Examples with BatchSchema org.apache.drill.exec.record.BatchSchema used on opensource projects

Example 11 with BatchSchema

use of org.apache.drill.exec.record.BatchSchema in project drill by apache.

the class RowSetTest method TestTopScalarArray.

@Test
public void TestTopScalarArray() {
    BatchSchema batchSchema = new SchemaBuilder().add("c", MinorType.INT).addArray("a", MinorType.INT).build();
    ExtendableRowSet rs1 = fixture.rowSet(batchSchema);
    RowSetWriter writer = rs1.writer();
    writer.column(0).setInt(10);
    ArrayWriter array = writer.column(1).array();
    array.setInt(100);
    array.setInt(110);
    writer.save();
    writer.column(0).setInt(20);
    array = writer.column(1).array();
    array.setInt(200);
    array.setInt(120);
    array.setInt(220);
    writer.save();
    writer.column(0).setInt(30);
    writer.save();
    writer.done();
    RowSetReader reader = rs1.reader();
    assertTrue(reader.next());
    assertEquals(10, reader.column(0).getInt());
    ArrayReader arrayReader = reader.column(1).array();
    assertEquals(2, arrayReader.size());
    assertEquals(100, arrayReader.getInt(0));
    assertEquals(110, arrayReader.getInt(1));
    assertTrue(reader.next());
    assertEquals(20, reader.column(0).getInt());
    arrayReader = reader.column(1).array();
    assertEquals(3, arrayReader.size());
    assertEquals(200, arrayReader.getInt(0));
    assertEquals(120, arrayReader.getInt(1));
    assertEquals(220, arrayReader.getInt(2));
    assertTrue(reader.next());
    assertEquals(30, reader.column(0).getInt());
    arrayReader = reader.column(1).array();
    assertEquals(0, arrayReader.size());
    assertFalse(reader.next());
    SingleRowSet rs2 = fixture.rowSetBuilder(batchSchema).add(10, new int[] { 100, 110 }).add(20, new int[] { 200, 120, 220 }).add(30, null).build();
    new RowSetComparison(rs1).verifyAndClear(rs2);
}

Example 12 with BatchSchema

use of org.apache.drill.exec.record.BatchSchema in project drill by apache.

the class RowSetTest method testDoubleRW.

private void testDoubleRW() {
    BatchSchema batchSchema = new SchemaBuilder().add("col", MinorType.FLOAT8).build();
    SingleRowSet rs = fixture.rowSetBuilder(batchSchema).add(0D).add(Double.MAX_VALUE).add(Double.MIN_VALUE).build();
    RowSetReader reader = rs.reader();
    assertTrue(reader.next());
    assertEquals(0, reader.column(0).getDouble(), 0.000001);
    assertTrue(reader.next());
    assertEquals(Double.MAX_VALUE, reader.column(0).getDouble(), 0.000001);
    assertTrue(reader.next());
    assertEquals(Double.MIN_VALUE, reader.column(0).getDouble(), 0.000001);
    assertFalse(reader.next());
    rs.clear();
}

Example 13 with BatchSchema

use of org.apache.drill.exec.record.BatchSchema in project drill by apache.

the class RowSetTest method testMap.

@Test
public void testMap() {
    BatchSchema batchSchema = new SchemaBuilder().add("a", MinorType.INT).addMap("b").add("c", MinorType.INT).add("d", MinorType.INT).buildMap().build();
    SingleRowSet rs = fixture.rowSetBuilder(batchSchema).add(10, 20, 30).add(40, 50, 60).build();
    RowSetReader reader = rs.reader();
    assertTrue(reader.next());
    assertEquals(10, reader.column(0).getInt());
    assertEquals(20, reader.column(1).getInt());
    assertEquals(30, reader.column(2).getInt());
    assertEquals(10, reader.column("a").getInt());
    assertEquals(30, reader.column("b.d").getInt());
    assertTrue(reader.next());
    assertEquals(40, reader.column(0).getInt());
    assertEquals(50, reader.column(1).getInt());
    assertEquals(60, reader.column(2).getInt());
    assertFalse(reader.next());
    rs.clear();
}

Example 14 with BatchSchema

use of org.apache.drill.exec.record.BatchSchema in project drill by apache.

the class SortRecordBatchBuilder method build.

public void build(FragmentContext context, VectorContainer outputContainer) throws SchemaChangeException {
    outputContainer.clear();
    if (batches.keySet().size() > 1) {
        throw new SchemaChangeException("Sort currently only supports a single schema.");
    }
    if (batches.size() > Character.MAX_VALUE) {
        throw new SchemaChangeException("Sort cannot work on more than %d batches at a time.", (int) Character.MAX_VALUE);
    }
    if (batches.keys().size() < 1) {
        assert false : "Invalid to have an empty set of batches with no schemas.";
    }
    final DrillBuf svBuffer = reservation.allocateBuffer();
    if (svBuffer == null) {
        throw new OutOfMemoryError("Failed to allocate direct memory for SV4 vector in SortRecordBatchBuilder.");
    }
    sv4 = new SelectionVector4(svBuffer, recordCount, Character.MAX_VALUE);
    BatchSchema schema = batches.keySet().iterator().next();
    List<RecordBatchData> data = batches.get(schema);
    // now we're going to generate the sv4 pointers
    switch(schema.getSelectionVectorMode()) {
        case NONE:
            {
                int index = 0;
                int recordBatchId = 0;
                for (RecordBatchData d : data) {
                    for (int i = 0; i < d.getRecordCount(); i++, index++) {
                        sv4.set(index, recordBatchId, i);
                    }
                    recordBatchId++;
                }
                break;
            }
        case TWO_BYTE:
            {
                int index = 0;
                int recordBatchId = 0;
                for (RecordBatchData d : data) {
                    for (int i = 0; i < d.getRecordCount(); i++, index++) {
                        sv4.set(index, recordBatchId, (int) d.getSv2().getIndex(i));
                    }
                    // might as well drop the selection vector since we'll stop using it now.
                    d.getSv2().clear();
                    recordBatchId++;
                }
                break;
            }
        default:
            throw new UnsupportedOperationException();
    }
    // next, we'll create lists of each of the vector types.
    ArrayListMultimap<MaterializedField, ValueVector> vectors = ArrayListMultimap.create();
    for (RecordBatchData rbd : batches.values()) {
        for (ValueVector v : rbd.getVectors()) {
            vectors.put(v.getField(), v);
        }
    }
    for (MaterializedField f : schema) {
        List<ValueVector> v = vectors.get(f);
        outputContainer.addHyperList(v, false);
    }
    outputContainer.buildSchema(SelectionVectorMode.FOUR_BYTE);
}

Example 15 with BatchSchema

use of org.apache.drill.exec.record.BatchSchema in project drill by apache.

the class ProducerConsumerBatch method load.

private boolean load(final RecordBatchData batch) {
    final VectorContainer newContainer = batch.getContainer();
    if (schema != null && newContainer.getSchema().equals(schema)) {
        container.zeroVectors();
        final BatchSchema schema = container.getSchema();
        for (int i = 0; i < container.getNumberOfColumns(); i++) {
            final MaterializedField field = schema.getColumn(i);
            final MajorType type = field.getType();
            final ValueVector vOut = container.getValueAccessorById(TypeHelper.getValueVectorClass(type.getMinorType(), type.getMode()), container.getValueVectorId(SchemaPath.getSimplePath(field.getPath())).getFieldIds()).getValueVector();
            final ValueVector vIn = newContainer.getValueAccessorById(TypeHelper.getValueVectorClass(type.getMinorType(), type.getMode()), newContainer.getValueVectorId(SchemaPath.getSimplePath(field.getPath())).getFieldIds()).getValueVector();
            final TransferPair tp = vIn.makeTransferPair(vOut);
            tp.transfer();
        }
        return false;
    } else {
        container.clear();
        for (final VectorWrapper<?> w : newContainer) {
            container.add(w.getValueVector());
        }
        container.buildSchema(SelectionVectorMode.NONE);
        schema = container.getSchema();
        return true;
    }
}