Changeset 54

Timestamp:

06/09/11 16:26:49 (13 years ago)

Author:

sherbold

Message:

• refactored web usage log parsing

Location:

trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web

Files:

• WeblogParser.java (added)
• commands/CMDloadSessionsFromClickstream.java (modified) (2 diffs)

trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/commands/CMDloadSessionsFromClickstream.java

@@ -1 +1 @@
 package de.ugoe.cs.eventbench.web.commands;
 
-import java.io.File;
-import java.io.FileReader;
+import java.io.FileNotFoundException;
+import java.io.IOException;
 import java.security.InvalidParameterException;
-import java.text.SimpleDateFormat;
-import java.util.ArrayList;
-import java.util.HashMap;
-import java.util.LinkedList;
+import java.text.ParseException;
 import java.util.List;
-import java.util.Map;
 
-import de.ugoe.cs.eventbench.web.data.WebEvent;
+import de.ugoe.cs.eventbench.data.GlobalDataContainer;
+import de.ugoe.cs.eventbench.web.WeblogParser;
 import de.ugoe.cs.util.console.Command;
 import de.ugoe.cs.util.console.Console;
@@ -25 +22 @@
                 String source = (String) parameters.get(0);
                 
-                // TODO needs to be variable
-                long timeout = 3600000; // 1 hour session-timeout
-                
-                try { // TODO dummy try/catch block
-                        File f = new File(source);
-                        FileReader reader = new FileReader(f);
-                        char[] buffer = new char[(int) f.length()];
-                        reader.read(buffer);
-                        reader.close();
-                        String[] lines = (new String(buffer)).split("\n");
-                        
-                        Map<String, List<Integer>> cookieSessionMap = new HashMap<String, List<Integer>>();
-                        int lastId = -1;
-                        
-                        List<List<WebEvent>> sessions = new ArrayList<List<WebEvent>>();
-                        
-                        SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
-                        
-                        for( String line : lines ) {
-                                String[] values = line.trim().split(" ");
-                                
-                                // use cookie as session identifier
-                                int cookieStart = values[0].lastIndexOf('.');
-                                String cookie = values[0].substring(cookieStart+1);
-                                String dateString = values[1].substring(1)+" "+values[2].substring(0, values[2].length()-1);
-                                long timestamp = dateFormat.parse(dateString).getTime();
-                                String uri = values[3];
-                                String ref = values[4];
-                                List<String> postedVars = new ArrayList<String>();
-                                for( int i=5 ; i<values.length ; i++ ) {
-                                        postedVars.add(values[i]);
-                                }
-                                
-                                        
-                                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);
-                                        sessions.add(new LinkedList<WebEvent>());
-                                } 
-                                Integer lastSessionIndex = sessionIds.get(sessionIds.size()-1);
-                                List<WebEvent> lastSession = sessions.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);
-                                        sessions.add(newSession);
-                                } else {
-                                        lastSession.add(event);
-                                }
-                        } 
-                        Console.traceln(sessions.toString());
-                } catch(Exception e) {
-                        e.printStackTrace();
+                WeblogParser parser = new WeblogParser();
+                try {
+                        parser.parseFile(source);
+                } catch (FileNotFoundException e) {
+                        Console.println(e.getMessage());
+                } catch (IOException e) {
+                        Console.println(e.getMessage());
+                } catch (ParseException e) {
+                        Console.println("Invalid format of date stamps.");
+                        Console.println(e.getMessage());
                 }
                 
+                GlobalDataContainer.getInstance().addData("sequences", parser.getSequences());
         }