Generics

core/src/main/java/zingg/hash/TrimLast1DigitDbl.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TrimLast1DigitDbl extends TrimLastDigitsDbl {
+public abstract class TrimLast1DigitDbl extends TrimLastDigitsDbl {
 
 	public TrimLast1DigitDbl() {
 		super(1);

core/src/main/java/zingg/hash/TrimLast1DigitInt.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TrimLast1DigitInt extends TrimLastDigitsInt {
+public abstract class TrimLast1DigitInt extends TrimLastDigitsInt {
 
 	public TrimLast1DigitInt() {
 		super(1);

core/src/main/java/zingg/hash/TrimLast2DigitsDbl.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TrimLast2DigitsDbl extends TrimLastDigitsDbl {
+public abstract class TrimLast2DigitsDbl extends TrimLastDigitsDbl {
 
 	public TrimLast2DigitsDbl() {
 		super(2);

core/src/main/java/zingg/hash/TrimLast2DigitsInt.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TrimLast2DigitsInt extends TrimLastDigitsInt {
+public abstract class TrimLast2DigitsInt extends TrimLastDigitsInt {
 
 	public TrimLast2DigitsInt() {
 		super(2);

core/src/main/java/zingg/hash/TrimLast3DigitsDbl.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TrimLast3DigitsDbl extends TrimLastDigitsDbl {
+public abstract class TrimLast3DigitsDbl extends TrimLastDigitsDbl {
 
 	public TrimLast3DigitsDbl() {
 		super(3);

core/src/main/java/zingg/hash/TrimLast3DigitsInt.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TrimLast3DigitsInt extends TrimLastDigitsInt {
+public abstract class TrimLast3DigitsInt extends TrimLastDigitsInt {
 
 	public TrimLast3DigitsInt() {
 		super(3);

core/src/main/java/zingg/hash/TrimLastDigitsDbl.java

@@ -1,20 +1,23 @@
 package zingg.hash;
 
-import org.apache.spark.sql.Row;
-import org.apache.spark.sql.api.java.UDF1;
-import org.apache.spark.sql.types.DataTypes;
-
-import zingg.client.ZFrame;
-
-public class TrimLastDigitsDbl extends HashFunction implements UDF1<Double, Double> {
+/**
+ * Base class for hash functions related to trimming of doubles
+ * @author vikasgupta
+ *
+ * @param <D>
+ * @param <R>
+ * @param <C>
+ * @param <T>
+ */
+public abstract class TrimLastDigitsDbl<D,R,C,T> extends HashFunction<D,R,C,T> {
 	int numDigits;
 	static final int[] POWERS_OF_10 = {1, 10, 100, 1000, 10000, 100000};
 	public TrimLastDigitsDbl(int count) {
-		super("trimLast" + count + "DigitsDbl", DataTypes.DoubleType, DataTypes.DoubleType, true);
+		super("trimLast" + count + "DigitsDbl");
 		this.numDigits = count;
 	}
 
-	@Override
+
 	public Double call(Double field) {
 		Double r = null;
 		if (field == null) {
@@ -24,39 +27,5 @@ public Double call(Double field) {
 		}
 		return r;
 	}
-
-	public Object apply(Row ds, String column) {
-		return call((Double) ds.getAs(column));
-	}
-
-	@Override
-	public ZFrame apply(ZFrame ds, String column, String newColumn) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object getAs(Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object getAs(Object df, Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object apply(Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object apply(Object df, Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
 
 }

core/src/main/java/zingg/hash/TrimLastDigitsInt.java

@@ -1,20 +1,22 @@
 package zingg.hash;
 
-import org.apache.spark.sql.Row;
-import org.apache.spark.sql.api.java.UDF1;
-import org.apache.spark.sql.types.DataTypes;
-
-import zingg.client.ZFrame;
-
-public class TrimLastDigitsInt<D,R,C,T> extends HashFunction<D,R,C,T> implements UDF1<Integer, Integer> {
+/**
+ * Base class for hash functions related to trimming of integers
+ * @author vikasgupta
+ *
+ * @param <D>
+ * @param <R>
+ * @param <C>
+ * @param <T>
+ */
+public abstract class TrimLastDigitsInt<D,R,C,T> extends HashFunction<D,R,C,T> {
 	int numDigits;
 	static final int[] POWERS_OF_10 = {1, 10, 100, 1000, 10000, 100000};
 	public TrimLastDigitsInt(int count) {
 		super("trimLast" + count + "DigitsInt");//, DataTypes.IntegerType, DataTypes.IntegerType, true);
 		this.numDigits = count;
 	}
 
-	@Override
 	public Integer call(Integer field) {
 		Integer r = null;
 		if (field == null) {
@@ -25,38 +27,5 @@ public Integer call(Integer field) {
 		return r;
 	}
 
-	public Object apply(Row ds, String column) {
-		return call((Integer) ds.getAs(column));
-	}
-
-	@Override
-	public ZFrame apply(ZFrame ds, String column, String newColumn) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object getAs(Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object getAs(Object df, Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
 
-	@Override
-	public Object apply(Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object apply(Object df, Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
 }

core/src/main/java/zingg/hash/TruncateDouble.java

@@ -1,12 +1,16 @@
 package zingg.hash;
 
-import org.apache.spark.sql.Row;
-import org.apache.spark.sql.api.java.UDF1;
-import org.apache.spark.sql.types.DataTypes;
-
-import zingg.client.ZFrame;
-
-public class TruncateDouble<D,R,C,T> extends HashFunction<D,R,C,T> {
+/**
+ * Base class for hash functions related to truncating of doubles
+ * 
+ * @author vikasgupta
+ *
+ * @param <D>
+ * @param <R>
+ * @param <C>
+ * @param <T>
+ */
+public abstract class TruncateDouble<D,R,C,T> extends HashFunction<D,R,C,T> {
 	int numDecimalPlaces;
 	static final int[] POWERS_OF_10 = {1, 10, 100, 1000, 10000, 100000};
 	public TruncateDouble(int numDecimalPlaces) {
@@ -25,36 +29,4 @@ public Double call(Double field) {
 		return r;
 	}
 
-
-
-	@Override
-	public ZFrame apply(ZFrame<D,R,C> ds, String column, String newColumn) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object getAs(Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object getAs(Object df, Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object apply(Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
-	@Override
-	public Object apply(Object df, Object r, String column) {
-		// TODO Auto-generated method stub
-		return null;
-	}
-
 }

core/src/main/java/zingg/hash/TruncateDoubleTo1Place.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TruncateDoubleTo1Place extends TruncateDouble {
+public abstract class TruncateDoubleTo1Place extends TruncateDouble {
 
 	public TruncateDoubleTo1Place() {
 		super(1);

core/src/main/java/zingg/hash/TruncateDoubleTo2Places.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TruncateDoubleTo2Places extends TruncateDouble {
+public abstract class TruncateDoubleTo2Places extends TruncateDouble {
 
 	public TruncateDoubleTo2Places() {
 		super(2);

core/src/main/java/zingg/hash/TruncateDoubleTo3Places.java

@@ -1,6 +1,6 @@
 package zingg.hash;
 
-public class TruncateDoubleTo3Places extends TruncateDouble {
+public abstract class TruncateDoubleTo3Places extends TruncateDouble {
 
 	public TruncateDoubleTo3Places() {
 		super(3);

core/src/main/java/zingg/spark/SparkLabelUpdater.java

@@ -10,30 +10,42 @@
 import org.apache.spark.sql.types.DataType;
 
 import zingg.LabelUpdater;
+import zingg.client.Arguments;
 import zingg.client.ZinggClientException;
 import zingg.client.ZinggOptions;
 import zingg.client.pipe.Pipe;
 import zingg.client.pipe.SparkPipe;
 
 
-
+/**
+ * Spark specific implementation of LabelUpdater
+ * 
+ * @author vikasgupta
+ *
+ */
 public class SparkLabelUpdater extends LabelUpdater<SparkSession, Dataset<Row>, Row, Column,DataType> {
 
 	public static String name = "zingg.spark.SparkLabelUpdater";
 	public static final Log LOG = LogFactory.getLog(SparkLabelUpdater.class);
 
 	public SparkLabelUpdater() {
 		setZinggOptions(ZinggOptions.UPDATE_LABEL);
+		setContext(new ZinggSparkContext());
 	}
 
-
+
+    @Override
+    public void init(Arguments args, String license)  throws ZinggClientException {
+        super.init(args, license);
+        getContext().init(license);
+    }
+
 	@Override
 	public void cleanup() throws ZinggClientException {
 		// TODO Auto-generated method stub
 
 	}
 
-
 	protected Pipe setSaveModeOnPipe(Pipe p) {
 		SparkPipe pipe = (SparkPipe) p;
 		pipe.setMode(SaveMode.Overwrite);

core/src/main/java/zingg/spark/SparkLabeller.java

@@ -6,15 +6,19 @@
 import org.apache.spark.sql.Dataset;
 import org.apache.spark.sql.Row;
 import org.apache.spark.sql.SparkSession;
-
 import org.apache.spark.sql.types.DataType;
 
 import zingg.Labeller;
 import zingg.client.Arguments;
 import zingg.client.ZinggClientException;
 import zingg.client.ZinggOptions;
 
-
+/**
+ * Spark specific implementation of Labeller
+ * 
+ * @author vikasgupta
+ *
+ */
 public class SparkLabeller extends Labeller<SparkSession, Dataset<Row>, Row, Column,DataType> {
 
 	public static String name = "zingg.spark.SparkLabeller";
@@ -28,11 +32,12 @@ public SparkLabeller() {
 		setContext(new ZinggSparkContext());
 	}
 
-	@Override
-	public void init(Arguments args, String license)  throws ZinggClientException {
-		super.init(args, license);
-		getContext().init(license);
-	}
+
+  @Override
+  public void init(Arguments args, String license)  throws ZinggClientException {
+    super.init(args, license);
+    getContext().init(license);
+  }
 
 
 	@Override