source: trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java @ 73

package de.ugoe.cs.eventbench.web;

import java.io.File;
import java.io.FileNotFoundException;
import java.io.FileReader;
import java.io.IOException;
import java.text.ParseException;
import java.text.SimpleDateFormat;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;

import de.ugoe.cs.eventbench.web.data.WebEvent;
import de.ugoe.cs.util.console.Console;

public class WeblogParser {
        
        private long timeout;
        
        private int minLength = 2;
        
        private List<List<WebEvent>> sequences;
        
        private static final String ROBOTFILTERFILE = "misc/robotfilter.txt";
        
        private String robotRegex = ".*";
        
        public WeblogParser() {
                timeout = 3600000; // 1 hour session-timeout as default
        }
        
        public WeblogParser(long timeout) {
                this.timeout = timeout;
        }
        
        public List<List<WebEvent>> getSequences() {
                return sequences; 
        }
        
        public void setTimeout(long timeout) {
                this.timeout = timeout;
        }
        
        public void setMinLength(int minLength) {
                this.minLength = minLength;
        }
        
        public void parseFile(String filename) throws IOException, FileNotFoundException, ParseException {
                File f = new File(filename);
                FileReader reader = new FileReader(f);
                char[] buffer = new char[(int) f.length()];
                reader.read(buffer);
                reader.close();
                String[] lines = (new String(buffer)).split("\r\n");
                
                Map<String, List<Integer>> cookieSessionMap = new HashMap<String, List<Integer>>();
                int lastId = -1;
                
                SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
                loadRobotRegex();
                
                sequences = new ArrayList<List<WebEvent>>();
                
                for( String line : lines ) {
                        String[] values = line.substring(1, line.length()-1).split("\" \"");
                        
                        // use cookie as session identifier
                        int cookieStart = values[0].lastIndexOf('.');
                        String cookie = values[0].substring(cookieStart+1);
                        String dateString = values[1];
                        long timestamp = dateFormat.parse(dateString).getTime();
                        String uri = values[2];
                        // String ref = values[3]; // referer is not yet used!
                        String agent = values[4]; // agent is not yet used!
                        List<String> postedVars = new ArrayList<String>();
                        if( values.length==6 ) { // post vars found
                                for( String postVar : values[5].split(" ") ) {
                                        postedVars.add(postVar);
                                }
                        }
                        if( !isRobot(agent) ) {
                                WebEvent event = new WebEvent(uri, timestamp, postedVars);
                                
                                // find session and add event
                                List<Integer> sessionIds = cookieSessionMap.get(cookie);
                                if( sessionIds==null ) {
                                        sessionIds = new ArrayList<Integer>();
                                        // start new session
                                        sessionIds.add(++lastId);
                                        cookieSessionMap.put(cookie, sessionIds);
                                        sequences.add(new LinkedList<WebEvent>());
                                } 
                                Integer lastSessionIndex = sessionIds.get(sessionIds.size()-1);
                                List<WebEvent> lastSession = sequences.get(lastSessionIndex);
                                long lastEventTime = timestamp;
                                if( !lastSession.isEmpty() ) {
                                        lastEventTime = lastSession.get(lastSession.size()-1).getTimestamp();
                                }
                                if( timestamp-lastEventTime>timeout ) {
                                        sessionIds.add(++lastId);
                                        List<WebEvent> newSession = new LinkedList<WebEvent>();
                                        newSession.add(event);
                                        sequences.add(newSession);
                                } else {
                                        lastSession.add(event);
                                }
                        }
                }
                Console.traceln(""+sequences.size()+ " user sequences found");
                // prune sequences shorter than min-length
                int i=0;
                while( i<sequences.size() ) {
                        if( sequences.get(i).size()<minLength ) {
                                sequences.remove(i);
                        } else {
                                Console.traceln(""+sequences.get(i).size());
                                if( sequences.get(i).size() > 152 ) {
                                        Console.traceln(sequences.get(i).toString().replaceAll(", ", "\n"));
                                }
                                i++;
                        }
                }
                Console.traceln(""+sequences.size()+ " remaining after pruning of sequences shorter than " + minLength);
        }
        
        private void loadRobotRegex() throws IOException, FileNotFoundException {
                File f = new File(ROBOTFILTERFILE);
                FileReader reader = new FileReader(f);
                char[] buffer = new char[(int) f.length()];
                reader.read(buffer);
                reader.close();
                String[] lines = (new String(buffer)).split("\r\n");
                StringBuilder regex = new StringBuilder();
                for( int i=0; i<lines.length; i++ ) {
                        regex.append("(.*"+lines[i]+".*)");
                        if( i!=lines.length-1 ) {
                                regex.append("|");
                        }
                }
                robotRegex = regex.toString();
        }
        
        private boolean isRobot(String agent) {
                return agent.matches(robotRegex);
        }
}