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);
                
                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), 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	1021	departureBean.stationName = station.getName();
67
68	torben	588	if (station.getRegional() != null) {
69	torben	978	DepartureBean tempBean = lookupDepartures(station.getRegional(), TrainType.REGIONAL, arrival);
70			departureBean.departureEntries.addAll( tempBean.departureEntries );
71			departureBean.notifications.addAll(tempBean.notifications);
72	torben	307	}
73
74	torben	588	if (station.getStrain() != null) {
75	torben	978	DepartureBean tempBean = lookupDepartures(station.getStrain(), TrainType.STOG, arrival);
76			departureBean.departureEntries.addAll( tempBean.departureEntries );
77			departureBean.notifications.addAll(tempBean.notifications);
78	torben	588	}
79
80	torben	978	Collections.sort( departureBean.departureEntries );
81	torben	588
82
83	torben	978	return departureBean;
84	torben	305	}
85
86	torben	978	public DepartureBean lookupDepartures(String stationcode, TrainType type, boolean arrival) throws Exception {
87	torben	580	if (useTempSite == false) {
88	torben	829	return lookupDeparturesNormalSite(stationcode, type, arrival);
89	torben	580	} else {
90	torben	972	//return lookupDeparturesFromTemporarySite(stationcode, type);
91			//TODO: find out what to to if they ever put a temp site up on trafikinfo.bane.dk
92			return null;
93	torben	580	}
94			}
95
96	torben	972	private String getTypeString(TrainType type) {
97			switch (type) {
98			case STOG:
99			return "S-Tog";
100			case REGIONAL:
101			return "Fjerntog";
102			default:
103			return ""; //Can not happen
104			}
105			}
106
107	torben	978	public DepartureBean lookupDeparturesNormalSite(String stationcode, TrainType type, boolean arrival) throws Exception {
108	torben	305
109	torben	978	DepartureBean departureBean = new DepartureBean();
110	torben	992
111	torben	970
112	torben	972	String typeString = getTypeString(type);
113	torben	970	String arrivalDeparture = (arrival==false) ? "Afgang" : "Ankomst";
114	torben	994
115			stationcode = URLEncoder.encode(stationcode,"ISO-8859-1");
116	torben	970	//String uri = "http://www.bane.dk/visStation.asp?ArtikelID=4275&W=" + type + "&S=" + stationcode;
117	torben	972	String uri = "http://trafikinfo.bane.dk/Trafikinformation/AfgangAnkomst/" + arrivalDeparture + "/" + stationcode + "/" + typeString + "/UdvidetVisning";
118	torben	970
119	torben	994
120
121	torben	992	//logger.info("URI: " + uri);
122			JsoupInvocation wrapper = new JsoupInvocation( new URL(uri), 2500);
123	torben	421	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
124	torben	305
125	torben	992	Document page = (Document) breaker.invoke(wrapper);
126	torben	305
127	torben	829	String tableName = arrival == false ? "afgangtabel" : "ankomsttabel";
128	torben	992	Element table = page.getElementById(tableName);
129	torben	829
130	torben	342	if (table != null) {
131	torben	992	Elements tableRows = table.getElementsByTag("tr");
132	torben	342
133	torben	1020	boolean tidsstregExists = (table.getElementsByAttributeValue("class", "Tidsstreg").size() > 0);
134			boolean passedTidsstreg = false;
135
136	torben	992	for (Element currentRow : tableRows) {
137			String rowClass = currentRow.attr("class");
138	torben	1020
139			if (tidsstregExists == true && passedTidsstreg == false) {
140			if (currentRow.getElementsByAttributeValue("class", "Tidsstreg").size() > 0) {
141			passedTidsstreg = true;
142			} else {
143			continue;
144			}
145			}
146
147	torben	342	if (rowClass != null && rowClass.toLowerCase().contains("station") ) {
148	torben	1020
149	torben	992	Elements fields = currentRow.getElementsByTag("td");
150	torben	342
151	torben	978	DepartureEntry departure = new DepartureEntry();
152	torben	342
153	torben	992	String time = fields.get(0).text();
154	torben	375	if (time.equals(""))
155			time = "0:00"; //Bane.dk bug work-around
156	torben	342	departure.setTime(time);
157
158			int updated = extractUpdated( fields.get(1) );
159			departure.setUpdated(updated);
160
161	torben	992	String trainNumber = fields.get(2).text();
162	torben	972	if (type == TrainType.STOG) //If it is S-train we need to extract the trainNumber
163	torben	349	trainNumber = trainNumber + " " + extractTrainNumber(fields.get(2));
164	torben	342	departure.setTrainNumber(trainNumber);
165
166	torben	992	String destination = fields.get(3).text();
167	torben	342	departure.setDestination(destination);
168
169	torben	992	String origin = fields.get(4).text();
170	torben	342	departure.setOrigin(origin);
171
172	torben	992	String location = fields.get(5).text();
173	torben	342	departure.setLocation(location);
174
175	torben	992	String status = fields.get(6).text().trim();
176	torben	342	departure.setStatus(status);
177
178			String note = extractNote( fields.get(7) );
179			departure.setNote(note);
180
181	torben	972	departure.setType(typeString);
182	torben	697
183	torben	978	departureBean.departureEntries.add( departure );
184	torben	342	}
185			}
186	torben	348	} else {
187			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
188	torben	305	}
189	torben	978
190	torben	992	Element notifDiv = page.getElementById("station_planlagte_text");
191	torben	978	if (notifDiv != null) {
192
193	torben	992	Elements tables = notifDiv.getElementsByTag("table");
194			for (Element tab : tables) {
195	torben	978
196	torben	992	Elements anchors = tab.getElementsByTag("a");
197	torben	978	if (anchors.size() == 2) {
198	torben	992	departureBean.notifications.add( anchors.get(1).text() );
199	torben	978	}
200			}
201
202			}
203
204
205			return departureBean;
206	torben	305	}
207
208	torben	974	/*
209			@Deprecated
210	torben	580	public List<DepartureBean> lookupDeparturesFromTemporarySite(String stationcode, String type) throws Exception {
211
212			List<DepartureBean> departureList = new ArrayList<DepartureBean>();
213
214	torben	591	final WebClient webClient = new WebClient(BrowserVersion.FIREFOX_3);
215	torben	580	webClient.setTimeout(2500);
216			webClient.setJavaScriptEnabled(false);
217
218
219			String uri = "http://bane.dk/lite/station.asp?w=" + type + "&s=" + stationcode;
220
221	torben	939	HtmlunitInvocation wrapper = new HtmlunitInvocation(webClient, uri);
222	torben	580	CircuitBreaker breaker = CircuitBreakerManager.getManager().getCircuitBreaker("banedk");
223
224			HtmlPage page = (HtmlPage) breaker.invoke(wrapper);
225
226			HtmlElement table = page.getElementById("traf_afgang");
227
228			if (table != null) {
229			DomNodeList<HtmlElement> tableRows = table.getElementsByTagName("tr");
230
231			boolean isFirst = true;
232
233			for (HtmlElement currentRow : tableRows) {
234			if (isFirst == true) { //skip table headers
235			isFirst = false;
236			continue;
237			}
238
239			DomNodeList<HtmlElement> fields = currentRow.getElementsByTagName("td");
240
241			DepartureBean departure = new DepartureBean();
242
243			String time = fields.get(0).asText().trim();
244	torben	583
245	torben	580	if (time.equals(""))
246			time = "0:00"; //Bane.dk bug work-around
247			departure.setTime(time);
248
249
250			String trainNumber = fields.get(1).asText();
251			departure.setTrainNumber(trainNumber);
252
253			String destination = fields.get(2).asText();
254			departure.setDestination(destination);
255
256			String origin = fields.get(3).asText();
257			departure.setOrigin(origin);
258
259			String status = fields.get(4).asText();
260			departure.setStatus(status);
261
262	torben	582	String note = fields.get(5).asText();
263	torben	580	departure.setNote(note);
264
265			departureList.add(departure);
266			}
267			} else {
268			logger.warning("No departures found for station=" + stationcode + ", type=" + type);
269			}
270	torben	591	webClient.closeAllWindows();
271	torben	580
272	torben	591
273	torben	580	return departureList;
274	torben	974	}*/
275	torben	580
276
277	torben	992	private int extractUpdated(Element updatedTd) { //extract the digit (in this case: 4) from "media/trafikinfo/opdater4.gif"
278	torben	305	int updated = -1;
279
280	torben	992	Elements updatedImgs = updatedTd.getElementsByTag("img");
281			String updatedStr = updatedImgs.get(0).attr("src");
282	torben	305
283			if (updatedStr != null) {
284			for (int i=0; i<updatedStr.length(); i++) {
285			char c = updatedStr.charAt(i);
286			if ( Character.isDigit(c)) {
287			updated = Character.digit(c, 10);
288			break;
289			}
290			}
291			}
292			return updated;
293			}
294
295	torben	992	private String extractNote(Element noteTd) {
296			String note = noteTd.text().trim();
297	torben	313
298	torben	992
299			Elements elems = noteTd.getElementsByClass("bemtype");
300	torben	313	if (elems.size() > 0 && note.charAt(note.length()-1) == 'i')
301			note = note.substring(0,note.length() -1 );
302
303			return note;
304			}
305
306	torben	992	private String extractTrainNumber(Element trainTd) {
307			Element anchorElement = trainTd.getElementsByTag("a").get(0);
308			String href = anchorElement.attr("href");
309	torben	349
310	torben	973	int pos = href.lastIndexOf('/');
311			String number = href.substring(pos+1);
312	torben	970
313	torben	349	return number;
314			}
315
316	torben	305	//test
317	torben	580	/*
318	torben	451	public static void main(String args[]) throws Exception {
319	torben	305	DepartureFetcher f = new DepartureFetcher();
320	torben	307	List<DepartureBean> deps = f.lookupDepartures("AR", "FJRN");
321	torben	305	for(DepartureBean d : deps) {
322			System.out.println( d.getTime() + ";" + d.getUpdated() + ";" + d.getTrainNumber() + ";" +
323			d.getDestination() + ";" + d.getOrigin() + ";" + d.getLocation() + ";" + d.getStatus() + ";" + d.getNote() );
324			}
325
326			System.out.println("--------------------------");
327	torben	580	}*/
328	torben	305	}