DepartureFetcher f = new DepartureFetcher();
322	List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
323	for(DepartureBean d : deps) {
324	System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
325	d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
326	}
327
328	System.out.println("--------------------------");
329	}*/
330	}