Search in sources :

Example 86 with SingleRowSet

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

the class TestIndirectReaders method testArray.

/**
 * More complex case with two levels of offset vector (one for the
 * array, another for the Varchar values.) Only the top level goes
 * through the indirection.
 */
@Test
public void testArray() {
    TupleMetadata schema = new SchemaBuilder().addArray("a", MinorType.VARCHAR).buildSchema();
    ExtendableRowSet rowSet = fixture.rowSet(schema);
    RowSetWriter writer = rowSet.writer();
    ArrayWriter aWriter = writer.array(0);
    ScalarWriter strWriter = aWriter.scalar();
    for (int i = 0; i < 10; i++) {
        for (int j = 0; j < 5; j++) {
            strWriter.setString("value" + i + "." + j);
        }
        writer.save();
    }
    SingleRowSet result = writer.done().toIndirect();
    SelectionVector2 sv2 = result.getSv2();
    for (int i = 0; i < 10; i++) {
        sv2.setIndex(i, 9 - i);
    }
    RowSetReader reader = result.reader();
    ArrayReader aReader = reader.array(0);
    ScalarReader strReader = aReader.scalar();
    for (int i = 9; i >= 0; i--) {
        assertTrue(reader.next());
        for (int j = 0; j < 5; j++) {
            assertTrue(aReader.next());
            assertEquals("value" + i + "." + j, strReader.getString());
        }
    }
    result.clear();
}
Also used : ScalarReader(org.apache.drill.exec.vector.accessor.ScalarReader) ArrayReader(org.apache.drill.exec.vector.accessor.ArrayReader) SingleRowSet(org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.exec.record.metadata.SchemaBuilder) SelectionVector2(org.apache.drill.exec.record.selection.SelectionVector2) ArrayWriter(org.apache.drill.exec.vector.accessor.ArrayWriter) ScalarWriter(org.apache.drill.exec.vector.accessor.ScalarWriter) ExtendableRowSet(org.apache.drill.exec.physical.rowSet.RowSet.ExtendableRowSet) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 87 with SingleRowSet

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

the class TestRowSet method testScalarStructure.

/**
 * Test the simplest constructs: a row with top-level scalar
 * columns.
 * <p>
 * The focus here is the structure of the readers and writers, along
 * with the row set loader and verifier that use those constructs.
 * That is, while this test uses the int vector, this test is not
 * focused on that vector.
 */
@Test
public void testScalarStructure() {
    final TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).buildSchema();
    final ExtendableRowSet rowSet = fixture.rowSet(schema);
    final RowSetWriter writer = rowSet.writer();
    // Required Int
    // Verify the invariants of the "full" and "simple" access paths
    assertEquals(ObjectType.SCALAR, writer.column("a").type());
    assertSame(writer.column("a"), writer.column(0));
    assertSame(writer.scalar("a"), writer.scalar(0));
    assertSame(writer.column("a").scalar(), writer.scalar("a"));
    assertSame(writer.column(0).scalar(), writer.scalar(0));
    assertEquals(ValueType.INTEGER, writer.scalar(0).valueType());
    try {
        writer.column(0).array();
        fail();
    } catch (final UnsupportedOperationException e) {
    // Expected
    }
    try {
        writer.column(0).tuple();
        fail();
    } catch (final UnsupportedOperationException e) {
    // Expected
    }
    // Test the various ways to get at the scalar writer.
    writer.column("a").scalar().setInt(10);
    writer.save();
    writer.scalar("a").setInt(20);
    writer.save();
    writer.column(0).scalar().setInt(30);
    writer.save();
    writer.scalar(0).setInt(40);
    writer.save();
    // Finish the row set and get a reader.
    final SingleRowSet actual = writer.done();
    final RowSetReader reader = actual.reader();
    // Verify invariants
    assertEquals(ObjectType.SCALAR, reader.column(0).type());
    assertSame(reader.column("a"), reader.column(0));
    assertSame(reader.scalar("a"), reader.scalar(0));
    assertSame(reader.column("a").scalar(), reader.scalar("a"));
    assertSame(reader.column(0).scalar(), reader.scalar(0));
    assertEquals(ValueType.INTEGER, reader.scalar(0).valueType());
    assertTrue(schema.metadata("a").isEquivalent(reader.column("a").schema()));
    // Test various accessors: full and simple
    assertTrue(reader.next());
    assertFalse(reader.column("a").scalar().isNull());
    assertEquals(10, reader.column("a").scalar().getInt());
    assertTrue(reader.next());
    assertFalse(reader.scalar("a").isNull());
    assertEquals(20, reader.scalar("a").getInt());
    assertTrue(reader.next());
    assertFalse(reader.column(0).scalar().isNull());
    assertEquals(30, reader.column(0).scalar().getInt());
    assertTrue(reader.next());
    assertFalse(reader.column(0).scalar().isNull());
    assertEquals(40, reader.scalar(0).getInt());
    assertFalse(reader.next());
    // Test the above again via the writer and reader
    // utility classes.
    final SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(10).addRow(20).addRow(30).addRow(40).build();
    RowSetUtilities.verify(expected, actual);
}
Also used : SingleRowSet(org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.exec.record.metadata.SchemaBuilder) ExtendableRowSet(org.apache.drill.exec.physical.rowSet.RowSet.ExtendableRowSet) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 88 with SingleRowSet

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

the class TestRowSet method example.

/**
 * The code below is not a test. Rather, it is a simple example of
 * how to write a batch of data using writers, then read it using
 * readers.
 */
@Test
public void example() {
    // Step 1: Define a schema. In a real app, this
    // will be provided by a reader, by an incoming batch,
    // etc.
    final TupleMetadata schema = new SchemaBuilder().add("a", MinorType.VARCHAR).addArray("b", MinorType.INT).addMap("c").add("c1", MinorType.INT).add("c2", MinorType.VARCHAR).resumeSchema().buildSchema();
    // Step 2: Create a batch. Done here because this is
    // a batch-oriented test. Done automatically in the
    // result set loader.
    final DirectRowSet drs = DirectRowSet.fromSchema(fixture.allocator(), schema);
    // Step 3: Create the writer.
    final RowSetWriter writer = drs.writer();
    // Step 4: Populate data. Here we do it the way an app would:
    // using the individual accessors. See tests above for the many
    // ways this can be done depending on the need of the app.
    // 
    // Write two rows:
    // ("fred", [10, 11], {12, "wilma"})
    // ("barney", [20, 21], {22, "betty"})
    // 
    // This example uses Java strings for Varchar. Real code might
    // use byte arrays.
    writer.scalar("a").setString("fred");
    final ArrayWriter bWriter = writer.array("b");
    bWriter.scalar().setInt(10);
    bWriter.scalar().setInt(11);
    final TupleWriter cWriter = writer.tuple("c");
    cWriter.scalar("c1").setInt(12);
    cWriter.scalar("c2").setString("wilma");
    writer.save();
    writer.scalar("a").setString("barney");
    bWriter.scalar().setInt(20);
    bWriter.scalar().setInt(21);
    cWriter.scalar("c1").setInt(22);
    cWriter.scalar("c2").setString("betty");
    writer.save();
    // Step 5: "Harvest" the batch. Done differently in the
    // result set loader.
    final SingleRowSet rowSet = writer.done();
    // Step 5: Create a reader.
    final RowSetReader reader = rowSet.reader();
    while (reader.next()) {
        final StringBuilder sb = new StringBuilder();
        sb.append(print(reader.scalar("a").getString()));
        final ArrayReader bReader = reader.array("b");
        while (bReader.next()) {
            sb.append(print(bReader.scalar().getInt()));
        }
        final TupleReader cReader = reader.tuple("c");
        sb.append(print(cReader.scalar("c1").getInt()));
        sb.append(print(cReader.scalar("c2").getString()));
        logger.debug(sb.toString());
    }
    // Step 7: Free memory.
    rowSet.clear();
}
Also used : ArrayReader(org.apache.drill.exec.vector.accessor.ArrayReader) TupleReader(org.apache.drill.exec.vector.accessor.TupleReader) SingleRowSet(org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet) TupleWriter(org.apache.drill.exec.vector.accessor.TupleWriter) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.exec.record.metadata.SchemaBuilder) ArrayWriter(org.apache.drill.exec.vector.accessor.ArrayWriter) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 89 with SingleRowSet

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

the class TestRowSet method testDictStructure.

@Test
public void testDictStructure() {
    final String dictName = "d";
    final TupleMetadata schema = new SchemaBuilder().add("id", MinorType.INT).addDict(dictName, MinorType.INT).value(// required int
    MinorType.VARCHAR).resumeSchema().buildSchema();
    final ExtendableRowSet rowSet = fixture.rowSet(schema);
    final RowSetWriter writer = rowSet.writer();
    // Dict
    // Pick out components and lightly test. (Assumes structure
    // tested earlier is still valid, so no need to exhaustively
    // test again.)
    assertEquals(ObjectType.ARRAY, writer.column(dictName).type());
    assertTrue(writer.column(dictName).schema().isDict());
    final ScalarWriter idWriter = writer.column(0).scalar();
    final DictWriter dictWriter = writer.column(1).dict();
    assertEquals(ValueType.INTEGER, dictWriter.keyType());
    assertEquals(ObjectType.SCALAR, dictWriter.valueType());
    final ScalarWriter keyWriter = dictWriter.keyWriter();
    final ScalarWriter valueWriter = dictWriter.valueWriter().scalar();
    assertEquals(ValueType.INTEGER, keyWriter.valueType());
    assertEquals(ValueType.STRING, valueWriter.valueType());
    // Write data
    idWriter.setInt(1);
    keyWriter.setInt(11);
    valueWriter.setString("a");
    // Advance to next entry position
    dictWriter.save();
    keyWriter.setInt(12);
    valueWriter.setString("b");
    dictWriter.save();
    writer.save();
    idWriter.setInt(2);
    keyWriter.setInt(21);
    valueWriter.setString("c");
    dictWriter.save();
    writer.save();
    idWriter.setInt(3);
    keyWriter.setInt(31);
    valueWriter.setString("d");
    dictWriter.save();
    keyWriter.setInt(32);
    valueWriter.setString("e");
    dictWriter.save();
    writer.save();
    // Finish the row set and get a reader.
    final SingleRowSet actual = writer.done();
    final RowSetReader reader = actual.reader();
    // Verify reader structure
    assertEquals(ObjectType.ARRAY, reader.column(dictName).type());
    final DictReader dictReader = reader.dict(1);
    assertEquals(ObjectType.ARRAY, dictReader.type());
    assertEquals(ValueType.INTEGER, dictReader.keyColumnType());
    assertEquals(ObjectType.SCALAR, dictReader.valueColumnType());
    // Row 1: get value reader with its position set to entry corresponding to a key
    assertTrue(reader.next());
    // dict itself is not null
    assertFalse(dictReader.isNull());
    dictReader.getAsString();
    final KeyAccessor keyAccessor = dictReader.keyAccessor();
    final ScalarReader valueReader = dictReader.valueReader().scalar();
    assertTrue(keyAccessor.find(12));
    assertEquals("b", valueReader.getString());
    assertTrue(keyAccessor.find(11));
    assertEquals("a", valueReader.getString());
    // compare entire dict
    Map<Object, Object> map = map(11, "a", 12, "b");
    assertEquals(map, dictReader.getObject());
    // Row 2
    assertTrue(reader.next());
    // the dict does not contain an entry with the key
    assertFalse(keyAccessor.find(22));
    assertTrue(keyAccessor.find(21));
    assertEquals("c", valueReader.getString());
    map = map(21, "c");
    assertEquals(map, dictReader.getObject());
    // Row 3
    assertTrue(reader.next());
    assertTrue(keyAccessor.find(31));
    assertEquals("d", valueReader.getString());
    assertFalse(keyAccessor.find(33));
    assertTrue(keyAccessor.find(32));
    assertEquals("e", valueReader.getString());
    map = map(31, "d", 32, "e");
    assertEquals(map, dictReader.getObject());
    assertFalse(reader.next());
    // Verify that the dict accessor's value count was set.
    final DictVector dictVector = (DictVector) actual.container().getValueVector(1).getValueVector();
    assertEquals(3, dictVector.getAccessor().getValueCount());
    final SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(1, map(11, "a", 12, "b")).addRow(2, map(21, "c")).addRow(3, map(31, "d", 32, "e")).build();
    RowSetUtilities.verify(expected, actual);
}
Also used : DictVector(org.apache.drill.exec.vector.complex.DictVector) RepeatedDictVector(org.apache.drill.exec.vector.complex.RepeatedDictVector) DictWriter(org.apache.drill.exec.vector.accessor.DictWriter) SingleRowSet(org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet) ScalarReader(org.apache.drill.exec.vector.accessor.ScalarReader) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.exec.record.metadata.SchemaBuilder) KeyAccessor(org.apache.drill.exec.vector.accessor.KeyAccessor) DictReader(org.apache.drill.exec.vector.accessor.DictReader) ScalarWriter(org.apache.drill.exec.vector.accessor.ScalarWriter) ExtendableRowSet(org.apache.drill.exec.physical.rowSet.RowSet.ExtendableRowSet) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 90 with SingleRowSet

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

the class TestHyperVectorReaders method testRepeated.

/**
 * Test an array to test the indirection from the repeated vector
 * to the array offsets vector and the array values vector. (Uses
 * varchar to add another level of indirection to the data offset
 * and data values vectors.)
 */
@Test
public void testRepeated() {
    TupleMetadata schema = new SchemaBuilder().addArray("a", MinorType.VARCHAR).buildSchema();
    SingleRowSet rowSet1 = fixture.rowSetBuilder(schema).addSingleCol(strArray("sixth", "6.1", "6.2")).addSingleCol(strArray("second", "2.1", "2.2", "2.3")).addSingleCol(strArray("fourth", "4.1")).build();
    SingleRowSet rowSet2 = fixture.rowSetBuilder(schema).addSingleCol(strArray("fifth", "51", "5.2")).addSingleCol(strArray("first", "1.1", "1.2", "1.3")).addSingleCol(strArray("third", "3.1")).build();
    // Build the hyper batch
    HyperRowSet hyperSet = HyperRowSetImpl.fromRowSets(fixture.allocator(), rowSet1, rowSet2);
    assertEquals(6, hyperSet.rowCount());
    SelectionVector4 sv4 = hyperSet.getSv4();
    sv4.set(0, 1, 1);
    sv4.set(1, 0, 1);
    sv4.set(2, 1, 2);
    sv4.set(3, 0, 2);
    sv4.set(4, 1, 0);
    sv4.set(5, 0, 0);
    SingleRowSet expected = fixture.rowSetBuilder(schema).addSingleCol(strArray("first", "1.1", "1.2", "1.3")).addSingleCol(strArray("second", "2.1", "2.2", "2.3")).addSingleCol(strArray("third", "3.1")).addSingleCol(strArray("fourth", "4.1")).addSingleCol(strArray("fifth", "51", "5.2")).addSingleCol(strArray("sixth", "6.1", "6.2")).build();
    RowSetUtilities.verify(expected, hyperSet);
}
Also used : SingleRowSet(org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet) HyperRowSet(org.apache.drill.exec.physical.rowSet.RowSet.HyperRowSet) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.exec.record.metadata.SchemaBuilder) SelectionVector4(org.apache.drill.exec.record.selection.SelectionVector4) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Aggregations

SingleRowSet (org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet)257 Test (org.junit.Test)241 SchemaBuilder (org.apache.drill.exec.record.metadata.SchemaBuilder)237 TupleMetadata (org.apache.drill.exec.record.metadata.TupleMetadata)234 SubOperatorTest (org.apache.drill.test.SubOperatorTest)207 RowSet (org.apache.drill.exec.physical.rowSet.RowSet)69 ScalarReader (org.apache.drill.exec.vector.accessor.ScalarReader)62 ResultSetLoader (org.apache.drill.exec.physical.resultSet.ResultSetLoader)61 RowSetLoader (org.apache.drill.exec.physical.resultSet.RowSetLoader)54 ValueVector (org.apache.drill.exec.vector.ValueVector)32 EvfTest (org.apache.drill.categories.EvfTest)30 ScalarWriter (org.apache.drill.exec.vector.accessor.ScalarWriter)29 RowSetBuilder (org.apache.drill.exec.physical.rowSet.RowSetBuilder)27 TupleWriter (org.apache.drill.exec.vector.accessor.TupleWriter)27 RowSetComparison (org.apache.drill.test.rowSet.RowSetComparison)27 ExtendableRowSet (org.apache.drill.exec.physical.rowSet.RowSet.ExtendableRowSet)25 RepeatedValueVector (org.apache.drill.exec.vector.complex.RepeatedValueVector)24 BatchSchemaBuilder (org.apache.drill.exec.record.BatchSchemaBuilder)23 VectorContainer (org.apache.drill.exec.record.VectorContainer)22 ArrayReader (org.apache.drill.exec.vector.accessor.ArrayReader)22