Examples with ResultSetOptions org.apache.drill.exec.physical.resultSet.impl.ResultSetLoaderImpl.ResultSetOptions used on opensource projects

Example 31 with ResultSetOptions

use of org.apache.drill.exec.physical.resultSet.impl.ResultSetLoaderImpl.ResultSetOptions in project drill by apache.

the class TestResultSetLoaderOverflow method testOversizeArray.

/**
 * 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();
}

Example 32 with ResultSetOptions

use of org.apache.drill.exec.physical.resultSet.impl.ResultSetLoaderImpl.ResultSetOptions in project drill by apache.

the class TestResultSetLoaderOverflow method testBatchSizeLimit.

/**
 * 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(// Data
    8 * 1024 * 1024 + // Offsets, doubled because of +1
    2 * ValueVector.MAX_ROW_COUNT * 4).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();
}

Example 33 with ResultSetOptions

use of org.apache.drill.exec.physical.resultSet.impl.ResultSetLoaderImpl.ResultSetOptions in project drill by apache.

the class TestResultSetLoaderOverflow method testVectorSizeLimit.

/**
 * 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();
}

Example 34 with ResultSetOptions

use of org.apache.drill.exec.physical.resultSet.impl.ResultSetLoaderImpl.ResultSetOptions in project drill by apache.

the class TestResultSetLoaderOverflow method testArrayOverflowWithOtherArrays.

/**
 * 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, Charsets.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);
            }
        }
        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();
}

Example 35 with ResultSetOptions

use of org.apache.drill.exec.physical.resultSet.impl.ResultSetLoaderImpl.ResultSetOptions in project drill by apache.

the class TestResultSetLoaderLimits method testLimit1.

/**
 * Pathological limit case: a single row.
 */
@Test
public void testLimit1() {
    // Start with a small limit.
    ResultSetOptions options = new ResultSetOptionBuilder().limit(1).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    assertTrue(rsLoader.startBatch());
    assertEquals(1, rsLoader.maxBatchSize());
    RowSetLoader rootWriter = rsLoader.writer();
    rootWriter.addColumn(SchemaBuilder.columnSchema("s", MinorType.VARCHAR, DataMode.REQUIRED));
    rootWriter.addRow("foo");
    assertTrue(rootWriter.isFull());
    assertFalse(rootWriter.start());
    RowSet result = fixture.wrap(rsLoader.harvest());
    assertEquals(1, result.rowCount());
    result.clear();
    assertTrue(rsLoader.atLimit());
    rsLoader.close();
}