[54] | 1 | package de.ugoe.cs.eventbench.web;
|
---|
| 2 |
|
---|
| 3 | import java.io.File;
|
---|
| 4 | import java.io.FileNotFoundException;
|
---|
| 5 | import java.io.FileReader;
|
---|
| 6 | import java.io.IOException;
|
---|
| 7 | import java.text.ParseException;
|
---|
| 8 | import java.text.SimpleDateFormat;
|
---|
| 9 | import java.util.ArrayList;
|
---|
| 10 | import java.util.HashMap;
|
---|
| 11 | import java.util.LinkedList;
|
---|
| 12 | import java.util.List;
|
---|
| 13 | import java.util.Map;
|
---|
| 14 |
|
---|
| 15 | import de.ugoe.cs.eventbench.web.data.WebEvent;
|
---|
| 16 |
|
---|
| 17 | public class WeblogParser {
|
---|
| 18 |
|
---|
| 19 | private long timeout;
|
---|
| 20 |
|
---|
| 21 | private List<List<WebEvent>> sequences;
|
---|
| 22 |
|
---|
| 23 | public WeblogParser() {
|
---|
| 24 | timeout = 3600000; // 1 hour session-timeout as default
|
---|
| 25 | }
|
---|
| 26 |
|
---|
| 27 | public WeblogParser(long timeout) {
|
---|
| 28 | this.timeout = timeout;
|
---|
| 29 | }
|
---|
| 30 |
|
---|
| 31 | public List<List<WebEvent>> getSequences() {
|
---|
| 32 | return sequences;
|
---|
| 33 | }
|
---|
| 34 |
|
---|
| 35 | public void parseFile(String filename) throws IOException, FileNotFoundException, ParseException {
|
---|
| 36 | File f = new File(filename);
|
---|
| 37 | FileReader reader = new FileReader(f);
|
---|
| 38 | char[] buffer = new char[(int) f.length()];
|
---|
| 39 | reader.read(buffer);
|
---|
| 40 | reader.close();
|
---|
| 41 | String[] lines = (new String(buffer)).split("\n");
|
---|
| 42 |
|
---|
| 43 | Map<String, List<Integer>> cookieSessionMap = new HashMap<String, List<Integer>>();
|
---|
| 44 | int lastId = -1;
|
---|
| 45 |
|
---|
| 46 | SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
|
---|
| 47 |
|
---|
| 48 | sequences = new ArrayList<List<WebEvent>>();
|
---|
| 49 |
|
---|
| 50 | for( String line : lines ) {
|
---|
| 51 | String[] values = line.trim().split(" ");
|
---|
| 52 |
|
---|
| 53 | // use cookie as session identifier
|
---|
| 54 | int cookieStart = values[0].lastIndexOf('.');
|
---|
| 55 | String cookie = values[0].substring(cookieStart+1);
|
---|
| 56 | String dateString = values[1].substring(1)+" "+values[2].substring(0, values[2].length()-1);
|
---|
| 57 | long timestamp = dateFormat.parse(dateString).getTime();
|
---|
| 58 | String uri = values[3];
|
---|
| 59 | // String ref = values[4]; // referer is not yet used!
|
---|
| 60 | List<String> postedVars = new ArrayList<String>();
|
---|
| 61 | for( int i=5 ; i<values.length ; i++ ) {
|
---|
| 62 | postedVars.add(values[i]);
|
---|
| 63 | }
|
---|
| 64 |
|
---|
| 65 | WebEvent event = new WebEvent(uri, timestamp, postedVars);
|
---|
| 66 |
|
---|
| 67 | // find session and add event
|
---|
| 68 | List<Integer> sessionIds = cookieSessionMap.get(cookie);
|
---|
| 69 | if( sessionIds==null ) {
|
---|
| 70 | sessionIds = new ArrayList<Integer>();
|
---|
| 71 | // start new session
|
---|
| 72 | sessionIds.add(++lastId);
|
---|
| 73 | cookieSessionMap.put(cookie, sessionIds);
|
---|
| 74 | sequences.add(new LinkedList<WebEvent>());
|
---|
| 75 | }
|
---|
| 76 | Integer lastSessionIndex = sessionIds.get(sessionIds.size()-1);
|
---|
| 77 | List<WebEvent> lastSession = sequences.get(lastSessionIndex);
|
---|
| 78 | long lastEventTime = timestamp;
|
---|
| 79 | if( !lastSession.isEmpty() ) {
|
---|
| 80 | lastEventTime = lastSession.get(lastSession.size()-1).getTimestamp();
|
---|
| 81 | }
|
---|
| 82 | if( timestamp-lastEventTime>timeout ) {
|
---|
| 83 | sessionIds.add(++lastId);
|
---|
| 84 | List<WebEvent> newSession = new LinkedList<WebEvent>();
|
---|
| 85 | newSession.add(event);
|
---|
| 86 | sequences.add(newSession);
|
---|
| 87 | } else {
|
---|
| 88 | lastSession.add(event);
|
---|
| 89 | }
|
---|
| 90 | }
|
---|
| 91 | }
|
---|
| 92 | }
|
---|