Changeset 177 for trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java

Timestamp:

09/10/11 00:30:54 (13 years ago)

Author:

sherbold

Message:

• Changed handling of invalid URIs when parsing web usage logs. Before, the parsing was aborted, now, the entry with the invalid URI is ignored.

File:

• trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java (modified) (3 diffs)

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

@@ -135 +135 @@
          * @throws ParseException
          *             thrown the date format is invalid
-         * @throws URISyntaxException
-         *             thrown if the URI is invalid
          */
         public void parseFile(String filename) throws IOException,
-                        FileNotFoundException, ParseException, URISyntaxException {
+                        FileNotFoundException, ParseException {
                 String[] lines = FileTools.getLinesFromFile(filename);
 
@@ -151 +149 @@
                 sequences = new ArrayList<List<WebEvent>>();
 
+                int lineCounter = 0;
                 for (String line : lines) {
+                        lineCounter++;
                         String[] values = line.substring(1, line.length() - 1).split(
                                         "\" \"");
@@ -176 +176 @@
                         }
                         if (!isRobot(agent)) {
-                                URI uri = new URI(uriString);
-
-                                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);
-                                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);
+                                try {
+                                        URI uri = new URI(uriString);
+                                        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);
+                                        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);
+                                        }
+                                } catch (URISyntaxException e) {
+                                        Console.traceln("Ignored line " + lineCounter + ": " + e.getMessage());
                                 }
                         }