Search in sources :

Example 6 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by axbaretto.

the class TestVariableWidthWriter method testSizeLimit.

/**
 * Test resize monitoring. Add a listener to an Varchar writer,
 * capture each resize, and refuse a resize when the s
 * of the vector exceeds 1 MB. This will trigger an overflow,
 * which will throw an exception which we then check for.
 */
@Test
public void testSizeLimit() {
    try (VarCharVector vector = allocVector(1000)) {
        TestIndex index = new TestIndex();
        VarCharColumnWriter writer = makeWriter(vector, index);
        writer.bindListener(new ColumnWriterListener() {

            // Because assumed array size is 10, so 10 * 1000 = 10,000
            // rounded to 16K
            int totalAlloc = 16384;

            @Override
            public void overflowed(ScalarWriter writer) {
                throw new IllegalStateException("overflow called");
            }

            @Override
            public boolean canExpand(ScalarWriter writer, int delta) {
                // System.out.println("Delta: " + delta);
                totalAlloc += delta;
                return totalAlloc < 1024 * 1024;
            }
        });
        writer.startWrite();
        byte[] value = new byte[423];
        Arrays.fill(value, (byte) 'X');
        try {
            for (int i = 0; ; i++) {
                index.index = i;
                writer.startRow();
                writer.setBytes(value, value.length);
                writer.saveRow();
            }
        } catch (IllegalStateException e) {
            assertTrue(e.getMessage().contains("overflow called"));
        }
    }
}
Also used : ColumnWriterListener(org.apache.drill.exec.vector.accessor.ScalarWriter.ColumnWriterListener) TestIndex(org.apache.drill.test.rowSet.test.TestFixedWidthWriter.TestIndex) VarCharColumnWriter(org.apache.drill.exec.vector.accessor.ColumnAccessors.VarCharColumnWriter) VarCharVector(org.apache.drill.exec.vector.VarCharVector) ScalarWriter(org.apache.drill.exec.vector.accessor.ScalarWriter) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 7 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by axbaretto.

the class RowSetTest method testMapStructure.

/**
 * Test a simple map structure at the top level of a row.
 *
 * @throws VectorOverflowException should never occur
 */
@Test
public void testMapStructure() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMap("m").addArray("b", MinorType.INT).resumeSchema().buildSchema();
    ExtendableRowSet rowSet = fixture.rowSet(schema);
    RowSetWriter writer = rowSet.writer();
    // Map and Int
    // Test Invariants
    assertEquals(ObjectType.SCALAR, writer.column("a").type());
    assertEquals(ObjectType.SCALAR, writer.column(0).type());
    assertEquals(ObjectType.TUPLE, writer.column("m").type());
    assertEquals(ObjectType.TUPLE, writer.column(1).type());
    assertSame(writer.column(1).tuple(), writer.tuple(1));
    TupleWriter mapWriter = writer.column(1).tuple();
    assertEquals(ObjectType.SCALAR, mapWriter.column("b").array().entry().type());
    assertEquals(ObjectType.SCALAR, mapWriter.column("b").array().entryType());
    ScalarWriter aWriter = writer.column("a").scalar();
    ScalarWriter bWriter = writer.column("m").tuple().column("b").array().entry().scalar();
    assertSame(bWriter, writer.tuple(1).array(0).scalar());
    assertEquals(ValueType.INTEGER, bWriter.valueType());
    try {
        writer.column(1).scalar();
        fail();
    } catch (UnsupportedOperationException e) {
    // Expected
    }
    try {
        writer.column(1).array();
        fail();
    } catch (UnsupportedOperationException e) {
    // Expected
    }
    // Write data
    aWriter.setInt(10);
    bWriter.setInt(11);
    bWriter.setInt(12);
    writer.save();
    aWriter.setInt(20);
    bWriter.setInt(21);
    bWriter.setInt(22);
    writer.save();
    aWriter.setInt(30);
    bWriter.setInt(31);
    bWriter.setInt(32);
    writer.save();
    // Finish the row set and get a reader.
    SingleRowSet actual = writer.done();
    RowSetReader reader = actual.reader();
    assertEquals(ObjectType.SCALAR, reader.column("a").type());
    assertEquals(ObjectType.SCALAR, reader.column(0).type());
    assertEquals(ObjectType.TUPLE, reader.column("m").type());
    assertEquals(ObjectType.TUPLE, reader.column(1).type());
    assertSame(reader.column(1).tuple(), reader.tuple(1));
    ScalarReader aReader = reader.column(0).scalar();
    TupleReader mReader = reader.column(1).tuple();
    assertEquals(ObjectType.SCALAR, mReader.column("b").array().entryType());
    ScalarElementReader bReader = mReader.column(0).elements();
    assertEquals(ValueType.INTEGER, bReader.valueType());
    assertTrue(reader.next());
    assertEquals(10, aReader.getInt());
    assertEquals(11, bReader.getInt(0));
    assertEquals(12, bReader.getInt(1));
    assertTrue(reader.next());
    assertEquals(20, aReader.getInt());
    assertEquals(21, bReader.getInt(0));
    assertEquals(22, bReader.getInt(1));
    assertTrue(reader.next());
    assertEquals(30, aReader.getInt());
    assertEquals(31, bReader.getInt(0));
    assertEquals(32, bReader.getInt(1));
    assertFalse(reader.next());
    // Verify that the map accessor's value count was set.
    @SuppressWarnings("resource") MapVector mapVector = (MapVector) actual.container().getValueVector(1).getValueVector();
    assertEquals(actual.rowCount(), mapVector.getAccessor().getValueCount());
    SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(10, objArray(intArray(11, 12))).addRow(20, objArray(intArray(21, 22))).addRow(30, objArray(intArray(31, 32))).build();
    new RowSetComparison(expected).verifyAndClearAll(actual);
}
Also used : RowSetWriter(org.apache.drill.test.rowSet.RowSetWriter) TupleReader(org.apache.drill.exec.vector.accessor.TupleReader) SingleRowSet(org.apache.drill.test.rowSet.RowSet.SingleRowSet) ScalarElementReader(org.apache.drill.exec.vector.accessor.ScalarElementReader) ScalarReader(org.apache.drill.exec.vector.accessor.ScalarReader) RowSetComparison(org.apache.drill.test.rowSet.RowSetComparison) TupleWriter(org.apache.drill.exec.vector.accessor.TupleWriter) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.test.rowSet.schema.SchemaBuilder) RowSetReader(org.apache.drill.test.rowSet.RowSetReader) ScalarWriter(org.apache.drill.exec.vector.accessor.ScalarWriter) ExtendableRowSet(org.apache.drill.test.rowSet.RowSet.ExtendableRowSet) RepeatedMapVector(org.apache.drill.exec.vector.complex.RepeatedMapVector) MapVector(org.apache.drill.exec.vector.complex.MapVector) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 8 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by axbaretto.

the class RowSetTest method testRepeatedMapStructure.

@Test
public void testRepeatedMapStructure() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMapArray("m").add("b", MinorType.INT).add("c", MinorType.INT).resumeSchema().buildSchema();
    ExtendableRowSet rowSet = fixture.rowSet(schema);
    RowSetWriter writer = rowSet.writer();
    // Map and Int
    // Pick out components and lightly test. (Assumes structure
    // tested earlier is still valid, so no need to exhaustively
    // test again.)
    assertEquals(ObjectType.SCALAR, writer.column("a").type());
    assertEquals(ObjectType.ARRAY, writer.column("m").type());
    ArrayWriter maWriter = writer.column(1).array();
    assertEquals(ObjectType.TUPLE, maWriter.entryType());
    TupleWriter mapWriter = maWriter.tuple();
    assertEquals(ObjectType.SCALAR, mapWriter.column("b").type());
    assertEquals(ObjectType.SCALAR, mapWriter.column("c").type());
    ScalarWriter aWriter = writer.column("a").scalar();
    ScalarWriter bWriter = mapWriter.scalar("b");
    ScalarWriter cWriter = mapWriter.scalar("c");
    assertEquals(ValueType.INTEGER, aWriter.valueType());
    assertEquals(ValueType.INTEGER, bWriter.valueType());
    assertEquals(ValueType.INTEGER, cWriter.valueType());
    // Write data
    aWriter.setInt(10);
    bWriter.setInt(101);
    cWriter.setInt(102);
    // Advance to next array position
    maWriter.save();
    bWriter.setInt(111);
    cWriter.setInt(112);
    maWriter.save();
    writer.save();
    aWriter.setInt(20);
    bWriter.setInt(201);
    cWriter.setInt(202);
    maWriter.save();
    bWriter.setInt(211);
    cWriter.setInt(212);
    maWriter.save();
    writer.save();
    aWriter.setInt(30);
    bWriter.setInt(301);
    cWriter.setInt(302);
    maWriter.save();
    bWriter.setInt(311);
    cWriter.setInt(312);
    maWriter.save();
    writer.save();
    // Finish the row set and get a reader.
    SingleRowSet actual = writer.done();
    RowSetReader reader = actual.reader();
    // Verify reader structure
    assertEquals(ObjectType.SCALAR, reader.column("a").type());
    assertEquals(ObjectType.ARRAY, reader.column("m").type());
    ArrayReader maReader = reader.column(1).array();
    assertEquals(ObjectType.TUPLE, maReader.entryType());
    TupleReader mapReader = maReader.tuple();
    assertEquals(ObjectType.SCALAR, mapReader.column("b").type());
    assertEquals(ObjectType.SCALAR, mapReader.column("c").type());
    ScalarReader aReader = reader.column("a").scalar();
    ScalarReader bReader = mapReader.scalar("b");
    ScalarReader cReader = mapReader.scalar("c");
    assertEquals(ValueType.INTEGER, aReader.valueType());
    assertEquals(ValueType.INTEGER, bReader.valueType());
    assertEquals(ValueType.INTEGER, cReader.valueType());
    // Row 1: use index accessors
    assertTrue(reader.next());
    assertEquals(10, aReader.getInt());
    TupleReader ixReader = maReader.tuple(0);
    assertEquals(101, ixReader.scalar(0).getInt());
    assertEquals(102, ixReader.scalar(1).getInt());
    ixReader = maReader.tuple(1);
    assertEquals(111, ixReader.scalar(0).getInt());
    assertEquals(112, ixReader.scalar(1).getInt());
    // Row 2: use common accessor with explicit positioning,
    // but access scalars through the map reader.
    assertTrue(reader.next());
    assertEquals(20, aReader.getInt());
    maReader.setPosn(0);
    assertEquals(201, mapReader.scalar(0).getInt());
    assertEquals(202, mapReader.scalar(1).getInt());
    maReader.setPosn(1);
    assertEquals(211, mapReader.scalar(0).getInt());
    assertEquals(212, mapReader.scalar(1).getInt());
    // Row 3: use common accessor for scalars
    assertTrue(reader.next());
    assertEquals(30, aReader.getInt());
    maReader.setPosn(0);
    assertEquals(301, bReader.getInt());
    assertEquals(302, cReader.getInt());
    maReader.setPosn(1);
    assertEquals(311, bReader.getInt());
    assertEquals(312, cReader.getInt());
    assertFalse(reader.next());
    // Verify that the map accessor's value count was set.
    @SuppressWarnings("resource") RepeatedMapVector mapVector = (RepeatedMapVector) actual.container().getValueVector(1).getValueVector();
    assertEquals(3, mapVector.getAccessor().getValueCount());
    // Verify the readers and writers again using the testing tools.
    SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(10, objArray(objArray(101, 102), objArray(111, 112))).addRow(20, objArray(objArray(201, 202), objArray(211, 212))).addRow(30, objArray(objArray(301, 302), objArray(311, 312))).build();
    new RowSetComparison(expected).verifyAndClearAll(actual);
}
Also used : RowSetWriter(org.apache.drill.test.rowSet.RowSetWriter) TupleReader(org.apache.drill.exec.vector.accessor.TupleReader) SingleRowSet(org.apache.drill.test.rowSet.RowSet.SingleRowSet) RepeatedMapVector(org.apache.drill.exec.vector.complex.RepeatedMapVector) ScalarReader(org.apache.drill.exec.vector.accessor.ScalarReader) ArrayReader(org.apache.drill.exec.vector.accessor.ArrayReader) RowSetComparison(org.apache.drill.test.rowSet.RowSetComparison) TupleWriter(org.apache.drill.exec.vector.accessor.TupleWriter) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.test.rowSet.schema.SchemaBuilder) ArrayWriter(org.apache.drill.exec.vector.accessor.ArrayWriter) RowSetReader(org.apache.drill.test.rowSet.RowSetReader) ScalarWriter(org.apache.drill.exec.vector.accessor.ScalarWriter) ExtendableRowSet(org.apache.drill.test.rowSet.RowSet.ExtendableRowSet) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 9 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by axbaretto.

the class TestFillEmpties method doFillEmptiesScalar.

private void doFillEmptiesScalar(MajorType majorType) {
    TupleMetadata schema = new SchemaBuilder().add("a", majorType).buildSchema();
    ExtendableRowSet rs = fixture.rowSet(schema);
    RowSetWriter writer = rs.writer();
    ScalarWriter colWriter = writer.scalar(0);
    ValueType valueType = colWriter.valueType();
    boolean nullable = majorType.getMode() == DataMode.OPTIONAL;
    for (int i = 0; i < ROW_COUNT; i++) {
        if (i % 5 == 0) {
            colWriter.setObject(RowSetUtilities.testDataFromInt(valueType, majorType, i));
        }
        writer.save();
    }
    SingleRowSet result = writer.done();
    RowSetReader reader = result.reader();
    ScalarReader colReader = reader.scalar(0);
    MinorType type = majorType.getMinorType();
    boolean isVariable = (type == MinorType.VARCHAR || type == MinorType.VAR16CHAR || type == MinorType.VARBINARY);
    for (int i = 0; i < ROW_COUNT; i++) {
        assertTrue(reader.next());
        if (i % 5 != 0) {
            if (nullable) {
                // Nullable types fill with nulls.
                assertTrue(colReader.isNull());
                continue;
            }
            if (isVariable) {
                // Variable width types fill with a zero-length value.
                assertEquals(0, colReader.getBytes().length);
                continue;
            }
        }
        // All other types fill with zero-bytes, interpreted as some form
        // of zero for each type.
        Object actual = colReader.getObject();
        Object expected = RowSetUtilities.testDataFromInt(valueType, majorType, i % 5 == 0 ? i : 0);
        RowSetUtilities.assertEqualValues(majorType.toString().replace('\n', ' ') + "[" + i + "]", valueType, expected, actual);
    }
    result.clear();
}
Also used : RowSetWriter(org.apache.drill.test.rowSet.RowSetWriter) SingleRowSet(org.apache.drill.test.rowSet.RowSet.SingleRowSet) ValueType(org.apache.drill.exec.vector.accessor.ValueType) ScalarReader(org.apache.drill.exec.vector.accessor.ScalarReader) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.test.rowSet.schema.SchemaBuilder) MinorType(org.apache.drill.common.types.TypeProtos.MinorType) RowSetReader(org.apache.drill.test.rowSet.RowSetReader) ScalarWriter(org.apache.drill.exec.vector.accessor.ScalarWriter) ExtendableRowSet(org.apache.drill.test.rowSet.RowSet.ExtendableRowSet)

Example 10 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by apache.

the class TestResultSetLoaderRepeatedList method test2DOverflow.

@Test
public void test2DOverflow() {
    final TupleMetadata schema = new SchemaBuilder().add("id", MinorType.INT).addRepeatedList("list2").addArray(MinorType.VARCHAR).resumeSchema().buildSchema();
    final ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().rowCountLimit(ValueVector.MAX_ROW_COUNT).readerSchema(schema).build();
    final ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    final RowSetLoader writer = rsLoader.writer();
    // Fill the batch with enough data to cause overflow.
    // Data must be large enough to cause overflow before 64K rows
    // Make a bit bigger to overflow early.
    final int outerSize = 7;
    final int innerSize = 5;
    final int strLength = ValueVector.MAX_BUFFER_SIZE / ValueVector.MAX_ROW_COUNT / outerSize / innerSize + 20;
    final byte[] value = new byte[strLength - 6];
    Arrays.fill(value, (byte) 'X');
    final String strValue = new String(value, Charsets.UTF_8);
    int rowCount = 0;
    int elementCount = 0;
    final ArrayWriter outerWriter = writer.array(1);
    final ArrayWriter innerWriter = outerWriter.array();
    final ScalarWriter elementWriter = innerWriter.scalar();
    rsLoader.startBatch();
    while (!writer.isFull()) {
        writer.start();
        writer.scalar(0).setInt(rowCount);
        for (int j = 0; j < outerSize; j++) {
            for (int k = 0; k < innerSize; k++) {
                elementWriter.setString(String.format("%s%06d", strValue, elementCount));
                elementCount++;
            }
            outerWriter.save();
        }
        writer.save();
        rowCount++;
    }
    // Number of rows should be driven by vector size.
    // Our row count should include the overflow row
    final int expectedCount = ValueVector.MAX_BUFFER_SIZE / (strLength * innerSize * outerSize);
    assertEquals(expectedCount + 1, rowCount);
    // Loader's row count should include only "visible" rows
    assertEquals(expectedCount, writer.rowCount());
    // Total count should include invisible and look-ahead rows.
    assertEquals(expectedCount + 1, rsLoader.totalRowCount());
    // Result should exclude the overflow row
    RowSet result = fixture.wrap(rsLoader.harvest());
    assertEquals(expectedCount, result.rowCount());
    // Verify the data.
    RowSetReader reader = result.reader();
    ArrayReader outerReader = reader.array(1);
    ArrayReader innerReader = outerReader.array();
    ScalarReader strReader = innerReader.scalar();
    int readRowCount = 0;
    int readElementCount = 0;
    while (reader.next()) {
        assertEquals(readRowCount, reader.scalar(0).getInt());
        for (int i = 0; i < outerSize; i++) {
            assertTrue(outerReader.next());
            for (int j = 0; j < innerSize; j++) {
                assertTrue(innerReader.next());
                assertEquals(String.format("%s%06d", strValue, readElementCount), strReader.getString());
                readElementCount++;
            }
            assertFalse(innerReader.next());
        }
        assertFalse(outerReader.next());
        readRowCount++;
    }
    assertEquals(readRowCount, result.rowCount());
    result.clear();
    // Write a few more rows to verify the overflow row.
    rsLoader.startBatch();
    for (int i = 0; i < 1000; i++) {
        writer.start();
        writer.scalar(0).setInt(rowCount);
        for (int j = 0; j < outerSize; j++) {
            for (int k = 0; k < innerSize; k++) {
                elementWriter.setString(String.format("%s%06d", strValue, elementCount));
                elementCount++;
            }
            outerWriter.save();
        }
        writer.save();
        rowCount++;
    }
    result = fixture.wrap(rsLoader.harvest());
    assertEquals(1001, result.rowCount());
    final int startCount = readRowCount;
    reader = result.reader();
    outerReader = reader.array(1);
    innerReader = outerReader.array();
    strReader = innerReader.scalar();
    while (reader.next()) {
        assertEquals(readRowCount, reader.scalar(0).getInt());
        for (int i = 0; i < outerSize; i++) {
            assertTrue(outerReader.next());
            for (int j = 0; j < innerSize; j++) {
                assertTrue(innerReader.next());
                elementWriter.setString(String.format("%s%06d", strValue, readElementCount));
                assertEquals(String.format("%s%06d", strValue, readElementCount), strReader.getString());
                readElementCount++;
            }
            assertFalse(innerReader.next());
        }
        assertFalse(outerReader.next());
        readRowCount++;
    }
    assertEquals(readRowCount - startCount, result.rowCount());
    result.clear();
    rsLoader.close();
}
Also used : SingleRowSet(org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet) RowSet(org.apache.drill.exec.physical.rowSet.RowSet) ScalarReader(org.apache.drill.exec.vector.accessor.ScalarReader) ArrayReader(org.apache.drill.exec.vector.accessor.ArrayReader) ResultSetLoader(org.apache.drill.exec.physical.resultSet.ResultSetLoader) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.exec.record.metadata.SchemaBuilder) RowSetLoader(org.apache.drill.exec.physical.resultSet.RowSetLoader) ArrayWriter(org.apache.drill.exec.vector.accessor.ArrayWriter) RowSetReader(org.apache.drill.exec.physical.rowSet.RowSetReader) ScalarWriter(org.apache.drill.exec.vector.accessor.ScalarWriter) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Aggregations

ScalarWriter (org.apache.drill.exec.vector.accessor.ScalarWriter)120 TupleMetadata (org.apache.drill.exec.record.metadata.TupleMetadata)69 SubOperatorTest (org.apache.drill.test.SubOperatorTest)68 Test (org.junit.Test)68 SchemaBuilder (org.apache.drill.exec.record.metadata.SchemaBuilder)51 SingleRowSet (org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet)44 ScalarReader (org.apache.drill.exec.vector.accessor.ScalarReader)31 ArrayWriter (org.apache.drill.exec.vector.accessor.ArrayWriter)26 RowSetLoader (org.apache.drill.exec.physical.resultSet.RowSetLoader)25 ResultSetLoader (org.apache.drill.exec.physical.resultSet.ResultSetLoader)24 TupleWriter (org.apache.drill.exec.vector.accessor.TupleWriter)23 ArrayReader (org.apache.drill.exec.vector.accessor.ArrayReader)22 RowSet (org.apache.drill.exec.physical.rowSet.RowSet)21 ExtendableRowSet (org.apache.drill.exec.physical.rowSet.RowSet.ExtendableRowSet)19 SchemaBuilder (org.apache.drill.test.rowSet.schema.SchemaBuilder)18 ColumnMetadata (org.apache.drill.exec.record.metadata.ColumnMetadata)17 TupleReader (org.apache.drill.exec.vector.accessor.TupleReader)17 SingleRowSet (org.apache.drill.test.rowSet.RowSet.SingleRowSet)14 RowSetReader (org.apache.drill.test.rowSet.RowSetReader)14 ResultSetLoader (org.apache.drill.exec.physical.rowSet.ResultSetLoader)13