Examples with RowSet org.apache.drill.test.rowSet.RowSet used on opensource projects

Example 51 with RowSet

use of org.apache.drill.test.rowSet.RowSet in project drill by axbaretto.

the class TestShortArrays method testSizer.

@Test
public void testSizer() {
    // Create a row set with less than one item, on
    // average, per array.
    BatchSchema schema = new SchemaBuilder().add("a", MinorType.INT).addArray("b", MinorType.INT).build();
    RowSetBuilder builder = fixture.rowSetBuilder(schema).addRow(1, intArray(10));
    for (int i = 2; i <= 10; i++) {
        builder.addRow(i, intArray());
    }
    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 bCol = sizer.columns().get("b");
    assertEquals(0.1, bCol.getCardinality(), 0.01);
    assertEquals(1, bCol.getElementCount());
    // Create a vector initializer using the sizer info.
    VectorInitializer vi = sizer.buildVectorInitializer();
    AllocationHint bHint = vi.hint("b");
    assertNotNull(bHint);
    assertEquals(bHint.elementCount, bCol.getCardinality(), 0.001);
    // Create a new batch, and new vector, using the sizer and
    // initializer inferred from the previous batch.
    SingleRowSet empty = fixture.rowSet(schema);
    vi.allocateBatch(empty.container(), 100);
    assertEquals(2, empty.container().getNumberOfColumns());
    @SuppressWarnings("resource") ValueVector bVector = empty.container().getValueVector(1).getValueVector();
    assertTrue(bVector instanceof RepeatedIntVector);
    assertEquals(16, ((RepeatedIntVector) bVector).getDataVector().getValueCapacity());
    rows.clear();
    empty.clear();
}

Example 52 with RowSet

use of org.apache.drill.test.rowSet.RowSet in project drill by axbaretto.

the class TestResultSetLoaderMapArray method testOmittedValues.

/**
 * Check that the "fill-empties" logic descends down into
 * a repeated map.
 */
@Test
public void testOmittedValues() {
    TupleMetadata schema = new SchemaBuilder().add("id", MinorType.INT).addMapArray("m").addNullable("a", MinorType.INT).addNullable("b", MinorType.VARCHAR).resumeSchema().buildSchema();
    ResultSetLoaderImpl.ResultSetOptions options = new OptionBuilder().setSchema(schema).setRowCountLimit(ValueVector.MAX_ROW_COUNT).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();
    int mapSkip = 5;
    int entrySkip = 3;
    int rowCount = 1000;
    int entryCount = 10;
    rsLoader.startBatch();
    ArrayWriter maWriter = rootWriter.array("m");
    TupleWriter mWriter = maWriter.tuple();
    for (int i = 0; i < rowCount; i++) {
        rootWriter.start();
        rootWriter.scalar(0).setInt(i);
        if (i % mapSkip != 0) {
            for (int j = 0; j < entryCount; j++) {
                if (j % entrySkip != 0) {
                    mWriter.scalar(0).setInt(i * entryCount + j);
                    mWriter.scalar(1).setString("b-" + i + "." + j);
                }
                maWriter.save();
            }
        }
        rootWriter.save();
    }
    RowSet result = fixture.wrap(rsLoader.harvest());
    assertEquals(rowCount, result.rowCount());
    RowSetReader reader = result.reader();
    ArrayReader maReader = reader.array("m");
    TupleReader mReader = maReader.tuple();
    for (int i = 0; i < rowCount; i++) {
        assertTrue(reader.next());
        assertEquals(i, reader.scalar(0).getInt());
        if (i % mapSkip == 0) {
            assertEquals(0, maReader.size());
            continue;
        }
        assertEquals(entryCount, maReader.size());
        for (int j = 0; j < entryCount; j++) {
            maReader.setPosn(j);
            if (j % entrySkip == 0) {
                assertTrue(mReader.scalar(0).isNull());
                assertTrue(mReader.scalar(1).isNull());
            } else {
                assertFalse(mReader.scalar(0).isNull());
                assertFalse(mReader.scalar(1).isNull());
                assertEquals(i * entryCount + j, mReader.scalar(0).getInt());
                assertEquals("b-" + i + "." + j, mReader.scalar(1).getString());
            }
        }
    }
    result.clear();
    rsLoader.close();
}

Example 53 with RowSet

use of org.apache.drill.test.rowSet.RowSet in project drill by axbaretto.

the class TestResultSetLoaderMapArray method testDoubleNestedArray.

/**
 * Test a doubly-nested arrays of maps.
 */
@Test
public void testDoubleNestedArray() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMapArray("m1").add("b", MinorType.INT).addMapArray("m2").add("c", MinorType.INT).addArray("d", MinorType.VARCHAR).resumeMap().resumeSchema().buildSchema();
    ResultSetLoaderImpl.ResultSetOptions options = new OptionBuilder().setSchema(schema).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();
    rsLoader.startBatch();
    ScalarWriter aWriter = rootWriter.scalar("a");
    ArrayWriter a1Writer = rootWriter.array("m1");
    TupleWriter m1Writer = a1Writer.tuple();
    ScalarWriter bWriter = m1Writer.scalar("b");
    ArrayWriter a2Writer = m1Writer.array("m2");
    TupleWriter m2Writer = a2Writer.tuple();
    ScalarWriter cWriter = m2Writer.scalar("c");
    ScalarWriter dWriter = m2Writer.array("d").scalar();
    for (int i = 0; i < 5; i++) {
        rootWriter.start();
        aWriter.setInt(i);
        for (int j = 0; j < 4; j++) {
            int a1Key = i + 10 + j;
            bWriter.setInt(a1Key);
            for (int k = 0; k < 3; k++) {
                int a2Key = a1Key * 10 + k;
                cWriter.setInt(a2Key);
                for (int l = 0; l < 2; l++) {
                    dWriter.setString("d-" + (a2Key * 10 + l));
                }
                a2Writer.save();
            }
            a1Writer.save();
        }
        rootWriter.save();
    }
    RowSet results = fixture.wrap(rsLoader.harvest());
    RowSetReader reader = results.reader();
    ScalarReader aReader = reader.scalar("a");
    ArrayReader a1Reader = reader.array("m1");
    TupleReader m1Reader = a1Reader.tuple();
    ScalarReader bReader = m1Reader.scalar("b");
    ArrayReader a2Reader = m1Reader.array("m2");
    TupleReader m2Reader = a2Reader.tuple();
    ScalarReader cReader = m2Reader.scalar("c");
    ScalarElementReader dReader = m2Reader.elements("d");
    for (int i = 0; i < 5; i++) {
        reader.next();
        assertEquals(i, aReader.getInt());
        for (int j = 0; j < 4; j++) {
            a1Reader.setPosn(j);
            int a1Key = i + 10 + j;
            assertEquals(a1Key, bReader.getInt());
            for (int k = 0; k < 3; k++) {
                a2Reader.setPosn(k);
                int a2Key = a1Key * 10 + k;
                assertEquals(a2Key, cReader.getInt());
                for (int l = 0; l < 2; l++) {
                    assertEquals("d-" + (a2Key * 10 + l), dReader.getString(l));
                }
            }
        }
    }
    rsLoader.close();
}

Example 54 with RowSet

use of org.apache.drill.test.rowSet.RowSet in project drill by axbaretto.

the class TestResultSetLoaderMaps method testNestedMapsRequired.

/**
 * Create nested maps. Then, add columns to each map
 * on the fly. Use required, variable-width columns since
 * those require the most processing and are most likely to
 * fail if anything is out of place.
 */
@Test
public void testNestedMapsRequired() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMap("m1").add("b", MinorType.VARCHAR).addMap("m2").add("c", MinorType.VARCHAR).resumeMap().resumeSchema().buildSchema();
    ResultSetLoaderImpl.ResultSetOptions options = new OptionBuilder().setSchema(schema).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    assertEquals(5, rsLoader.schemaVersion());
    RowSetLoader rootWriter = rsLoader.writer();
    rsLoader.startBatch();
    rootWriter.addRow(10, objArray("b1", objArray("c1")));
    // Validate first batch
    RowSet actual = fixture.wrap(rsLoader.harvest());
    assertEquals(5, rsLoader.schemaVersion());
    SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(10, objArray("b1", objArray("c1"))).build();
    new RowSetComparison(expected).verifyAndClearAll(actual);
    // Now add columns in the second batch.
    rsLoader.startBatch();
    rootWriter.addRow(20, objArray("b2", objArray("c2")));
    TupleWriter m1Writer = rootWriter.tuple("m1");
    m1Writer.addColumn(SchemaBuilder.columnSchema("d", MinorType.VARCHAR, DataMode.REQUIRED));
    TupleWriter m2Writer = m1Writer.tuple("m2");
    m2Writer.addColumn(SchemaBuilder.columnSchema("e", MinorType.VARCHAR, DataMode.REQUIRED));
    rootWriter.addRow(30, objArray("b3", objArray("c3", "e3"), "d3"));
    // And another set while the write proceeds.
    m1Writer.addColumn(SchemaBuilder.columnSchema("f", MinorType.VARCHAR, DataMode.REQUIRED));
    m2Writer.addColumn(SchemaBuilder.columnSchema("g", MinorType.VARCHAR, DataMode.REQUIRED));
    rootWriter.addRow(40, objArray("b4", objArray("c4", "e4", "g4"), "d4", "e4"));
    // Validate second batch
    actual = fixture.wrap(rsLoader.harvest());
    assertEquals(9, rsLoader.schemaVersion());
    TupleMetadata expectedSchema = new SchemaBuilder().add("a", MinorType.INT).addMap("m1").add("b", MinorType.VARCHAR).addMap("m2").add("c", MinorType.VARCHAR).add("e", MinorType.VARCHAR).add("g", MinorType.VARCHAR).resumeMap().add("d", MinorType.VARCHAR).add("f", MinorType.VARCHAR).resumeSchema().buildSchema();
    expected = fixture.rowSetBuilder(expectedSchema).addRow(20, objArray("b2", objArray("c2", "", ""), "", "")).addRow(30, objArray("b3", objArray("c3", "e3", ""), "d3", "")).addRow(40, objArray("b4", objArray("c4", "e4", "g4"), "d4", "e4")).build();
    new RowSetComparison(expected).verifyAndClearAll(actual);
    rsLoader.close();
}

Example 55 with RowSet

use of org.apache.drill.test.rowSet.RowSet in project drill by axbaretto.

the class TestExternalSort method testNewColumns.

private void testNewColumns(boolean testLegacy) throws Exception {
    final int record_count = 10000;
    final String tableDirName = "newColumns";
    {
        final BatchSchema schema = new SchemaBuilder().add("a", TypeProtos.MinorType.INT).add("b", TypeProtos.MinorType.INT).build();
        final RowSetBuilder rowSetBuilder = new RowSetBuilder(allocator, schema);
        for (int i = 0; i <= record_count; i += 2) {
            rowSetBuilder.addRow(i, i);
        }
        final RowSet rowSet = rowSetBuilder.build();
        final File tableFile = createTableFile(tableDirName, "a.json");
        new JsonFileBuilder(rowSet).build(tableFile);
        rowSet.clear();
    }
    {
        final BatchSchema schema = new SchemaBuilder().add("a", TypeProtos.MinorType.INT).add("c", TypeProtos.MinorType.INT).build();
        final RowSetBuilder rowSetBuilder = new RowSetBuilder(allocator, schema);
        for (int i = 1; i <= record_count; i += 2) {
            rowSetBuilder.addRow(i, i);
        }
        final RowSet rowSet = rowSetBuilder.build();
        final File tableFile = createTableFile(tableDirName, "b.json");
        new JsonFileBuilder(rowSet).build(tableFile);
        rowSet.clear();
    }
    // Test framework currently doesn't handle changing schema (i.e. new
    // columns) on the client side
    TestBuilder builder = testBuilder().sqlQuery("select a, b, c from dfs.`%s` order by a desc", tableDirName).ordered().optionSettingQueriesForTestQuery(getOptions(testLegacy)).baselineColumns("a", "b", "c");
    for (int i = record_count; i >= 0; ) {
        builder.baselineValues((long) i, (long) i--, null);
        if (i >= 0) {
            builder.baselineValues((long) i, null, (long) i--);
        }
    }
    builder.go();
    // TODO: Useless test: just dumps to console
    test("select * from dfs.`%s` order by a desc", tableDirName);
}