Examples with RepeatedMapVector org.apache.drill.exec.vector.complex.RepeatedMapVector used on opensource projects

Example 6 with RepeatedMapVector

use of org.apache.drill.exec.vector.complex.RepeatedMapVector in project drill by apache.

the class TestResultSetLoaderMapArray method testBasics.

@Test
public void testBasics() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMapArray("m").add("c", MinorType.INT).add("d", MinorType.VARCHAR).resumeSchema().buildSchema();
    ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().readerSchema(schema).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();
    // Verify structure and schema
    TupleMetadata actualSchema = rootWriter.tupleSchema();
    assertEquals(2, actualSchema.size());
    assertTrue(actualSchema.metadata(1).isArray());
    assertTrue(actualSchema.metadata(1).isMap());
    assertEquals(2, actualSchema.metadata("m").tupleSchema().size());
    assertEquals(2, actualSchema.column("m").getChildren().size());
    TupleWriter mapWriter = rootWriter.array("m").tuple();
    assertSame(actualSchema.metadata("m").tupleSchema(), mapWriter.schema().tupleSchema());
    assertSame(mapWriter.tupleSchema(), mapWriter.schema().tupleSchema());
    assertSame(mapWriter.tupleSchema().metadata(0), mapWriter.scalar(0).schema());
    assertSame(mapWriter.tupleSchema().metadata(1), mapWriter.scalar(1).schema());
    // Write a couple of rows with arrays.
    rsLoader.startBatch();
    rootWriter.addRow(10, mapArray(mapValue(110, "d1.1"), mapValue(120, "d2.2"))).addRow(20, mapArray()).addRow(30, mapArray(mapValue(310, "d3.1"), mapValue(320, "d3.2"), mapValue(330, "d3.3")));
    // Verify the first batch
    RowSet actual = fixture.wrap(rsLoader.harvest());
    RepeatedMapVector mapVector = (RepeatedMapVector) actual.container().getValueVector(1).getValueVector();
    MaterializedField mapField = mapVector.getField();
    assertEquals(2, mapField.getChildren().size());
    Iterator<MaterializedField> iter = mapField.getChildren().iterator();
    assertTrue(mapWriter.scalar(0).schema().schema().isEquivalent(iter.next()));
    assertTrue(mapWriter.scalar(1).schema().schema().isEquivalent(iter.next()));
    SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(10, mapArray(mapValue(110, "d1.1"), mapValue(120, "d2.2"))).addRow(20, mapArray()).addRow(30, mapArray(mapValue(310, "d3.1"), mapValue(320, "d3.2"), mapValue(330, "d3.3"))).build();
    RowSetUtilities.verify(expected, actual);
    // In the second, create a row, then add a map member.
    // Should be back-filled to empty for the first row.
    rsLoader.startBatch();
    rootWriter.addRow(40, mapArray(mapValue(410, "d4.1"), mapValue(420, "d4.2")));
    mapWriter.addColumn(SchemaBuilder.columnSchema("e", MinorType.VARCHAR, DataMode.OPTIONAL));
    rootWriter.addRow(50, mapArray(mapValue(510, "d5.1", "e5.1"), mapValue(520, "d5.2", null))).addRow(60, mapArray(mapValue(610, "d6.1", "e6.1"), mapValue(620, "d6.2", null), mapValue(630, "d6.3", "e6.3")));
    // Verify the second batch
    actual = fixture.wrap(rsLoader.harvest());
    mapVector = (RepeatedMapVector) actual.container().getValueVector(1).getValueVector();
    mapField = mapVector.getField();
    assertEquals(3, mapField.getChildren().size());
    TupleMetadata expectedSchema = new SchemaBuilder().add("a", MinorType.INT).addMapArray("m").add("c", MinorType.INT).add("d", MinorType.VARCHAR).addNullable("e", MinorType.VARCHAR).resumeSchema().buildSchema();
    expected = fixture.rowSetBuilder(expectedSchema).addRow(40, mapArray(mapValue(410, "d4.1", null), mapValue(420, "d4.2", null))).addRow(50, mapArray(mapValue(510, "d5.1", "e5.1"), mapValue(520, "d5.2", null))).addRow(60, mapArray(mapValue(610, "d6.1", "e6.1"), mapValue(620, "d6.2", null), mapValue(630, "d6.3", "e6.3"))).build();
    RowSetUtilities.verify(expected, actual);
    rsLoader.close();
}

Example 7 with RepeatedMapVector

use of org.apache.drill.exec.vector.complex.RepeatedMapVector in project drill by apache.

the class FlattenRecordBatch method getFlattenFieldTransferPair.

/**
 * The data layout is the same for the actual data within a repeated field, as
 * it is in a scalar vector for the same sql type. For example, a repeated int
 * vector has a vector of offsets into a regular int vector to represent the
 * lists. As the data layout for the actual values in the same in the repeated
 * vector as in the scalar vector of the same type, we can avoid making
 * individual copies for the column being flattened, and just use vector
 * copies between the inner vector of the repeated field to the resulting
 * scalar vector from the flatten operation. This is completed after we
 * determine how many records will fit (as we will hit either a batch end, or
 * the end of one of the other vectors while we are copying the data of the
 * other vectors alongside each new flattened value coming out of the repeated
 * field.)
 */
private TransferPair getFlattenFieldTransferPair(FieldReference reference) {
    TypedFieldId fieldId = incoming.getValueVectorId(popConfig.getColumn());
    Class<?> vectorClass = incoming.getSchema().getColumn(fieldId.getFieldIds()[0]).getValueClass();
    ValueVector flattenField = incoming.getValueAccessorById(vectorClass, fieldId.getFieldIds()).getValueVector();
    TransferPair tp = null;
    if (flattenField instanceof AbstractRepeatedMapVector) {
        tp = ((AbstractRepeatedMapVector) flattenField).getTransferPairToSingleMap(reference.getAsNamePart().getName(), oContext.getAllocator());
    } else if (!(flattenField instanceof RepeatedValueVector)) {
        if (incoming.getRecordCount() != 0) {
            throw UserException.unsupportedError().message("Flatten does not support inputs of non-list values.").build(logger);
        }
        logger.error("Cannot cast {} to RepeatedValueVector", flattenField);
        // when incoming recordCount is 0, don't throw exception since the type being seen here is not solid
        ValueVector vv = new RepeatedMapVector(flattenField.getField(), oContext.getAllocator(), null);
        tp = RepeatedValueVector.class.cast(vv).getTransferPair(reference.getAsNamePart().getName(), oContext.getAllocator());
    } else {
        ValueVector vvIn = RepeatedValueVector.class.cast(flattenField).getDataVector();
        // vvIn may be null because of fast schema return for repeated list vectors
        if (vvIn != null) {
            tp = vvIn.getTransferPair(reference.getAsNamePart().getName(), oContext.getAllocator());
        }
    }
    return tp;
}

Example 8 with RepeatedMapVector

use of org.apache.drill.exec.vector.complex.RepeatedMapVector in project drill by apache.

the class FlattenRecordBatch method setFlattenVector.

private void setFlattenVector() {
    TypedFieldId typedFieldId = incoming.getValueVectorId(popConfig.getColumn());
    MaterializedField field = incoming.getSchema().getColumn(typedFieldId.getFieldIds()[0]);
    RepeatedValueVector vector;
    ValueVector inVV = incoming.getValueAccessorById(field.getValueClass(), typedFieldId.getFieldIds()).getValueVector();
    if (!(inVV instanceof RepeatedValueVector)) {
        if (incoming.getRecordCount() != 0) {
            throw UserException.unsupportedError().message("Flatten does not support inputs of non-list values.").build(logger);
        }
        // when incoming recordCount is 0, don't throw exception since the type being seen here is not solid
        logger.error("setFlattenVector cast failed and recordcount is 0, create empty vector anyway.");
        vector = new RepeatedMapVector(field, oContext.getAllocator(), null);
    } else {
        vector = RepeatedValueVector.class.cast(inVV);
    }
    flattener.setFlattenField(vector);
}

Example 9 with RepeatedMapVector

use of org.apache.drill.exec.vector.complex.RepeatedMapVector in project drill by apache.

the class TestValueVector method testVectors.

/**
 * Convenience method that allows running tests on various {@link ValueVector vector} instances.
 *
 * @param test test function to execute
 */
private void testVectors(VectorVerifier test) throws Exception {
    final MaterializedField[] fields = { MaterializedField.create(EMPTY_SCHEMA_PATH, UInt4Holder.TYPE), MaterializedField.create(EMPTY_SCHEMA_PATH, BitHolder.TYPE), MaterializedField.create(EMPTY_SCHEMA_PATH, VarCharHolder.TYPE), MaterializedField.create(EMPTY_SCHEMA_PATH, NullableVarCharHolder.TYPE), MaterializedField.create(EMPTY_SCHEMA_PATH, RepeatedListVector.TYPE), MaterializedField.create(EMPTY_SCHEMA_PATH, MapVector.TYPE), MaterializedField.create(EMPTY_SCHEMA_PATH, RepeatedMapVector.TYPE) };
    final ValueVector[] vectors = { new UInt4Vector(fields[0], allocator), new BitVector(fields[1], allocator), new VarCharVector(fields[2], allocator), new NullableVarCharVector(fields[3], allocator), new RepeatedListVector(fields[4], allocator, null), new MapVector(fields[5], allocator, null), new RepeatedMapVector(fields[6], allocator, null) };
    try {
        for (final ValueVector vector : vectors) {
            test.verify(vector);
        }
    } finally {
        AutoCloseables.close(vectors);
    }
}

Example 10 with RepeatedMapVector

use of org.apache.drill.exec.vector.complex.RepeatedMapVector in project drill by apache.

the class TestRecordBatchSizer method testEmptyBatchRepeatedMap.

@Test
public void testEmptyBatchRepeatedMap() {
    TupleMetadata schema = new SchemaBuilder().addMapArray("map").add("key", MinorType.INT).add("value", MinorType.VARCHAR).resumeSchema().buildSchema();
    RowSetBuilder builder = fixture.rowSetBuilder(schema);
    RowSet rows = builder.build();
    // Run the record batch sizer on the resulting batch.
    RecordBatchSizer sizer = new RecordBatchSizer(rows.container());
    assertEquals(1, sizer.columns().size());
    /**
     * stdDataSize:50+4, stdNetSize:50+4+4+4, dataSizePerEntry:0,
     * netSizePerEntry: 0,
     * totalDataSize:0, totalNetSize:0,
     * valueCount:0,
     * elementCount:0, cardinality:0, isVariableWidth:true
     */
    verifyColumnValues(sizer.columns().get("map"), 54, 62, 0, 0, 0, 0, 0, 0, 0, false);
    // Verify memory allocation is done correctly based on std size for empty batch.
    SingleRowSet empty = fixture.rowSet(schema);
    VectorAccessible accessible = empty.vectorAccessible();
    UInt4Vector offsetVector;
    for (VectorWrapper<?> vw : accessible) {
        ValueVector v = vw.getValueVector();
        RecordBatchSizer.ColumnSize colSize = sizer.getColumn(v.getField().getName());
        // Allocates to nearest power of two
        colSize.allocateVector(v, testRowCount);
        RepeatedMapVector mapVector = (RepeatedMapVector) v;
        offsetVector = ((RepeatedValueVector) mapVector).getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCount) << 1), offsetVector.getValueCapacity());
        ValueVector keyVector = mapVector.getChild("key");
        ValueVector valueVector1 = mapVector.getChild("value");
        assertEquals(((Integer.highestOneBit(testRowCount * STD_REPETITION_FACTOR) << 1)), keyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCount * STD_REPETITION_FACTOR) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(testRowCount * STD_REPETITION_FACTOR << 1) - 1, valueVector1.getValueCapacity());
        // Allocates the same as value passed since it is already power of two.
        colSize.allocateVector(v, testRowCountPowerTwo - 1);
        mapVector = (RepeatedMapVector) v;
        offsetVector = ((RepeatedValueVector) mapVector).getOffsetVector();
        assertEquals(testRowCountPowerTwo, offsetVector.getValueCapacity());
        keyVector = mapVector.getChild("key");
        valueVector1 = mapVector.getChild("value");
        assertEquals(Integer.highestOneBit(testRowCountPowerTwo * STD_REPETITION_FACTOR) << 1, keyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(Integer.highestOneBit((int) (testRowCountPowerTwo * STD_REPETITION_FACTOR)) << 1, offsetVector.getValueCapacity());
        assertEquals((Integer.highestOneBit(testRowCountPowerTwo * STD_REPETITION_FACTOR << 1)) - 1, valueVector1.getValueCapacity());
        // Allocate for max rows.
        colSize.allocateVector(v, ValueVector.MAX_ROW_COUNT - 1);
        mapVector = (RepeatedMapVector) v;
        offsetVector = ((RepeatedValueVector) mapVector).getOffsetVector();
        assertEquals(ValueVector.MAX_ROW_COUNT, offsetVector.getValueCapacity());
        keyVector = mapVector.getChild("key");
        valueVector1 = mapVector.getChild("value");
        assertEquals(Integer.highestOneBit(ValueVector.MAX_ROW_COUNT * STD_REPETITION_FACTOR) << 1, keyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(Integer.highestOneBit(ValueVector.MAX_ROW_COUNT * STD_REPETITION_FACTOR) << 1, offsetVector.getValueCapacity());
        assertEquals((Integer.highestOneBit(ValueVector.MAX_ROW_COUNT * STD_REPETITION_FACTOR) << 1) - 1, valueVector1.getValueCapacity());
        // Allocate for 0 rows. should atleast do allocation for 1 row.
        colSize.allocateVector(v, 0);
        mapVector = (RepeatedMapVector) v;
        offsetVector = ((RepeatedValueVector) mapVector).getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT, offsetVector.getValueCapacity());
        keyVector = mapVector.getChild("key");
        valueVector1 = mapVector.getChild("value");
        assertEquals(ValueVector.MIN_ROW_COUNT, keyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT + 1, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MIN_ROW_COUNT, valueVector1.getValueCapacity());
        v.clear();
    }
    empty.clear();
    rows.clear();
}