exec/java-exec/src/test/java/org/apache/drill/exec/physical/resultSet/impl/TestResultSetLoaderOverflow.java

/*
 * Licensed to the Apache Software Foundation (ASF) under one
 * or more contributor license agreements.  See the NOTICE file
 * distributed with this work for additional information
 * regarding copyright ownership.  The ASF licenses this file
 * to you under the Apache License, Version 2.0 (the
 * "License"); you may not use this file except in compliance
 * with the License.  You may obtain a copy of the License at
 *
 * http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
package org.apache.drill.exec.physical.resultSet.impl;

import static org.junit.Assert.assertArrayEquals;
import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertTrue;
import static org.junit.Assert.fail;

import java.nio.charset.StandardCharsets;
import java.util.Arrays;

import org.apache.drill.categories.RowSetTest;
import org.apache.drill.common.exceptions.UserException;
import org.apache.drill.common.types.TypeProtos.DataMode;
import org.apache.drill.common.types.TypeProtos.MinorType;
import org.apache.drill.exec.physical.impl.validate.BatchValidator;
import org.apache.drill.exec.physical.resultSet.ResultSetLoader;
import org.apache.drill.exec.physical.resultSet.RowSetLoader;
import org.apache.drill.exec.physical.resultSet.impl.ResultSetLoaderImpl.ResultSetOptions;
import org.apache.drill.exec.physical.rowSet.RowSet;
import org.apache.drill.exec.physical.rowSet.RowSetReader;
import org.apache.drill.exec.record.MaterializedField;
import org.apache.drill.exec.record.VectorContainer;
import org.apache.drill.exec.record.metadata.SchemaBuilder;
import org.apache.drill.exec.record.metadata.TupleMetadata;
import org.apache.drill.exec.vector.ValueVector;
import org.apache.drill.exec.vector.accessor.ArrayReader;
import org.apache.drill.exec.vector.accessor.ScalarReader;
import org.apache.drill.exec.vector.accessor.ScalarWriter;
import org.apache.drill.test.SubOperatorTest;
import org.junit.Test;
import org.junit.experimental.categories.Category;

/**
 * Exercise the vector overflow functionality for the result set loader.
 */

@Category(RowSetTest.class)
public class TestResultSetLoaderOverflow extends SubOperatorTest {

  /**
   * Test that the writer detects a vector overflow. The offending column
   * value should be moved to the next batch.
   */

  @Test
  public void testVectorSizeLimit() {
    TupleMetadata schema = new SchemaBuilder()
        .add("s", MinorType.VARCHAR)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    rsLoader.startBatch();
    byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');

    // Number of rows should be driven by vector size.
    // Our row count should include the overflow row

    int expectedCount = ValueVector.MAX_BUFFER_SIZE / value.length;
    {
      int count = 0;
      while (! rootWriter.isFull()) {
        rootWriter.start();
        rootWriter.scalar(0).setBytes(value, value.length);
        rootWriter.save();
        count++;
      }

      assertEquals(expectedCount + 1, count);

      // Loader's row count should include only "visible" rows

      assertEquals(expectedCount, rootWriter.rowCount());

      // Total count should include invisible and look-ahead rows.

      assertEquals(expectedCount + 1, rsLoader.totalRowCount());

      // Result should exclude the overflow row

      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(expectedCount, result.rowCount());
      result.clear();
    }

    // Next batch should start with the overflow row

    {
      rsLoader.startBatch();
      assertEquals(1, rootWriter.rowCount());
      assertEquals(expectedCount + 1, rsLoader.totalRowCount());
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(1, result.rowCount());
      result.clear();
    }

    rsLoader.close();
  }

  /**
   * Test that the writer detects a vector overflow. The offending column
   * value should be moved to the next batch.
   */

  @Test
  public void testBatchSizeLimit() {
    TupleMetadata schema = new SchemaBuilder()
        .add("s", MinorType.VARCHAR)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .batchSizeLimit(
            8 * 1024 * 1024 + // Data
            2 * ValueVector.MAX_ROW_COUNT * 4) // Offsets, doubled because of +1
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    rsLoader.startBatch();
    byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');

    // Our row count should include the overflow row

    int expectedCount = 8 * 1024 * 1024 / value.length;

    // First batch, with overflow

    {
      int count = 0;
      while (! rootWriter.isFull()) {
        rootWriter.start();
        rootWriter.scalar(0).setBytes(value, value.length);
        rootWriter.save();
        count++;
      }
      assertEquals(expectedCount + 1, count);

      // Loader's row count should include only "visible" rows

      assertEquals(expectedCount, rootWriter.rowCount());

      // Total count should include invisible and look-ahead rows.

      assertEquals(expectedCount + 1, rsLoader.totalRowCount());

      // Result should exclude the overflow row

      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(expectedCount, result.rowCount());
      result.clear();
    }

    // Next batch should start with the overflow row

    {
      rsLoader.startBatch();
      assertEquals(1, rootWriter.rowCount());
      assertEquals(expectedCount + 1, rsLoader.totalRowCount());
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(1, result.rowCount());
      result.clear();
    }

    rsLoader.close();
  }

  /**
   * Load a batch to overflow. Then, close the loader with the overflow
   * batch unharvested. The Loader should release the memory allocated
   * to the unused overflow vectors.
   */

  @Test
  public void testCloseWithOverflow() {
    TupleMetadata schema = new SchemaBuilder()
        .add("s", MinorType.VARCHAR)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    rsLoader.startBatch();
    byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');
    int count = 0;
    while (! rootWriter.isFull()) {
      rootWriter.start();
      rootWriter.scalar(0).setBytes(value, value.length);
      rootWriter.save();
      count++;
    }

    assertTrue(count < ValueVector.MAX_ROW_COUNT);

    // Harvest the full batch

    VectorContainer container = rsLoader.harvest();
    BatchValidator.validate(container);
    RowSet result = fixture.wrap(container);
    result.clear();

    // Close without harvesting the overflow batch.

    rsLoader.close();
  }

  /**
   * Case where a single array fills up the vector to the maximum size
   * limit. Overflow won't work here; the attempt will fail with a user
   * exception.
   */

  @Test
  public void testOversizeArray() {
    TupleMetadata schema = new SchemaBuilder()
        .addArray("s", MinorType.VARCHAR)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    // Create a single array as the column value in the first row. When
    // this overflows, an exception is thrown since overflow is not possible.

    rsLoader.startBatch();
    byte[] value = new byte[473];
    Arrays.fill(value, (byte) 'X');
    rootWriter.start();
    ScalarWriter array = rootWriter.array(0).scalar();
    try {
      for (int i = 0; i < ValueVector.MAX_ROW_COUNT; i++) {
        array.setBytes(value, value.length);
      }
      fail();
    } catch (UserException e) {
      assertTrue(e.getMessage().contains("column value is larger than the maximum"));
    }
    rsLoader.close();
  }

  /**
   * Test a row with a single array column which overflows. Verifies
   * that all the fiddly bits about offset vectors and so on works
   * correctly. Run this test (the simplest case) if you change anything
   * about the array handling code.
   */

  @Test
  public void testSizeLimitOnArray() {
    TupleMetadata schema = new SchemaBuilder()
        .addArray("s", MinorType.VARCHAR)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    // Fill batch with rows of with a single array, three values each. Tack on
    // a suffix to each so we can be sure the proper data is written and moved
    // to the overflow batch.

    rsLoader.startBatch();
    byte[] value = new byte[473];
    Arrays.fill(value, (byte) 'X');
    String strValue = new String(value, StandardCharsets.UTF_8);
    int valuesPerArray = 13;
    int count = 0;

    {
      int rowSize = 0;
      int totalSize = 0;
      while (rootWriter.start()) {
        totalSize += rowSize;
        rowSize = 0;
        ScalarWriter array = rootWriter.array(0).scalar();
        for (int i = 0; i < valuesPerArray; i++) {
          String cellValue = strValue + (count + 1) + "." + i;
          array.setString(cellValue);
          rowSize += cellValue.length();
        }
        rootWriter.save();
        count++;
      }

      // Row count should include the overflow row.

      int expectedCount = count - 1;

      // Size without overflow row should fit in the vector, size
      // with overflow should not.

      assertTrue(totalSize <= ValueVector.MAX_BUFFER_SIZE);
      assertTrue(totalSize + rowSize > ValueVector.MAX_BUFFER_SIZE);

      // Result should exclude the overflow row. Last row
      // should hold the last full array.

      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(expectedCount, result.rowCount());
      RowSetReader reader = result.reader();
      reader.setPosition(expectedCount - 1);
      ArrayReader arrayReader = reader.array(0);
      ScalarReader strReader = arrayReader.scalar();
      assertEquals(valuesPerArray, arrayReader.size());
      for (int i = 0; i < valuesPerArray; i++) {
        assertTrue(arrayReader.next());
        String cellValue = strValue + (count - 1) + "." + i;
        assertEquals(cellValue, strReader.getString());
      }
      result.clear();
    }

    // Next batch should start with the overflow row.
    // The only row in this next batch should be the whole
    // array being written at the time of overflow.

    {
      rsLoader.startBatch();
      assertEquals(1, rootWriter.rowCount());
      assertEquals(count, rsLoader.totalRowCount());
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(1, result.rowCount());
      RowSetReader reader = result.reader();
      reader.next();
      ArrayReader arrayReader = reader.array(0);
      ScalarReader strReader = arrayReader.scalar();
      assertEquals(valuesPerArray, arrayReader.size());
      for (int i = 0; i < valuesPerArray; i++) {
        assertTrue(arrayReader.next());
        String cellValue = strValue + count + "." + i;
        assertEquals(cellValue, strReader.getString());
      }
      result.clear();
    }

    rsLoader.close();
  }

  /**
   * Test the complete set of array overflow cases:
   * <ul>
   * <li>Array a is written before the column that has overflow,
   * and must be copied, in its entirety, to the overflow row.</li>
   * <li>Column b causes the overflow.</li>
   * <li>Column c is written after the overflow, and should go
   * to the look-ahead row.</li>
   * <li>Column d is written for a while, then has empties before
   * the overflow row, but is written in the overflow row.<li>
   * <li>Column e is like d, but is not written in the overflow
   * row.</li>
   */

  @Test
  public void testArrayOverflowWithOtherArrays() {
    TupleMetadata schema = new SchemaBuilder()
        .addArray("a", MinorType.INT)
        .addArray("b", MinorType.VARCHAR)
        .addArray("c", MinorType.INT)
        .addArray("d", MinorType.INT)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    // Fill batch with rows of with a single array, three values each. Tack on
    // a suffix to each so we can be sure the proper data is written and moved
    // to the overflow batch.

    byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');
    String strValue = new String(value, StandardCharsets.UTF_8);

    int aCount = 3;
    int bCount = 11;
    int cCount = 5;
    int dCount = 7;

    int cCutoff = ValueVector.MAX_BUFFER_SIZE / value.length / bCount / 2;

    ScalarWriter aWriter = rootWriter.array("a").scalar();
    ScalarWriter bWriter = rootWriter.array("b").scalar();
    ScalarWriter cWriter = rootWriter.array("c").scalar();
    ScalarWriter dWriter = rootWriter.array("d").scalar();

    int count = 0;
    rsLoader.startBatch();
    while (rootWriter.start()) {
      for (int i = 0; i < aCount; i++) {
        aWriter.setInt(count * aCount + i);
      }
      for (int i = 0; i < bCount; i++) {
        String cellValue = strValue + (count * bCount + i);
        bWriter.setString(cellValue);
      }
      if (count < cCutoff) {
        for (int i = 0; i < cCount; i++) {
          cWriter.setInt(count * cCount + i);
        }
      }

      // Relies on fact that isFull becomes true right after
      // a vector overflows; don't have to wait for saveRow().

      if (count < cCutoff || rootWriter.isFull()) {
        for (int i = 0; i < dCount; i++) {
          dWriter.setInt(count * dCount + i);
        }
      }
      rootWriter.save();
      count++;
    }

    // Verify

    {
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(count - 1, result.rowCount());
      RowSetReader reader = result.reader();
      ArrayReader aArray = reader.array("a");
      ScalarReader aReader = aArray.scalar();
      ArrayReader bArray = reader.array("b");
      ScalarReader bReader = bArray.scalar();
      ArrayReader cArray = reader.array("c");
      ScalarReader cReader = cArray.scalar();
      ArrayReader dArray = reader.array("d");
      ScalarReader dReader = dArray.scalar();

      while (reader.next()) {
        int rowId = reader.offset();
        assertEquals(aCount, aArray.size());
        for (int i = 0; i < aCount; i++) {
          assertTrue(aArray.next());
          assertEquals(rowId * aCount + i, aReader.getInt());
        }
        assertEquals(bCount, bArray.size());
        for (int i = 0; i < bCount; i++) {
          assertTrue(bArray.next());
          String cellValue = strValue + (rowId * bCount + i);
          assertEquals(cellValue, bReader.getString());
        }
        if (rowId < cCutoff) {
          assertEquals(cCount, cArray.size());
          for (int i = 0; i < cCount; i++) {
            assertTrue(cArray.next());
            assertEquals(rowId * cCount + i, cReader.getInt());
          }
          assertEquals(dCount, dArray.size());
          for (int i = 0; i < dCount; i++) {
            assertTrue(dArray.next());
            assertEquals(rowId * dCount + i, dReader.getInt());
          }
        } else {
          assertEquals(0, cArray.size());
          assertEquals(0, dArray.size());
        }
      }
      result.clear();
    }
    int firstCount = count - 1;

    // One row is in the batch. Write more, skipping over the
    // initial few values for columns c and d. Column d has a
    // roll-over value, c has an empty roll-over.

    rsLoader.startBatch();
    for (int j = 0; j < 5; j++) {
      rootWriter.start();
      for (int i = 0; i < aCount; i++) {
        aWriter.setInt(count * aCount + i);
      }
      for (int i = 0; i < bCount; i++) {
        String cellValue = strValue + (count * bCount + i);
        bWriter.setString(cellValue);
      }
      if (j > 3) {
        for (int i = 0; i < cCount; i++) {
          cWriter.setInt(count * cCount + i);
        }
        for (int i = 0; i < dCount; i++) {
          dWriter.setInt(count * dCount + i);
        }
      }
      rootWriter.save();
      count++;
    }

    {
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(6, result.rowCount());
      RowSetReader reader = result.reader();
      ArrayReader aArray = reader.array("a");
      ScalarReader aReader = aArray.scalar();
      ArrayReader bArray = reader.array("b");
      ScalarReader bReader = bArray.scalar();
      ArrayReader cArray = reader.array("c");
      ScalarReader cReader = cArray.scalar();
      ArrayReader dArray = reader.array("d");
      ScalarReader dReader = dArray.scalar();

      int j = 0;
      while (reader.next()) {
        int rowId = firstCount + reader.offset();
        assertEquals(aCount, aArray.size());
        for (int i = 0; i < aCount; i++) {
          assertTrue(aArray.next());
          assertEquals("Index " + i, rowId * aCount + i, aReader.getInt());
        }
        assertEquals(bCount, bArray.size());
        for (int i = 0; i < bCount; i++) {
          assertTrue(bArray.next());
          String cellValue = strValue + (rowId * bCount + i);
          assertEquals(cellValue, bReader.getString());
        }
        if (j > 4) {
          assertEquals(cCount, cArray.size());
          for (int i = 0; i < cCount; i++) {
            assertTrue(cArray.next());
            assertEquals(rowId * cCount + i, cReader.getInt());
          }
        } else {
          assertEquals(0, cArray.size());
        }
        if (j == 0 || j > 4) {
          assertEquals(dCount, dArray.size());
          for (int i = 0; i < dCount; i++) {
            assertTrue(dArray.next());
            assertEquals(rowId * dCount + i, dReader.getInt());
          }
        } else {
          assertEquals(0, dArray.size());
        }
        j++;
      }
      result.clear();
    }

    rsLoader.close();
  }

  /**
   * Create an array that contains more than 64K values. Drill has no numeric
   * limit on array lengths. (Well, it does, but the limit is about 2 billion
   * which, even for bytes, is too large to fit into a vector...)
   */

  @Test
  public void testLargeArray() {
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator());
    RowSetLoader rootWriter = rsLoader.writer();
    MaterializedField field = SchemaBuilder.columnSchema("a", MinorType.INT, DataMode.REPEATED);
    rootWriter.addColumn(field);

    // Create a single array as the column value in the first row. When
    // this overflows, an exception is thrown since overflow is not possible.

    rsLoader.startBatch();
    rootWriter.start();
    ScalarWriter array = rootWriter.array(0).scalar();
    try {
      for (int i = 0; i < Integer.MAX_VALUE; i++) {
        array.setInt(i+1);
      }
      fail();
    } catch (UserException e) {
      // Expected
    }
    rsLoader.close();
  }

  /**
   * Test the case that an array has "missing values" before the overflow.
   */

  @Test
  public void testMissingArrayValues() {
    TupleMetadata schema = new SchemaBuilder()
        .add("a", MinorType.INT)
        .add("b", MinorType.VARCHAR)
        .addArray("c", MinorType.INT)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');

    int blankAfter = ValueVector.MAX_BUFFER_SIZE / 512 * 2 / 3;
    ScalarWriter cWriter = rootWriter.array("c").scalar();

    rsLoader.startBatch();
    int rowId = 0;
    while (rootWriter.start()) {
      rootWriter.scalar("a").setInt(rowId);
      rootWriter.scalar("b").setBytes(value, value.length);
      if (rowId < blankAfter) {
        for (int i = 0; i < 3; i++) {
          cWriter.setInt(rowId * 3 + i);
        }
      }
      rootWriter.save();
      rowId++;
    }

    VectorContainer container = rsLoader.harvest();
    BatchValidator.validate(container);
    RowSet result = fixture.wrap(container);
    assertEquals(rowId - 1, result.rowCount());
    RowSetReader reader = result.reader();
    ArrayReader cArray = reader.array("c");
    ScalarReader cReader = cArray.scalar();
    while (reader.next()) {
      assertEquals(reader.offset(), reader.scalar("a").getInt());
      assertArrayEquals(value, reader.scalar("b").getBytes());
      if (reader.offset() < blankAfter) {
        assertEquals(3, cArray.size());
        for (int i = 0; i < 3; i++) {
          assertTrue(cArray.next());
          assertEquals(reader.offset() * 3 + i, cReader.getInt());
        }
      } else {
        assertEquals(0, cArray.size());
      }
    }
    result.clear();
    rsLoader.close();
  }

  @Test
  public void testOverflowWithNullables() {
    TupleMetadata schema = new SchemaBuilder()
        .add("n", MinorType.INT)
        .addNullable("a", MinorType.VARCHAR)
        .addNullable("b", MinorType.VARCHAR)
        .addNullable("c", MinorType.VARCHAR)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    rsLoader.startBatch();
    byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');
    int count = 0;
    while (! rootWriter.isFull()) {
      rootWriter.start();
      rootWriter.scalar(0).setInt(count);
      rootWriter.scalar(1).setNull();
      rootWriter.scalar(2).setBytes(value, value.length);
      rootWriter.scalar(3).setNull();
      rootWriter.save();
      count++;
    }

    // Result should exclude the overflow row

    {
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(count - 1, result.rowCount());

      RowSetReader reader = result.reader();
      while (reader.next()) {
        assertEquals(reader.offset(), reader.scalar(0).getInt());
        assertTrue(reader.scalar(1).isNull());
        assertArrayEquals(value, reader.scalar(2).getBytes());
        assertTrue(reader.scalar(3).isNull());
      }
      result.clear();
    }

    // Next batch should start with the overflow row

    rsLoader.startBatch();
    {
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      RowSetReader reader = result.reader();
      assertEquals(1, result.rowCount());
      assertTrue(reader.next());
      assertEquals(count - 1, reader.scalar(0).getInt());
      assertTrue(reader.scalar(1).isNull());
      assertArrayEquals(value, reader.scalar(2).getBytes());
      assertTrue(reader.scalar(3).isNull());
      result.clear();
    }

    rsLoader.close();
  }

  @Test
  public void testVectorSizeLimitWithAppend() {
    TupleMetadata schema = new SchemaBuilder()
        .add("s", MinorType.VARCHAR)
        .buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder()
        .rowCountLimit(ValueVector.MAX_ROW_COUNT)
        .readerSchema(schema)
        .build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();

    rsLoader.startBatch();
    byte[] head = "abc".getBytes();
    byte[] tail = new byte[523];
    Arrays.fill(tail, (byte) 'X');

    String expected = new String(head, StandardCharsets.UTF_8);
    expected += new String(tail, StandardCharsets.UTF_8);
    expected += new String(tail, StandardCharsets.UTF_8);

    int count = 0;
    ScalarWriter colWriter = rootWriter.scalar(0);
    while (! rootWriter.isFull()) {
      rootWriter.start();
      colWriter.setBytes(head, head.length);
      colWriter.appendBytes(tail, tail.length);
      colWriter.appendBytes(tail, tail.length);
      rootWriter.save();
      count++;
    }

    // Number of rows should be driven by vector size.
    // Our row count should include the overflow row

    int valueLength = head.length + 2 * tail.length;
    int expectedCount = ValueVector.MAX_BUFFER_SIZE / valueLength;
    assertEquals(expectedCount + 1, count);

    // Loader's row count should include only "visible" rows

    assertEquals(expectedCount, rootWriter.rowCount());

    // Total count should include invisible and look-ahead rows.

    assertEquals(expectedCount + 1, rsLoader.totalRowCount());

    // Result should exclude the overflow row

    {
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(expectedCount, result.rowCount());

      // Verify that the values were, in fact, appended.

      RowSetReader reader = result.reader();
      while (reader.next()) {
        assertEquals(expected, reader.scalar(0).getString());
      }
      result.clear();
    }

    // Next batch should start with the overflow row

    rsLoader.startBatch();
    assertEquals(1, rootWriter.rowCount());
    assertEquals(expectedCount + 1, rsLoader.totalRowCount());
    {
      VectorContainer container = rsLoader.harvest();
      BatchValidator.validate(container);
      RowSet result = fixture.wrap(container);
      assertEquals(1, result.rowCount());
      RowSetReader reader = result.reader();
      while (reader.next()) {
        assertEquals(expected, reader.scalar(0).getString());
      }
      result.clear();
    }

    rsLoader.close();
  }
}