Changeset 224 for trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java

Timestamp:

09/30/11 16:38:42 (13 years ago)

Author:

sherbold

Message:

• added parameter maxLength to command loadSessionsFromClickstream and the WeblogParser? for pruning of unexplained very long sequences

File:

• trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java (modified) (6 diffs)

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

@@ -45 +45 @@
         /**
          * <p>
+         * Maximal length of a session. All longer sessions will be prunde. Default:
+         * 100
+         * </p>
+         */
+        private int maxLength = 100;
+
+        /**
+         * <p>
          * Collection of generated sequences.
          * </p>
@@ -121 +129 @@
         public void setMinLength(int minLength) {
                 this.minLength = minLength;
+        }
+
+        /**
+         * <p>
+         * Sets the maximal length of a session. All sessions that contain more
+         * events will be pruned.
+         * </p>
+         * 
+         * @param maxLength
+         *            new maximal length
+         */
+        public void setMaxLength(int maxLength) {
+                this.maxLength = maxLength;
         }
 
@@ -181 +202 @@
                                         String path = uri.getPath();
                                         List<String> getVars = extractGetVarsFromUri(uri);
-        
+
                                         WebEvent event = new WebEvent(path, timestamp, postedVars,
                                                         getVars);
-        
+
                                         // find session and add event
                                         List<Integer> sessionIds = cookieSessionMap.get(cookie);
@@ -196 +217 @@
                                         Integer lastSessionIndex = sessionIds
                                                         .get(sessionIds.size() - 1);
-                                        List<WebEvent> lastSession = sequences.get(lastSessionIndex);
+                                        List<WebEvent> lastSession = sequences
+                                                        .get(lastSessionIndex);
                                         long lastEventTime = timestamp;
                                         if (!lastSession.isEmpty()) {
@@ -211 +233 @@
                                         }
                                 } catch (URISyntaxException e) {
-                                        Console.traceln("Ignored line " + lineCounter + ": " + e.getMessage());
+                                        Console.traceln("Ignored line " + lineCounter + ": "
+                                                        + e.getMessage());
                                 }
                         }
                 }
-                pruneShortSequences();
+                pruneSequences();
         }
 
@@ -223 +246 @@
          * </p>
          */
-        private void pruneShortSequences() {
+        private void pruneSequences() {
                 Console.traceln("" + sequences.size() + " user sequences found");
-                // prune sequences shorter than min-length
+                // prune sequences shorter than min-length and longer than maxLength
                 int i = 0;
                 while (i < sequences.size()) {
-                        if (sequences.get(i).size() < minLength) {
+                        if ((sequences.get(i).size() < minLength)
+                                        || sequences.get(i).size() > maxLength) {
                                 sequences.remove(i);
                         } else {