ARROW-13792 [Java]: The toString representation is incorrect for unsigned integer vectors

java/vector/src/main/java/org/apache/arrow/vector/UInt1Vector.java

@@ -29,6 +29,7 @@
 import org.apache.arrow.vector.types.pojo.Field;
 import org.apache.arrow.vector.types.pojo.FieldType;
 import org.apache.arrow.vector.util.TransferPair;
+import org.apache.arrow.vector.util.ValueVectorUtility;
 
 /**
  * UInt1Vector implements a fixed width (1 bytes) vector of
@@ -328,7 +329,10 @@ public long getValueAsLong(int index) {
     return this.get(index) & PROMOTION_MASK;
   }
 
-
+  @Override
+  public String toString() {
+    return ValueVectorUtility.getToString(this, 0, getValueCount(), (v, i) -> v.getObjectNoOverflow(i));
+  }
 
   private class TransferImpl implements TransferPair {
     UInt1Vector to;

java/vector/src/main/java/org/apache/arrow/vector/UInt2Vector.java

@@ -29,6 +29,7 @@
 import org.apache.arrow.vector.types.pojo.Field;
 import org.apache.arrow.vector.types.pojo.FieldType;
 import org.apache.arrow.vector.util.TransferPair;
+import org.apache.arrow.vector.util.ValueVectorUtility;
 
 /**
  * UInt2Vector implements a fixed width (2 bytes) vector of
@@ -305,6 +306,12 @@ public long getValueAsLong(int index) {
     return this.get(index);
   }
 
+  @Override
+  public String toString() {
+    return ValueVectorUtility.getToString(this, 0, getValueCount(), (v, i) ->
+      v.isNull(i) ? "null" : Integer.toString(v.get(i) & 0x0000ffff));
+  }
+
   private class TransferImpl implements TransferPair {
     UInt2Vector to;
 

java/vector/src/main/java/org/apache/arrow/vector/UInt4Vector.java

@@ -29,6 +29,7 @@
 import org.apache.arrow.vector.types.pojo.Field;
 import org.apache.arrow.vector.types.pojo.FieldType;
 import org.apache.arrow.vector.util.TransferPair;
+import org.apache.arrow.vector.util.ValueVectorUtility;
 
 /**
  * UInt4Vector implements a fixed width (4 bytes) vector of
@@ -300,6 +301,11 @@ public long getValueAsLong(int index) {
     return this.get(index) & PROMOTION_MASK;
   }
 
+  @Override
+  public String toString() {
+    return ValueVectorUtility.getToString(this, 0, getValueCount(), (v, i) -> v.getObjectNoOverflow(i));
+  }
+
   private class TransferImpl implements TransferPair {
     UInt4Vector to;
 

java/vector/src/main/java/org/apache/arrow/vector/UInt8Vector.java

@@ -31,6 +31,7 @@
 import org.apache.arrow.vector.types.pojo.Field;
 import org.apache.arrow.vector.types.pojo.FieldType;
 import org.apache.arrow.vector.util.TransferPair;
+import org.apache.arrow.vector.util.ValueVectorUtility;
 
 /**
  * UInt8Vector implements a fixed width vector (8 bytes) of
@@ -296,6 +297,11 @@ public long getValueAsLong(int index) {
     return this.get(index);
   }
 
+  @Override
+  public String toString() {
+    return ValueVectorUtility.getToString(this, 0, getValueCount(), (v, i) -> v.getObjectNoOverflow(i));
+  }
+
   private class TransferImpl implements TransferPair {
     UInt8Vector to;
 

java/vector/src/main/java/org/apache/arrow/vector/util/ValueVectorUtility.java

@@ -19,6 +19,8 @@
 
 import static org.apache.arrow.vector.validate.ValidateUtil.validateOrThrow;
 
+import java.util.function.BiFunction;
+
 import org.apache.arrow.util.Preconditions;
 import org.apache.arrow.vector.BaseFixedWidthVector;
 import org.apache.arrow.vector.ValueVector;
@@ -37,7 +39,7 @@ private ValueVectorUtility() {
 
   /**
    * Get the toString() representation of vector suitable for debugging.
-   * Note since vectors may have millions of values, this method only show max 20 values.
+   * Note since vectors may have millions of values, this method only shows max 20 values.
    * Examples as below (v represents value):
    * <li>
    *   vector with 0 value:
@@ -52,7 +54,20 @@ private ValueVectorUtility() {
    *  [v0, v1, v2, v3, v4, v5, v6, v7, v8, v9, ..., v90, v91, v92, v93, v94, v95, v96, v97, v98, v99]
    * </li>
    */
-  public static String getToString(ValueVector vector, int start, int end) {
+  public static <V extends ValueVector> String getToString(V vector, int start, int end) {
+    return getToString(vector, start, end, (v, i) -> v.getObject(i));
+  }
+
+  /**
+   * Get the toString() representation of vector suitable for debugging.
+   * Note since vectors may have millions of values, this method only shows at most 20 values.
+   * @param vector the vector for which to get toString representation.
+   * @param start the starting index, inclusive.
+   * @param end the end index, exclusive.
+   * @param valueToString the function to transform individual elements to strings.
+   */
+  public static <V extends ValueVector> String getToString(
+      V vector, int start, int end, BiFunction<V, Integer, Object> valueToString) {
     Preconditions.checkNotNull(vector);
     final int length = end - start;
     Preconditions.checkArgument(length >= 0);
@@ -77,7 +92,7 @@ public static String getToString(ValueVector vector, int start, int end) {
         i = end - window - 1;
         skipComma = true;
       } else {
-        sb.append(vector.getObject(i));
+        sb.append(valueToString.apply(vector, i));
       }
 
       if (i == end - 1) {

java/vector/src/test/java/org/apache/arrow/vector/TestValueVector.java

@@ -2791,6 +2791,38 @@ public void testToString() {
     }
   }
 
+  @Test
+  public void testUInt1VectorToString() {
+    try (final UInt1Vector uInt1Vector = new UInt1Vector("uInt1Vector", allocator)) {
+      setVector(uInt1Vector, (byte) 0xff);
+      assertEquals("[255]", uInt1Vector.toString());
+    }
+  }
+
+  @Test
+  public void testUInt2VectorToString() {
+    try (final UInt2Vector uInt2Vector = new UInt2Vector("uInt2Vector", allocator)) {
+      setVector(uInt2Vector, (char) 0xffff);
+      assertEquals("[65535]", uInt2Vector.toString());
+    }
+  }
+
+  @Test
+  public void testUInt4VectorToString() {
+    try (final UInt4Vector uInt4Vector = new UInt4Vector("uInt4Vector", allocator)) {
+      setVector(uInt4Vector, 0xffffffff);
+      assertEquals("[4294967295]", uInt4Vector.toString());
+    }
+  }
+
+  @Test
+  public void testUInt8VectorToString() {
+    try (final UInt8Vector uInt8Vector = new UInt8Vector("uInt8Vector", allocator)) {
+      setVector(uInt8Vector, 0xffffffffffffffffL);
+      assertEquals("[18446744073709551615]", uInt8Vector.toString());
+    }
+  }
+
   @Test
   public void testUnloadVariableWidthVector() {
     try (final VarCharVector varCharVector = new VarCharVector("var char", allocator)) {