Changeset 68 for trunk

Timestamp:

06/15/11 16:54:11 (13 years ago)

Author:

sherbold

Message:

• changed weblog parser such that session without a minimum length are pruned to remove single page visits from the sessions (e.g., through robots)
• changed loadSessionsFromClickstream such that the session timeout and the minimum session length are optional parameters

Location:

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

Files:

• WeblogParser.java (modified) (3 diffs)
• commands/CMDloadSessionsFromClickstream.java (modified) (2 diffs)

trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java

@@ -14 +14 @@
 
 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;
@@ -31 +34 @@
         public List<List<WebEvent>> getSequences() {
                 return sequences; 
+        }
+        
+        public void setTimeout(long timeout) {
+                this.timeout = timeout;
+        }
+        
+        public void setMinLength(int minLength) {
+                this.minLength = minLength;
         }
         
@@ -89 +100 @@
                         }
                 }
+                Console.traceln(""+sequences.size()+ " user sequences found");
+                // prune sequences shorter than min-length
+                for( int i=0; i<sequences.size(); i++ ) {
+                        if( sequences.get(i).size()<minLength ) {
+                                sequences.remove(i);
+                        }
+                }
+                Console.traceln(""+sequences.size()+ " remaining after pruning of sequences shorter than " + minLength);
         }
 }

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

@@ -16 +16 @@
         @Override
         public void run(List<Object> parameters) {
-                // TODO Auto-generated method stub
                 if( parameters.size() < 1 ) {
                         throw new InvalidParameterException();
                 }
                 String source = (String) parameters.get(0);
+                int timeout = -1;
+                int minLength = -1;
+                if( parameters.size() ==3 ) {
+                        timeout = Integer.parseInt((String) parameters.get(1));
+                        minLength = Integer.parseInt((String) parameters.get(2));
+                }
                 
-                WeblogParser parser = new WeblogParser();
+                WeblogParser parser = new WeblogParser();;
+                if( timeout!=-1 ) {
+                        parser.setTimeout(timeout);
+                        parser.setMinLength(minLength);
+                }
                 try {
                         parser.parseFile(source);
@@ -39 +48 @@
         @Override
         public void help() {
-                // TODO Auto-generated method stub
-                
+                Console.println("Usage: loadSessionsFromClickstream <filename> {<timeout> <minSessionLength>}");
         }