Changeset 111 for trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java

Timestamp:

07/07/11 11:04:27 (13 years ago)

Author:

sherbold

Message:

• URI of web sessions is not parsed and split into path and GET parameters; of the GET parameters only the name is used to define the type

File:

• trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java (modified) (5 diffs)

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

@@ -3 +3 @@
 import java.io.FileNotFoundException;
 import java.io.IOException;
+import java.net.URI;
+import java.net.URISyntaxException;
 import java.text.ParseException;
 import java.text.SimpleDateFormat;
@@ -47 +49 @@
         }
         
-        public void parseFile(String filename) throws IOException, FileNotFoundException, ParseException {
+        public void parseFile(String filename) throws IOException, FileNotFoundException, ParseException, URISyntaxException {
                 String[] lines = FileTools.getLinesFromFile(filename);
                 
@@ -66 +68 @@
                         String dateString = values[1];
                         long timestamp = dateFormat.parse(dateString).getTime();
-                        String uri = values[2];
+                        String uriString = values[2];
                         // String ref = values[3]; // referer is not yet used!
                         String agent;
@@ -77 +79 @@
                         List<String> postedVars = new ArrayList<String>();
                         if( values.length==6 ) { // post vars found
-                                for( String postVar : values[5].split(" ") ) {
+                                for( String postVar : values[5].trim().split(" ") ) {
                                         postedVars.add(postVar);
                                 }
                         }
                         if( !isRobot(agent) ) {
-                                WebEvent event = new WebEvent(uri, timestamp, postedVars);
+                                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
@@ -141 +148 @@
                 return agent.matches(robotRegex);
         }
+        
+        private List<String> extractGetVarsFromUri(URI uri) {
+                List<String> getVars = new ArrayList<String>();
+                String query = uri.getQuery();
+                if( query!=null ) {
+                        String[] paramPairs = query.split("&");
+                        for( String paramPair : paramPairs ) {
+                                String[] paramSplit = paramPair.split("=");
+                                getVars.add(paramSplit[0]);
+                        }
+                }
+                return getVars;
+        }
 }