Merge pull request #11 from nRo/develop

Develop

Author: nRo

README.md

@@ -24,7 +24,7 @@ Add this to you pom.xml
     <dependency>
         <groupId>de.unknownreality</groupId>
         <artifactId>dataframe</artifactId>
-        <version>0.7.5</version>
+        <version>0.7.5.1</version>
     </dependency>
 ...
 </dependencies>
@@ -52,7 +52,7 @@ To build the library from sources:
     <dependency>
         <groupId>de.unknownreality</groupId>
         <artifactId>dataframe</artifactId>
-        <version>0.7.5-SNAPSHOT</version>
+        <version>0.7.5.1-SNAPSHOT</version>
     </dependency>
 ...
 </dependencies>

pom.xml

@@ -6,7 +6,7 @@
 
     <groupId>de.unknownreality</groupId>
     <artifactId>dataframe</artifactId>
-    <version>0.7.5-SNAPSHOT</version>
+    <version>0.7.5.1-SNAPSHOT</version>
     <properties>
         <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
         <maven.compiler.source>1.8</maven.compiler.source>

src/main/java/de/unknownreality/dataframe/ColumnSelection.java

@@ -2,11 +2,11 @@
 
 import de.unknownreality.dataframe.filter.FilterPredicate;
 
-public class ColumnSelection<T extends DataFrame> {
+public class ColumnSelection{
     private DataFrameColumn[] columns;
-    private T dataFrame;
+    private DataFrame dataFrame;
 
-    public ColumnSelection(T dataFrame, DataFrameColumn... columns){
+    public ColumnSelection(DataFrame dataFrame, DataFrameColumn... columns){
         this.dataFrame = dataFrame;
         this.columns = columns;
     }
@@ -19,7 +19,7 @@ public ColumnSelection(T dataFrame, DataFrameColumn... columns){
      * @param value   input value
      * @return new dataframe
      */
-    public T where(String colName, Comparable value){
+    public DataFrame where(String colName, Comparable value){
         DataRows rows = dataFrame.selectRows(colName,value);
         return createDataFrame(rows);
     }
@@ -31,7 +31,7 @@ public T where(String colName, Comparable value){
      * @param predicate input predicate
      * @return new dataframe
      */
-    public T where(FilterPredicate predicate){
+    public DataFrame where(FilterPredicate predicate){
         DataRows rows = dataFrame.selectRows(predicate);
         return createDataFrame(rows);
     }
@@ -43,7 +43,7 @@ public T where(FilterPredicate predicate){
      * @param predicateString input predicate string
      * @return new dataframe
      */
-    public T where(String predicateString){
+    public DataFrame where(String predicateString){
         DataRows rows = dataFrame.selectRows(predicateString);
         return createDataFrame(rows);
     }
@@ -56,7 +56,7 @@ public T where(String predicateString){
      * @param values index values
      * @return new dataframe
      */
-    public T whereIndex(String indexName, Comparable... values){
+    public DataFrame whereIndex(String indexName, Comparable... values){
         DataRows rows = dataFrame.selectRowsByIndex(indexName, values);
         return createDataFrame(rows);
     }
@@ -66,19 +66,14 @@ public T whereIndex(String indexName, Comparable... values){
      *
      * @return new dataframe
      */
-    public T allRows(){
+    public DataFrame allRows(){
         DataRows rows = dataFrame.getRows();
         return createDataFrame(rows);
     }
 
     @SuppressWarnings("unchecked")
-    private T createDataFrame(DataRows rows){
-        T df;
-        try {
-            df = (T)dataFrame.getClass().newInstance();
-        } catch (Exception e) {
-            throw new DataFrameRuntimeException("error creating dataframe instance",e);
-        }
+    private DataFrame createDataFrame(DataRows rows){
+        DataFrame df = new DefaultDataFrame();
         for(DataFrameColumn column : columns){
             df.addColumn(column.copyEmpty());
         }

src/main/java/de/unknownreality/dataframe/DataFrame.java

@@ -89,7 +89,7 @@ public interface DataFrame extends DataContainer<DataFrameHeader, DataRow> {
      * @param columnNames selected columns
      * @return {@link ColumnSelection} for row selection
      */
-    ColumnSelection<? extends DataFrame> selectColumns(String... columnNames);
+    ColumnSelection selectColumns(String... columnNames);
 
     /**
      * Returns a {@link ColumnSelection} for the selected columns.
@@ -98,7 +98,7 @@ public interface DataFrame extends DataContainer<DataFrameHeader, DataRow> {
      * @param columns selected columns
      * @return {@link ColumnSelection} for row selection
      */
-    ColumnSelection<? extends DataFrame> selectColumns(DataFrameColumn... columns);
+    ColumnSelection selectColumns(DataFrameColumn... columns);
 
     /**
      * Adds a column to the data frame.

src/main/java/de/unknownreality/dataframe/DataRow.java

@@ -54,6 +54,9 @@ public int size() {
      */
     @Override
     public boolean isCompatible(Comparable value, String headerName) {
+        if(value == null || value instanceof Values.NA){
+            return true;
+        }
         Class<? extends Comparable> type = getHeader().getType(headerName);
         if(Number.class.isAssignableFrom(type)){
             return Number.class.isAssignableFrom(value.getClass());

src/main/java/de/unknownreality/dataframe/DefaultDataFrame.java

@@ -148,7 +148,7 @@ public DefaultDataFrame replaceColumn(DataFrameColumn existing, DataFrameColumn
 
     @Override
     @SuppressWarnings("unchecked")
-    public ColumnSelection<DefaultDataFrame> selectColumns(String... columnNames) {
+    public ColumnSelection selectColumns(String... columnNames) {
         DataFrameColumn[] columns = new DataFrameColumn[columnNames.length];
         for (int i = 0; i < columnNames.length; i++) {
             columns[i] = getColumn(columnNames[i]);
@@ -158,14 +158,19 @@ public ColumnSelection<DefaultDataFrame> selectColumns(String... columnNames) {
 
     @Override
     @SuppressWarnings("unchecked")
-    public ColumnSelection<DefaultDataFrame> selectColumns(DataFrameColumn... columns) {
+    public ColumnSelection selectColumns(DataFrameColumn... columns) {
         return new ColumnSelection(this, columns);
     }
 
 
     @Override
     @SuppressWarnings("unchecked")
     public DefaultDataFrame addColumn(DataFrameColumn column) {
+        if(column.size() == 0 && size != 0){
+            column.appendAll(
+                    Arrays.asList(new Values.NA[size])
+            );
+        }
         if (columns != null && column.size() != size) {
             throw new DataFrameRuntimeException("column lengths must be equal");
         }
@@ -1142,7 +1147,7 @@ public void setValue(int col, int row, Comparable newValue) {
         if (col >= columns.length || row > size) {
             throw new DataFrameRuntimeException("index out of bounds");
         }
-        if (newValue == Values.NA) {
+        if (newValue == null || newValue == Values.NA) {
             columns[col].setNA(row);
         } else {
             columns[col].set(row, newValue);

src/main/java/de/unknownreality/dataframe/column/BasicColumn.java

@@ -29,6 +29,7 @@
 import de.unknownreality.dataframe.DataFrameRuntimeException;
 import de.unknownreality.dataframe.MapFunction;
 import de.unknownreality.dataframe.Values;
+import de.unknownreality.dataframe.common.ValueComparator;
 
 import java.lang.reflect.Array;
 import java.util.*;
@@ -82,13 +83,14 @@ protected void doSort(Comparator<T> comparator) {
 
     @Override
     protected void doSort() {
-        Arrays.sort(values, 0, size());
+        Arrays.sort(values,0,size(), ValueComparator.COMPARATOR);
+
     }
 
 
     @Override
     protected void doSet(int index, T value) {
-        if (value == Values.NA) {
+        if (value == null || value == Values.NA) {
             doSetNA(index);
             return;
         }

src/main/java/de/unknownreality/dataframe/common/ValueComparator.java

@@ -0,0 +1,27 @@
+package de.unknownreality.dataframe.common;
+
+import java.util.Comparator;
+
+public class ValueComparator implements Comparator<Comparable> {
+
+    public final static ValueComparator COMPARATOR = new ValueComparator(1);
+    private int naOrder = 1;
+
+    public ValueComparator(int naOrder){
+        this.naOrder = naOrder;
+    }
+
+    @Override
+    public int compare(Comparable o1, Comparable o2) {
+        if(o1 == null && o2 == null){
+            return 0;
+        }
+        if(o1 == null) {
+            return -1*naOrder;
+        }
+        if(o2 == null){
+            return 1*naOrder;
+        }
+        return o1.compareTo(o2);
+    }
+}

src/test/java/de/unknownreality/dataframe/frame/DataFrameTest.java

@@ -93,8 +93,11 @@ public void testCreation() {
 
         }
         dataFrame = dataFrame.concat(dataFrame2);
+        dataFrame.addStringColumn("null_test");
         Assert.assertEquals(2, dataFrame.size());
         Assert.assertEquals(2d, dataFrame.getRow(1).toDouble("value"), 0d);
+        Assert.assertEquals("NA", dataFrame.getRow(1).getString("null_test"));
+        Assert.assertEquals(true, dataFrame.getRow(1).isNA("null_test"));
 
 
         Assert.assertEquals(valueColumn, dataFrame2.getDoubleColumn("value"));
@@ -176,10 +179,13 @@ public void testReader() throws IOException {
 
         DataRow row = df.getRow(1);
         row.set("A", 999);
-        df.update(row);
 
-        Assert.assertEquals(new Integer(999), df.getRow(1).getInteger("A"));
+        Assert.assertEquals(Integer.valueOf(999), df.getRow(1).getInteger("A"));
+        row.set("A", null);
+        Assert.assertEquals(true, df.getRow(1).isNA("A"));
 
+        row.set("A", Values.NA);
+        Assert.assertEquals(true, df.getRow(1).isNA("A"));
         df.getColumn("A").sort();
     }
 
@@ -371,7 +377,13 @@ public void testNA() throws IOException {
         DataFrame df = DataFrameLoader.load(csv,csvReader);
         List<DataRow> r = df.getRows();
         int i = 0;
+        df.addStringColumn("test");
+        df.addDoubleColumn("test2");
+
         for (DataRow row : df.rows()) {
+            Assert.assertEquals(Values.NA, row.get("test"));
+            Assert.assertEquals(Values.NA, row.get("test2"));
+
             if (i == 2) {
                 Assert.assertEquals(true, row.isNA("B"));
             } else {