Examples with SingleRowSet org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet used on opensource projects

Example 76 with SingleRowSet

use of org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet in project drill by apache.

the class TestRecordBatchSizer method testEmptyBatchNestedMap.

@Test
public void testEmptyBatchNestedMap() {
    TupleMetadata schema = new SchemaBuilder().addMap("map").add("key", MinorType.INT).add("value", MinorType.VARCHAR).addMap("childMap").add("childKey", MinorType.INT).add("childValue", MinorType.VARCHAR).resumeMap().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)*2, stdNetSize:(50+4)*2+4+4, dataSizePerEntry:0,
     * netSizePerEntry: 0,
     * totalDataSize:0, totalNetSize:0,
     * valueCount:0,
     * elementCount:0, cardinality:0, isVariableWidth:true
     */
    verifyColumnValues(sizer.columns().get("map"), 108, 116, 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);
        MapVector mapVector = (MapVector) v;
        ValueVector keyVector = mapVector.getChild("key");
        ValueVector valueVector1 = mapVector.getChild("value");
        assertEquals((Integer.highestOneBit(testRowCount) << 1), keyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCount) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(testRowCount << 1) - 1, valueVector1.getValueCapacity());
        MapVector childMapVector = (MapVector) mapVector.getChild("childMap");
        ValueVector childKeyVector = childMapVector.getChild("childKey");
        ValueVector childValueVector1 = childMapVector.getChild("childValue");
        assertEquals((Integer.highestOneBit(testRowCount) << 1), childKeyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCount) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(testRowCount << 1) - 1, childValueVector1.getValueCapacity());
        // Allocates the same as value passed since it is already power of two.
        colSize.allocateVector(v, testRowCountPowerTwo - 1);
        mapVector = (MapVector) v;
        keyVector = mapVector.getChild("key");
        valueVector1 = mapVector.getChild("value");
        assertEquals(testRowCountPowerTwo, keyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(testRowCountPowerTwo, offsetVector.getValueCapacity());
        assertEquals(testRowCountPowerTwo - 1, valueVector1.getValueCapacity());
        childMapVector = (MapVector) mapVector.getChild("childMap");
        childKeyVector = childMapVector.getChild("childKey");
        childValueVector1 = childMapVector.getChild("childValue");
        assertEquals(testRowCountPowerTwo, childKeyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(testRowCountPowerTwo, offsetVector.getValueCapacity());
        assertEquals(testRowCountPowerTwo - 1, childValueVector1.getValueCapacity());
        // Allocate for max rows.
        colSize.allocateVector(v, ValueVector.MAX_ROW_COUNT - 1);
        mapVector = (MapVector) v;
        keyVector = mapVector.getChild("key");
        valueVector1 = mapVector.getChild("value");
        assertEquals(ValueVector.MAX_ROW_COUNT, keyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(ValueVector.MAX_ROW_COUNT, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MAX_ROW_COUNT - 1, valueVector1.getValueCapacity());
        childMapVector = (MapVector) mapVector.getChild("childMap");
        childKeyVector = childMapVector.getChild("childKey");
        childValueVector1 = childMapVector.getChild("childValue");
        assertEquals(ValueVector.MAX_ROW_COUNT, childKeyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(ValueVector.MAX_ROW_COUNT, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MAX_ROW_COUNT - 1, childValueVector1.getValueCapacity());
        // Allocate for 0 rows. should atleast do allocation for 1 row.
        colSize.allocateVector(v, 0);
        mapVector = (MapVector) v;
        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());
        childMapVector = (MapVector) mapVector.getChild("childMap");
        childKeyVector = childMapVector.getChild("childKey");
        childValueVector1 = childMapVector.getChild("childValue");
        assertEquals(ValueVector.MIN_ROW_COUNT, childKeyVector.getValueCapacity());
        offsetVector = ((VariableWidthVector) valueVector1).getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT + 1, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MIN_ROW_COUNT, childValueVector1.getValueCapacity());
        v.clear();
    }
    empty.clear();
    rows.clear();
}

Example 77 with SingleRowSet

use of org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet in project drill by apache.

the class TestRecordBatchSizer method testSizerVariableWidth.

@Test
public void testSizerVariableWidth() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.VARCHAR).buildSchema();
    RowSetBuilder builder = fixture.rowSetBuilder(schema);
    StringBuilder stringBuilder = new StringBuilder();
    // a, aa, aaa, ... aaaaaaaaaa. totalSize = (10*11)/2 = 55
    for (long i = 0; i < 10; i++) {
        stringBuilder.append("a");
        builder.addRow(stringBuilder.toString());
    }
    RowSet rows = builder.build();
    // Run the record batch sizer on the resulting batch.
    RecordBatchSizer sizer = new RecordBatchSizer(rows.container());
    assertEquals(1, sizer.columns().size());
    ColumnSize aColumn = sizer.columns().get("a");
    /**
     * stdDataSize:50, stdNetSize:50+4, dataSizePerEntry:8, netSizePerEntry:8,
     * totalDataSize:(10*11)/2, totalNetSize:(10*11)/2 + 4*10, valueCount:10,
     * elementCount:10, cardinality:1, isVariableWidth:true
     */
    verifyColumnValues(aColumn, 50, 54, 6, 10, 55, 95, 10, 10, 1, true);
    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);
        offsetVector = ((VariableWidthVector) v).getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCount) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(testRowCount << 1) - 1, v.getValueCapacity());
        v.clear();
        // Allocates the same as value passed since it is already power of two.
        // -1 is done for adjustment needed for offset vector.
        colSize.allocateVector(v, testRowCountPowerTwo - 1);
        offsetVector = ((VariableWidthVector) v).getOffsetVector();
        assertEquals(testRowCountPowerTwo, offsetVector.getValueCapacity());
        assertEquals(testRowCountPowerTwo - 1, v.getValueCapacity());
        v.clear();
        // Allocate for max rows.
        colSize.allocateVector(v, ValueVector.MAX_ROW_COUNT - 1);
        offsetVector = ((VariableWidthVector) v).getOffsetVector();
        assertEquals(ValueVector.MAX_ROW_COUNT, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MAX_ROW_COUNT - 1, v.getValueCapacity());
        v.clear();
        // Allocate for 0 rows. should atleast do allocation for 1 row.
        colSize.allocateVector(v, 0);
        offsetVector = ((VariableWidthVector) v).getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT + 1, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MIN_ROW_COUNT, v.getValueCapacity());
        v.clear();
    }
    empty.clear();
    rows.clear();
}

Example 78 with SingleRowSet

use of org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet in project drill by apache.

the class TestRecordBatchSizer method testEmptyBatchRepeatedVariableWidth.

@Test
public void testEmptyBatchRepeatedVariableWidth() {
    TupleMetadata schema = new SchemaBuilder().addArray("b", MinorType.VARCHAR).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());
    ColumnSize bColumn = sizer.columns().get("b");
    /**
     * stdDataSize:50*5, stdNetSize:50*5+4*10+4, dataSizePerEntry:0, netSizePerEntry:0,
     * totalDataSize:0, totalNetSize:0, valueCount:0,
     * elementCount:0, cardinality:0, isVariableWidth:true
     */
    verifyColumnValues(bColumn, 250, 274, 0, 0, 0, 0, 0, 0, 0, true);
    // Verify memory allocation is done correctly based on std size for empty batch.
    SingleRowSet empty = fixture.rowSet(schema);
    VectorAccessible accessible = empty.vectorAccessible();
    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 - 1);
        UInt4Vector offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals(Integer.highestOneBit(testRowCount) << 1, offsetVector.getValueCapacity());
        VariableWidthVector vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals((Integer.highestOneBit((testRowCount - 1) * STD_REPETITION_FACTOR) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit((testRowCount - 1) * STD_REPETITION_FACTOR << 1) - 1, vwVector.getValueCapacity());
        v.clear();
        // Allocates the same as value passed since it is already power of two.
        colSize.allocateVector(v, testRowCountPowerTwo);
        offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCountPowerTwo) << 1), offsetVector.getValueCapacity());
        vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals((Integer.highestOneBit((int) (testRowCountPowerTwo * STD_REPETITION_FACTOR)) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(testRowCountPowerTwo * STD_REPETITION_FACTOR << 1) - 1, vwVector.getValueCapacity());
        v.clear();
        // Allocate for max rows.
        colSize.allocateVector(v, ValueVector.MAX_ROW_COUNT);
        offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals(ValueVector.MAX_ROW_COUNT << 1, offsetVector.getValueCapacity());
        vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals((Integer.highestOneBit((int) (ValueVector.MAX_ROW_COUNT * STD_REPETITION_FACTOR)) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(ValueVector.MAX_ROW_COUNT * STD_REPETITION_FACTOR << 1) - 1, vwVector.getValueCapacity());
        v.clear();
        // Allocate for 0 rows. should atleast do allocation for 1 row.
        colSize.allocateVector(v, 0);
        offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT + 1, offsetVector.getValueCapacity());
        vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT + 1, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MIN_ROW_COUNT, vwVector.getValueCapacity());
        v.clear();
    }
    empty.clear();
    rows.clear();
}

Example 79 with SingleRowSet

use of org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet in project drill by apache.

the class TestRecordBatchSizer method testEmptyBatchFixedWidth.

@Test
public void testEmptyBatchFixedWidth() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.BIGINT).add("b", MinorType.FLOAT8).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(2, sizer.columns().size());
    ColumnSize aColumn = sizer.columns().get("a");
    /**
     * stdDataSize:8, stdNetSize:8, dataSizePerEntry:0, netSizePerEntry:0,
     * totalDataSize:0, totalNetSize:0, valueCount:0,
     * elementCount:0, cardinality:0, isVariableWidth:false
     */
    verifyColumnValues(aColumn, 8, 8, 0, 0, 0, 0, 0, 0, 0, false);
    ColumnSize bColumn = sizer.columns().get("b");
    verifyColumnValues(bColumn, 8, 8, 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();
    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);
        assertEquals((Integer.highestOneBit(testRowCount) << 1), v.getValueCapacity());
        v.clear();
        // Allocates the same as value passed since it is already power of two.
        colSize.allocateVector(v, testRowCountPowerTwo);
        assertEquals(testRowCountPowerTwo, v.getValueCapacity());
        v.clear();
        // Allocate for max rows.
        colSize.allocateVector(v, ValueVector.MAX_ROW_COUNT);
        assertEquals(ValueVector.MAX_ROW_COUNT, v.getValueCapacity());
        v.clear();
        // Allocate for 0 rows. should atleast do allocation for 1 row.
        colSize.allocateVector(v, 0);
        assertEquals(ValueVector.MIN_ROW_COUNT, v.getValueCapacity());
        v.clear();
    }
    rows.clear();
    empty.clear();
}

Example 80 with SingleRowSet

use of org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet in project drill by apache.

the class TestRecordBatchSizer method testSizerRepeatedVariableWidth.

@Test
public void testSizerRepeatedVariableWidth() {
    TupleMetadata schema = new SchemaBuilder().addArray("b", MinorType.VARCHAR).buildSchema();
    RowSetBuilder builder = fixture.rowSetBuilder(schema);
    // size = (5*6)/2 = 15
    String[] newString = new String[] { "a", "aa", "aaa", "aaaa", "aaaaa" };
    for (long i = 0; i < 10; i++) {
        builder.addRow((Object) (newString));
    }
    RowSet rows = builder.build();
    // Run the record batch sizer on the resulting batch.
    RecordBatchSizer sizer = new RecordBatchSizer(rows.container());
    assertEquals(1, sizer.columns().size());
    ColumnSize bColumn = sizer.columns().get("b");
    /**
     * stdDataSize:50*5, stdNetSize:50*5+4*5+4, dataSizePerEntry:(5*6)/2, netSizePerEntry:(5*6)/2+5*4+4,
     * totalDataSize:(5*6)/2 * 10, totalNetSize: ((5*6)/2+5*4+4)*10, valueCount:10,
     * elementCount:50, cardinality:5, isVariableWidth:true
     */
    verifyColumnValues(bColumn, 250, 274, 15, 39, 150, 390, 10, 50, 5, true);
    SingleRowSet empty = fixture.rowSet(schema);
    VectorAccessible accessible = empty.vectorAccessible();
    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 - 1);
        UInt4Vector offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals(Integer.highestOneBit(testRowCount) << 1, offsetVector.getValueCapacity());
        VariableWidthVector vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals((Integer.highestOneBit((testRowCount - 1) * 5) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit((testRowCount - 1) * 5 << 1) - 1, vwVector.getValueCapacity());
        v.clear();
        // Allocates the same as value passed since it is already power of two.
        colSize.allocateVector(v, testRowCountPowerTwo);
        offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCountPowerTwo) << 1), offsetVector.getValueCapacity());
        vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals((Integer.highestOneBit(testRowCountPowerTwo * 5) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(testRowCountPowerTwo * 5 << 1) - 1, vwVector.getValueCapacity());
        v.clear();
        // Allocate for max rows.
        colSize.allocateVector(v, ValueVector.MAX_ROW_COUNT);
        offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals(ValueVector.MAX_ROW_COUNT << 1, offsetVector.getValueCapacity());
        vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals((Integer.highestOneBit(ValueVector.MAX_ROW_COUNT * 5) << 1), offsetVector.getValueCapacity());
        assertEquals(Integer.highestOneBit(ValueVector.MAX_ROW_COUNT * 5 << 1) - 1, vwVector.getValueCapacity());
        v.clear();
        // Allocate for 0 rows. should atleast do allocation for 1 row.
        colSize.allocateVector(v, 0);
        offsetVector = ((RepeatedValueVector) v).getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT + 1, offsetVector.getValueCapacity());
        vwVector = ((VariableWidthVector) ((RepeatedValueVector) v).getDataVector());
        offsetVector = vwVector.getOffsetVector();
        assertEquals(ValueVector.MIN_ROW_COUNT + 1, offsetVector.getValueCapacity());
        assertEquals(ValueVector.MIN_ROW_COUNT, vwVector.getValueCapacity());
        v.clear();
    }
    empty.clear();
    rows.clear();
}