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;
        
        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();
                 
            
            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), 2500);
            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	torben	305	package dk.thoerup.traininfoservice.banedk;
2
3	torben	978
4	torben	992	import java.net.URL;
5	torben	994	import java.net.URLEncoder;
6	torben	307	import java.util.Collections;
7	torben	428	import java.util.Map;
8	torben	348	import java.util.logging.Logger;
9	torben	305
10	torben	992	import org.jsoup.nodes.Document;
11			import org.jsoup.nodes.Element;
12			import org.jsoup.select.Elements;
13	torben	305
14	torben	468	import dk.thoerup.circuitbreaker.CircuitBreaker;
15			import dk.thoerup.circuitbreaker.CircuitBreakerManager;
16	torben	588	import dk.thoerup.traininfoservice.StationBean;
17			import dk.thoerup.traininfoservice.StationDAO;
18	torben	711	import dk.thoerup.traininfoservice.Statistics;
19	torben	307
20	torben	305	public class DepartureFetcher {
21	torben	348
22	torben	972	enum TrainType{
23			STOG,
24			REGIONAL
25			}
26
27	torben	348	Logger logger = Logger.getLogger(DepartureFetcher.class.getName());
28	torben	387
29	torben	978	Map<String, DepartureBean> cache;
30	torben	387
31	torben	588	StationDAO stationDao = new StationDAO();
32
33	torben	580	private boolean useTempSite;
34
35	torben	584	public DepartureFetcher(boolean tempSite, int cacheTimeout) {
36	torben	580	useTempSite = tempSite;
37	torben	978	cache = new TimeoutMap<String,DepartureBean>(cacheTimeout);
38	torben	580	}
39
40
41	torben	307
42	torben	387
43	torben	978	public DepartureBean cachedLookupDepartures(int stationID, boolean arrival) throws Exception {
44	torben	829	final String key = "" + stationID + ":" + arrival;
45	torben	980
46	torben	978	DepartureBean departureBean = cache.get(key);
47	torben	308
48	torben	387
49	torben	978	if (departureBean == null) {
50			departureBean = lookupDepartures(stationID,arrival);
51			cache.put(key, departureBean);
52	torben	387	} else {
53	torben	711	Statistics.getInstance().incrementDepartureCacheHits();
54	torben	829	logger.info("Departure: Cache hit " + key); //remove before production
55	torben	387	}
56	torben	980	return departureBean;
57	torben	387	}
58
59
60	torben	978	public DepartureBean lookupDepartures(int stationID, boolean arrival) throws Exception {
61	torben	307
62	torben	978	DepartureBean departureBean = new DepartureBean();
63
64	torben	588	StationBean station = stationDao.getById(stationID);
65	torben	307
66	torben	588	if (station.getRegional() != null) {
67	torben	978	DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
68			departureBean.departureEntries.addAll( tempBean.departureEntries );
69			departureBean.notifications.addAll(tempBean.notifications);
70	torben	307	}
71
72	torben	588	if (station.getStrain() != null) {
73	torben	978	DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
74			departureBean.departureEntries.addAll( tempBean.departureEntries );
75			departureBean.notifications.addAll(tempBean.notifications);
76	torben	588	}
77
78	torben	978	Collections.sort( departureBean.departureEntries );
79	torben	588
80
81	torben	978	return departureBean;
82	torben	305	}
83
84	torben	978	public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
85	torben	580	if (useTempSite == false) {
86	torben	829	return lookupDeparturesNormalSite(stationcode, type, arrival);
87	torben	580	} else {
88	torben	972	//return lookupDeparturesFromTemporarySite(stationcode, type);
89			//TODO: find out what to to if they ever put a temp site up on trafikinfo.bane.dk
90			return null;
91	torben	580	}
92			}
93
94	torben	972	private String getTypeString(TrainType type) {
95			switch (type) {
96			case STOG:
97			return "S-Tog";
98			case REGIONAL:
99			return "Fjerntog";
100			default:
101			return ""; //Can not happen
102			}
103			}
104
105	torben	978	public DepartureBean lookupDeparturesNormalSite(String stationcode, TrainType type, boolean arrival) throws Exception {
106	torben	305
107	torben	978	DepartureBean departureBean = new DepartureBean();
108	torben	992
109	torben	970
110	torben	972	String typeString = getTypeString(type);
111	torben	970	String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
112	torben	994
113			stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
114	torben	970	//String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
115	torben	972	String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
116	torben	970
117	torben	994
118
119	torben	992	//logger.info("URI: " + uri);
120			JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), 2500);
121	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
122	torben	305
123	torben	992	Document page = (Document) breaker.invoke(wrapper);
124	torben	305
125	torben	829	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
126	torben	992	Element table = page.getElementById(tableName);
127	torben	829
128	torben	342	if (table != null) {
129	torben	992	Elements tableRows = table.getElementsByTag("tr");
130	torben	342
131	torben	1020	boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
132			boolean passedTidsstreg = false;
133
134	torben	992	for (Element currentRow : tableRows) {
135			String rowClass = currentRow.attr("class");
136	torben	1020
137			if (tidsstregExists == true && passedTidsstreg == false) {
138			if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
139			passedTidsstreg = true;
140			} else {
141			continue;
142			}
143			}
144
145	torben	342	if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
146	torben	1020
147	torben	992	Elements fields = currentRow.getElementsByTag("td");
148	torben	342
149	torben	978	DepartureEntry departure = new DepartureEntry();
150	torben	342
151	torben	992	String time = fields.get(0).text();
152	torben	375	if (time.equals(""))
153			time = "0:00"; //Bane.dk bug work-around
154	torben	342	departure.setTime(time);
155
156			int updated = extractUpdated( fields.get(1) );
157			departure.setUpdated(updated);
158
159	torben	992	String trainNumber = fields.get(2).text();
160	torben	972	if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
161	torben	349	trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
162	torben	342	departure.setTrainNumber(trainNumber);
163
164	torben	992	String destination = fields.get(3).text();
165	torben	342	departure.setDestination(destination);
166
167	torben	992	String origin = fields.get(4).text();
168	torben	342	departure.setOrigin(origin);
169
170	torben	992	String location = fields.get(5).text();
171	torben	342	departure.setLocation(location);
172
173	torben	992	String status = fields.get(6).text().trim();
174	torben	342	departure.setStatus(status);
175
176			String note = extractNote( fields.get(7) );
177			departure.setNote(note);
178
179	torben	972	departure.setType(typeString);
180	torben	697
181	torben	978	departureBean.departureEntries.add( departure );
182	torben	342	}
183			}
184	torben	348	} else {
185			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
186	torben	305	}
187	torben	978
188	torben	992	Element notifDiv = page.getElementById("station_planlagte_text");
189	torben	978	if (notifDiv != null) {
190
191	torben	992	Elements tables = notifDiv.getElementsByTag("table");
192			for (Element tab : tables) {
193	torben	978
194	torben	992	Elements anchors = tab.getElementsByTag("a");
195	torben	978	if (anchors.size() == 2) {
196	torben	992	departureBean.notifications.add( anchors.get(1).text() );
197	torben	978	}
198			}
199
200			}
201
202
203			return departureBean;
204	torben	305	}
205
206	torben	974	/*
207			@Deprecated
208	torben	580	public List<DepartureBean> lookupDeparturesFromTemporarySite(String stationcode, String type) throws Exception {
209
210			List<DepartureBean> departureList = new ArrayList<DepartureBean>();
211
212	torben	591	final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3);
213	torben	580	webClient.setTimeout(2500);
214			webClient.setJavaScriptEnabled(false);
215
216
217			String uri = "http://bane.dk/lite/station.asp?w=" + type + "&s=" + stationcode;
218
219	torben	939	HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
220	torben	580	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
221
222			HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
223
224			HtmlElement table = page.getElementById("traf_afgang");
225
226			if (table != null) {
227			DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr");
228
229			boolean isFirst = true;
230
231			for (HtmlElement currentRow : tableRows) {
232			if (isFirst == true) { //skip table headers
233			isFirst = false;
234			continue;
235			}
236
237			DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
238
239			DepartureBean departure = new DepartureBean();
240
241			String time = fields.get(0).asText().trim();
242	torben	583
243	torben	580	if (time.equals(""))
244			time = "0:00"; //Bane.dk bug work-around
245			departure.setTime(time);
246
247
248			String trainNumber = fields.get(1).asText();
249			departure.setTrainNumber(trainNumber);
250
251			String destination = fields.get(2).asText();
252			departure.setDestination(destination);
253
254			String origin = fields.get(3).asText();
255			departure.setOrigin(origin);
256
257			String status = fields.get(4).asText();
258			departure.setStatus(status);
259
260	torben	582	String note = fields.get(5).asText();
261	torben	580	departure.setNote(note);
262
263			departureList.add(departure);
264			}
265			} else {
266			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
267			}
268	torben	591	webClient.closeAllWindows();
269	torben	580
270	torben	591
271	torben	580	return departureList;
272	torben	974	}*/
273	torben	580
274
275	torben	992	private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
276	torben	305	int updated = -1;
277
278	torben	992	Elements updatedImgs = updatedTd.getElementsByTag("img");
279			String updatedStr = updatedImgs.get(0).attr("src");
280	torben	305
281			if (updatedStr != null) {
282			for (int i=0; i<updatedStr.length(); i++) {
283			char c = updatedStr.charAt(i);
284			if ( Character.isDigit(c)) {
285			updated = Character.digit(c, 10);
286			break;
287			}
288			}
289			}
290			return updated;
291			}
292
293	torben	992	private String extractNote(Element noteTd) {
294			String note = noteTd.text().trim();
295	torben	313
296	torben	992
297			Elements elems = noteTd.getElementsByClass("bemtype");
298	torben	313	if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
299			note = note.substring(0,note.length() -1 );
300
301			return note;
302			}
303
304	torben	992	private String extractTrainNumber(Element trainTd) {
305			Element anchorElement = trainTd.getElementsByTag("a").get(0);
306			String href = anchorElement.attr("href");
307	torben	349
308	torben	973	int pos = href.lastIndexOf('/');
309			String number = href.substring(pos+1);
310	torben	970
311	torben	349	return number;
312			}
313
314	torben	305	//test
315	torben	580	/*
316	torben	451	public static void main(String args[]) throws Exception {
317	torben	305	DepartureFetcher f = new DepartureFetcher();
318	torben	307	List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
319	torben	305	for(DepartureBean d : deps) {
320			System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
321			d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
322			}
323
324			System.out.println("--------------------------");
325	torben	580	}*/
326	torben	305	}