Merge pull request #823 from zinggAI/enterprise

canopy changes

common/core/src/main/java/zingg/common/core/block/Block.java

@@ -66,16 +66,13 @@ public void setDupes(ZFrame<D,R,C> dupes) {
 	/**
 	 * @return the types
 	 * 
-	 *         public Class[] getTypes() { return types; }
 	 */
 
 	/**
 	 * @param types
-	 *            the types to set
+	 * the types to set
 	 * 
-	 *            public void setTypes(Class[] types) { this.types = types; }
-	 * 
-	 *            /**
+	 *           
 	 * @return the maxSize
 	 */
 	public long getMaxSize() {
@@ -84,7 +81,7 @@ public long getMaxSize() {
 
 	/**
 	 * @param maxSize
-	 *            the maxSize to set
+	 *  the maxSize to set
 	 */
 	public void setMaxSize(long maxSize) {
 		this.maxSize = maxSize;
@@ -102,17 +99,24 @@ protected void setFunctionsMap(ListMap<T, HashFunction<D,R,C,T>> m) {
 		this.functionsMap = m;
 	}
 
+	protected Canopy<R> getCanopy(){
+		return new Canopy<R>();
+	}
 
 	public Canopy<R>getNodeFromCurrent(Canopy<R>node, HashFunction<D,R,C,T> function,
 			FieldDefinition context) {
-		Canopy<R>trial = new Canopy<R>();
+		Canopy<R>trial = getCanopy();
 		trial = node.copyTo(trial);
 		// node.training, node.dupeN, function, context);
 		trial.function = function;
 		trial.context = context;
 		return trial;
 	}
 
+	public void estimateElimCount(Canopy<R> c, long elimCount) {
+		c.estimateElimCount();
+	}
+
 	public abstract T getDataTypeFromString(String t);
 
 	public Canopy<R>getBestNode(Tree<Canopy<R>> tree, Canopy<R>parent, Canopy<R>node,
@@ -122,14 +126,18 @@ protected void setFunctionsMap(ListMap<T, HashFunction<D,R,C,T>> m) {
 		Canopy<R>best = null;
 
 		for (FieldDefinition field : fieldsOfInterest) {
-			LOG.debug("Trying for " + field + " with data type " + field.getDataType() + " and real dt " 
-				+ getDataTypeFromString(field.getDataType()));
+			if (LOG.isDebugEnabled()){
+				LOG.debug("Trying for " + field + " with data type " + field.getDataType() + " and real dt " 
+					+ getDataTypeFromString(field.getDataType()));
+			}
 			//Class type = FieldClass.getFieldClassClass(field.getFieldClass());
 			FieldDefinition context = field;
 			if (least ==0) break;//how much better can it get?
 			// applicable functions
 			List<HashFunction<D,R,C,T>> functions = functionsMap.get(getDataTypeFromString(field.getDataType()));
-			LOG.debug("functions are " + functions);
+			if (LOG.isDebugEnabled()){
+				LOG.debug("functions are " + functions);
+			}
 
 			if (functions != null) {
 
@@ -140,11 +148,13 @@ protected void setFunctionsMap(ListMap<T, HashFunction<D,R,C,T>> m) {
 							//!childless.contains(function, field.fieldName)
 							) 
 							{
-						LOG.debug("Evaluating field " + field.fieldName
+						if (LOG.isDebugEnabled()){
+							LOG.debug("Evaluating field " + field.fieldName
 								+ " and function " + function + " for " + field.dataType);
+						}
 						Canopy<R>trial = getNodeFromCurrent(node, function,
 								context);
-						trial.estimateElimCount();
+						estimateElimCount(trial, least);
 						long elimCount = trial.getElimCount();
 
 
@@ -178,7 +188,9 @@ protected void setFunctionsMap(ListMap<T, HashFunction<D,R,C,T>> m) {
 								}*/
 							}
 							else {
-								LOG.debug("No child " + function);
+								if (LOG.isDebugEnabled()){
+									LOG.debug("No child " + function);
+								}
 								//childless.add(function, field.fieldName);
 							}
 

common/core/src/main/java/zingg/common/core/block/Canopy.java

@@ -20,19 +20,19 @@ public class Canopy<R> implements Serializable {
 	public static final Log LOG = LogFactory.getLog(Canopy.class);
 
 	// created by function edge leading from parent to this node
-	HashFunction function;
+	protected HashFunction function;
 	// aplied on field
-	FieldDefinition context;
+	protected FieldDefinition context;
 	// list of duplicates passed from parent
-	List<R> dupeN;
+	protected List<R> dupeN;
 	// number of duplicates eliminated after function applied on fn context
-	long elimCount;
+	protected long elimCount;
 	// hash of canopy
-	Object hash;
+	protected Object hash;
 	// training set
-	List<R> training;
+	protected List<R> training;
 	// duplicates remaining after function is applied
-	List<R> dupeRemaining;
+	protected List<R> dupeRemaining;
 
 	public Canopy() {
 	}

common/core/src/main/java/zingg/common/core/hash/FirstChars.java

@@ -32,7 +32,7 @@ public String call(String field) {
 				r = field.trim().substring(0, endIndex);
 			}
 			}
-			LOG.debug("Applying " + this.getName() + " on " + field + " and returning " + r);
+			//LOG.debug("Applying " + this.getName() + " on " + field + " and returning " + r);
 			return r;
 	 }
 

spark/core/src/test/java/zingg/block/TestBlock.java

@@ -50,6 +50,7 @@ public void testTree() throws Throwable {
 		// primary deciding is unique year so identityInteger should have been picked
 		Canopy<Row> head = blockingTree.getHead();
 		assertEquals("identityInteger", head.getFunction().getName());
+		blockingTree.toString();
 
 	}