Resolved a memory leak on the AutoCloseConnector class. Updated the T…

…ypeInference class to reduce memory consumption. Refactored the TextClassifier class and improved its speed.

CHANGELOG.md

@@ -1,10 +1,13 @@
 CHANGELOG
 =========
 
-Version 0.6.1 - Build 20150526
+Version 0.6.1 - Build 20151231
 ------------------------------
 
-- Resolving minor bugs (Unreleased Resource problem on tests).
+- Fixed a minor issue related to Unreleased Resources on tests.
+- Resolved a memory leak on the AutoCloseConnector class. Shutdown hooks are now removed when close() is called.
+- Updated the TypeInference class to reduce memory consumption.
+- Refactored the TextClassifier class and improved its speed.
 
 Version 0.6.0 - Build 20150502
 ------------------------------

TODO.txt

@@ -1,6 +1,7 @@
 CODE IMPROVEMENTS
 =================
 
+- Add support to MapDB 2.0.
 - Improve Serialization by setting the serialVersionUID in every serializable class?
 - Create better Exceptions and Exception messages.
 - Add multithreading support.

src/main/java/com/datumbox/applications/nlp/TextClassifier.java

@@ -154,7 +154,8 @@ public void predict(Dataset testDataset) {
         //ensure db loaded
         knowledgeBase.load();
 
-        getPredictions(testDataset);
+        preprocessTestDataset(testDataset);
+        mlmodel.predict(testDataset);
     }
 
     /**
@@ -226,7 +227,8 @@ public BaseMLmodel.ValidationMetrics validate(Dataset testDataset) {
         //ensure db loaded
         knowledgeBase.load();
 
-        BaseMLmodel.ValidationMetrics vm = getPredictions(testDataset);
+        preprocessTestDataset(testDataset);
+        BaseMLmodel.ValidationMetrics vm = mlmodel.validate(testDataset);
 
         return vm;
     }
@@ -294,7 +296,7 @@ protected void _fit(Dataset trainingDataset) {
         }
     }
 
-    private BaseMLmodel.ValidationMetrics getPredictions(Dataset testDataset) {
+    private void preprocessTestDataset(Dataset testDataset) {
         TextClassifier.TrainingParameters trainingParameters = knowledgeBase.getTrainingParameters();
         DatabaseConfiguration dbConf = knowledgeBase.getDbConf();
 
@@ -321,20 +323,10 @@ private BaseMLmodel.ValidationMetrics getPredictions(Dataset testDataset) {
             featureSelection.transform(testDataset);
         }
 
-
         //initialize mlmodel
         if(mlmodel==null) {
             mlmodel = BaseMLmodel.newInstance(trainingParameters.getMLmodelClass(), dbName, dbConf); 
         }
-
-        //call predict of the mlmodel for the new dataset
-        BaseMLmodel.ValidationMetrics vm = mlmodel.validate(testDataset);
-
-        if(transformData) {
-            dataTransformer.denormalize(testDataset); //optional denormization
-        }
-
-        return vm;
     }
 
 }

src/main/java/com/datumbox/common/dataobjects/TypeInference.java

@@ -27,7 +27,6 @@ public class TypeInference {
      * Internal DataTypes used by the Framework.
      */
     public enum DataType {
-        //NOTE: DO NOT CHANGE THE ORDER OF THE ENUMS!!!
         /**
          * Boolean/Dummy Variable.
          * Stored as boolean.
@@ -126,15 +125,24 @@ private boolean isInstance(Object v) {
      * @return 
      */
     public static DataType getDataType(Object v) {
-        for(DataType dataType : DataType.values()) {
-            if(dataType.isInstance(v)) {
-                return dataType;
-            }
+        //NOTE: DO NOT CHANGE THE ORDER OF THE IFS!!!
+        if(DataType.BOOLEAN.isInstance(v)) {
+            return DataType.BOOLEAN;
+        }
+        else if(DataType.ORDINAL.isInstance(v)) {
+            return DataType.ORDINAL;
+        }
+        else if(DataType.NUMERICAL.isInstance(v)) {
+            return DataType.NUMERICAL;
+        }
+        else if(DataType.CATEGORICAL.isInstance(v)) {
+            return DataType.CATEGORICAL;
+        }
+        else {
+            //We will reach this point ONLY if the value is null. 
+            //This is because the last enum value in the loop is Categorical which uses the Object class.
+            return null; 
         }
-
-        //We will reach this point ONLY if the value is null. 
-        //This is because the last enum value in the loop is Categorical which uses the Object class.
-        return null; 
     }
 
     /**

src/main/java/com/datumbox/common/persistentstorage/AutoCloseConnector.java

@@ -16,6 +16,7 @@
 package com.datumbox.common.persistentstorage;
 
 import com.datumbox.common.persistentstorage.interfaces.DatabaseConnector;
+import java.util.concurrent.atomic.AtomicBoolean;
 
 /**
  * Any class that inherits from the abstract AutoCloseConnector class can be used
@@ -27,18 +28,25 @@
  */
 public abstract class AutoCloseConnector implements DatabaseConnector, AutoCloseable {
 
-    private boolean isClosed = false;
+    private final AtomicBoolean isClosed = new AtomicBoolean(false);
+
+    private Thread hook;
 
     /**
      * Protected Constructor which is responsible for adding the Shutdown hook.
      */
     protected AutoCloseConnector() {
-        Runtime.getRuntime().addShutdownHook(new Thread() {
+        hook = new Thread(new Runnable() {
             @Override
             public void run() {
+                AutoCloseConnector.this.hook = null;
+                if(AutoCloseConnector.this.isClosed()) {
+                    return;
+                }
                 AutoCloseConnector.this.close();
             }
         });
+        Runtime.getRuntime().addShutdownHook(hook);
     }
 
     /**
@@ -48,23 +56,29 @@ public void run() {
      */
     @Override
     public boolean isClosed() {
-        return isClosed;
+        return isClosed.get();
     }
 
     /**
      * Marks the connector as closed.
      */
     @Override
     public void close() {
-        isClosed = true;
+        if(isClosed() == false && hook != null) {
+            //remove hook to save memory
+            Runtime.getRuntime().removeShutdownHook(hook);
+            hook = null;
+        }
+        isClosed.set(true);
     }
 
     /**
      * Ensures the connection is not closed.
      */
     protected void ensureNotClosed() {
-        if(isClosed) {
+        if(isClosed()) {
             throw new RuntimeException("The connector is already closed");
         }
     }
+
 }