Examples with ScalarWriter org.apache.drill.exec.vector.accessor.ScalarWriter used on opensource projects

Example 76 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by apache.

the class TestResultSetLoaderOverflow method testOversizeArray.

/**
 * Case where a single array fills up the vector to the maximum size
 * limit. Overflow won't work here; the attempt will fail with a user
 * exception.
 */
@Test
public void testOversizeArray() {
    TupleMetadata schema = new SchemaBuilder().addArray("s", MinorType.VARCHAR).buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder().rowCountLimit(ValueVector.MAX_ROW_COUNT).readerSchema(schema).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();
    // Create a single array as the column value in the first row. When
    // this overflows, an exception is thrown since overflow is not possible.
    rsLoader.startBatch();
    byte[] value = new byte[473];
    Arrays.fill(value, (byte) 'X');
    rootWriter.start();
    ScalarWriter array = rootWriter.array(0).scalar();
    try {
        for (int i = 0; i < ValueVector.MAX_ROW_COUNT; i++) {
            array.setBytes(value, value.length);
        }
        fail();
    } catch (UserException e) {
        assertTrue(e.getMessage().contains("column value is larger than the maximum"));
    }
    rsLoader.close();
}

Example 77 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by apache.

the class TestResultSetLoaderOverflow method testArrayOverflowWithOtherArrays.

/**
 * Test the complete set of array overflow cases:
 * <ul>
 * <li>Array a is written before the column that has overflow,
 * and must be copied, in its entirety, to the overflow row.</li>
 * <li>Column b causes the overflow.</li>
 * <li>Column c is written after the overflow, and should go
 * to the look-ahead row.</li>
 * <li>Column d is written for a while, then has empties before
 * the overflow row, but is written in the overflow row.<li>
 * <li>Column e is like d, but is not written in the overflow
 * row.</li>
 */
@Test
public void testArrayOverflowWithOtherArrays() {
    TupleMetadata schema = new SchemaBuilder().addArray("a", MinorType.INT).addArray("b", MinorType.VARCHAR).addArray("c", MinorType.INT).addArray("d", MinorType.INT).buildSchema();
    ResultSetOptions options = new ResultSetOptionBuilder().rowCountLimit(ValueVector.MAX_ROW_COUNT).readerSchema(schema).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();
    // Fill batch with rows of with a single array, three values each. Tack on
    // a suffix to each so we can be sure the proper data is written and moved
    // to the overflow batch.
    byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');
    String strValue = new String(value, Charsets.UTF_8);
    int aCount = 3;
    int bCount = 11;
    int cCount = 5;
    int dCount = 7;
    int cCutoff = ValueVector.MAX_BUFFER_SIZE / value.length / bCount / 2;
    ScalarWriter aWriter = rootWriter.array("a").scalar();
    ScalarWriter bWriter = rootWriter.array("b").scalar();
    ScalarWriter cWriter = rootWriter.array("c").scalar();
    ScalarWriter dWriter = rootWriter.array("d").scalar();
    int count = 0;
    rsLoader.startBatch();
    while (rootWriter.start()) {
        for (int i = 0; i < aCount; i++) {
            aWriter.setInt(count * aCount + i);
        }
        for (int i = 0; i < bCount; i++) {
            String cellValue = strValue + (count * bCount + i);
            bWriter.setString(cellValue);
        }
        if (count < cCutoff) {
            for (int i = 0; i < cCount; i++) {
                cWriter.setInt(count * cCount + i);
            }
        }
        if (count < cCutoff || rootWriter.isFull()) {
            for (int i = 0; i < dCount; i++) {
                dWriter.setInt(count * dCount + i);
            }
        }
        rootWriter.save();
        count++;
    }
    // Verify
    {
        VectorContainer container = rsLoader.harvest();
        BatchValidator.validate(container);
        RowSet result = fixture.wrap(container);
        assertEquals(count - 1, result.rowCount());
        RowSetReader reader = result.reader();
        ArrayReader aArray = reader.array("a");
        ScalarReader aReader = aArray.scalar();
        ArrayReader bArray = reader.array("b");
        ScalarReader bReader = bArray.scalar();
        ArrayReader cArray = reader.array("c");
        ScalarReader cReader = cArray.scalar();
        ArrayReader dArray = reader.array("d");
        ScalarReader dReader = dArray.scalar();
        while (reader.next()) {
            int rowId = reader.offset();
            assertEquals(aCount, aArray.size());
            for (int i = 0; i < aCount; i++) {
                assertTrue(aArray.next());
                assertEquals(rowId * aCount + i, aReader.getInt());
            }
            assertEquals(bCount, bArray.size());
            for (int i = 0; i < bCount; i++) {
                assertTrue(bArray.next());
                String cellValue = strValue + (rowId * bCount + i);
                assertEquals(cellValue, bReader.getString());
            }
            if (rowId < cCutoff) {
                assertEquals(cCount, cArray.size());
                for (int i = 0; i < cCount; i++) {
                    assertTrue(cArray.next());
                    assertEquals(rowId * cCount + i, cReader.getInt());
                }
                assertEquals(dCount, dArray.size());
                for (int i = 0; i < dCount; i++) {
                    assertTrue(dArray.next());
                    assertEquals(rowId * dCount + i, dReader.getInt());
                }
            } else {
                assertEquals(0, cArray.size());
                assertEquals(0, dArray.size());
            }
        }
        result.clear();
    }
    int firstCount = count - 1;
    // One row is in the batch. Write more, skipping over the
    // initial few values for columns c and d. Column d has a
    // roll-over value, c has an empty roll-over.
    rsLoader.startBatch();
    for (int j = 0; j < 5; j++) {
        rootWriter.start();
        for (int i = 0; i < aCount; i++) {
            aWriter.setInt(count * aCount + i);
        }
        for (int i = 0; i < bCount; i++) {
            String cellValue = strValue + (count * bCount + i);
            bWriter.setString(cellValue);
        }
        if (j > 3) {
            for (int i = 0; i < cCount; i++) {
                cWriter.setInt(count * cCount + i);
            }
            for (int i = 0; i < dCount; i++) {
                dWriter.setInt(count * dCount + i);
            }
        }
        rootWriter.save();
        count++;
    }
    {
        VectorContainer container = rsLoader.harvest();
        BatchValidator.validate(container);
        RowSet result = fixture.wrap(container);
        assertEquals(6, result.rowCount());
        RowSetReader reader = result.reader();
        ArrayReader aArray = reader.array("a");
        ScalarReader aReader = aArray.scalar();
        ArrayReader bArray = reader.array("b");
        ScalarReader bReader = bArray.scalar();
        ArrayReader cArray = reader.array("c");
        ScalarReader cReader = cArray.scalar();
        ArrayReader dArray = reader.array("d");
        ScalarReader dReader = dArray.scalar();
        int j = 0;
        while (reader.next()) {
            int rowId = firstCount + reader.offset();
            assertEquals(aCount, aArray.size());
            for (int i = 0; i < aCount; i++) {
                assertTrue(aArray.next());
                assertEquals("Index " + i, rowId * aCount + i, aReader.getInt());
            }
            assertEquals(bCount, bArray.size());
            for (int i = 0; i < bCount; i++) {
                assertTrue(bArray.next());
                String cellValue = strValue + (rowId * bCount + i);
                assertEquals(cellValue, bReader.getString());
            }
            if (j > 4) {
                assertEquals(cCount, cArray.size());
                for (int i = 0; i < cCount; i++) {
                    assertTrue(cArray.next());
                    assertEquals(rowId * cCount + i, cReader.getInt());
                }
            } else {
                assertEquals(0, cArray.size());
            }
            if (j == 0 || j > 4) {
                assertEquals(dCount, dArray.size());
                for (int i = 0; i < dCount; i++) {
                    assertTrue(dArray.next());
                    assertEquals(rowId * dCount + i, dReader.getInt());
                }
            } else {
                assertEquals(0, dArray.size());
            }
            j++;
        }
        result.clear();
    }
    rsLoader.close();
}

Example 78 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by apache.

the class TestResultSetLoaderDicts method testValueOverflow.

@Test
public void testValueOverflow() {
    TupleMetadata schema = new SchemaBuilder().addDict("d", MinorType.INT).value(MinorType.VARCHAR).resumeSchema().buildSchema();
    ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().rowCountLimit(ValueVector.MAX_ROW_COUNT).readerSchema(schema).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();
    rsLoader.startBatch();
    byte[] value = new byte[523];
    Arrays.fill(value, (byte) 'X');
    // number of entries in each dict
    int dictSize = 4;
    // Number of rows should be driven by vector size.
    // Our row count should include the overflow row
    DictWriter dictWriter = rootWriter.dict(0);
    ScalarWriter keyWriter = dictWriter.keyWriter();
    ScalarWriter valueWriter = dictWriter.valueWriter().scalar();
    int expectedCount = ValueVector.MAX_BUFFER_SIZE / (value.length * dictSize);
    {
        int count = 0;
        while (!rootWriter.isFull()) {
            rootWriter.start();
            for (int i = 0; i < dictSize; i++) {
                // acts as a placeholder, the actual value is not important
                keyWriter.setInt(0);
                valueWriter.setBytes(value, value.length);
                // not necessary for scalars, just for completeness
                dictWriter.save();
            }
            rootWriter.save();
            count++;
        }
        assertEquals(expectedCount + 1, count);
        // Loader's row count should include only "visible" rows
        assertEquals(expectedCount, rootWriter.rowCount());
        // Total count should include invisible and look-ahead rows.
        assertEquals(expectedCount + 1, rsLoader.totalRowCount());
        // Result should exclude the overflow row
        VectorContainer container = rsLoader.harvest();
        BatchValidator.validate(container);
        RowSet result = fixture.wrap(container);
        assertEquals(expectedCount, result.rowCount());
        result.clear();
    }
    // Next batch should start with the overflow row
    {
        rsLoader.startBatch();
        assertEquals(1, rootWriter.rowCount());
        assertEquals(expectedCount + 1, rsLoader.totalRowCount());
        VectorContainer container = rsLoader.harvest();
        BatchValidator.validate(container);
        RowSet result = fixture.wrap(container);
        assertEquals(1, result.rowCount());
        result.clear();
    }
    rsLoader.close();
}

Example 79 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by apache.

the class TestResultSetLoaderDicts method testBasics.

@Test
public void testBasics() {
    final TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addDict("d", MinorType.INT).value(MinorType.VARCHAR).resumeSchema().buildSchema();
    final ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().readerSchema(schema).build();
    final ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    assertFalse(rsLoader.isProjectionEmpty());
    final RowSetLoader rootWriter = rsLoader.writer();
    // Verify structure and schema
    assertEquals(4, rsLoader.schemaVersion());
    final TupleMetadata actualSchema = rootWriter.tupleSchema();
    assertEquals(2, actualSchema.size());
    assertTrue(actualSchema.metadata(1).isDict());
    assertEquals(2, actualSchema.metadata("d").tupleSchema().size());
    assertEquals(2, actualSchema.column("d").getChildren().size());
    rsLoader.startBatch();
    // Write a row the way that clients will do.
    final ScalarWriter aWriter = rootWriter.scalar("a");
    final DictWriter dictWriter = rootWriter.dict("d");
    final ScalarWriter keyWriter = dictWriter.keyWriter();
    final ScalarWriter valueWriter = dictWriter.valueWriter().scalar();
    rootWriter.start();
    aWriter.setInt(10);
    keyWriter.setInt(110);
    valueWriter.setString("fred");
    dictWriter.save();
    keyWriter.setInt(111);
    valueWriter.setString("george");
    dictWriter.save();
    rootWriter.save();
    // Write another using the test-time conveniences
    rootWriter.addRow(20, map(210, "barney", 211, "bart", 212, "jerry"));
    // Harvest the batch
    final RowSet actual = fixture.wrap(rsLoader.harvest());
    assertEquals(4, rsLoader.schemaVersion());
    assertEquals(2, actual.rowCount());
    final DictVector dictVector = (DictVector) actual.container().getValueVector(1).getValueVector();
    assertEquals(2, dictVector.getAccessor().getValueCount());
    // Validate data
    final SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(10, map(110, "fred", 111, "george")).addRow(20, map(210, "barney", 211, "bart", 212, "jerry")).build();
    RowSetUtilities.verify(expected, actual);
    rsLoader.close();
}

Example 80 with ScalarWriter

use of org.apache.drill.exec.vector.accessor.ScalarWriter in project drill by apache.

the class TestResultSetLoaderMaps method testNameSpace.

/**
 * Verify that map name spaces (and implementations) are
 * independent.
 */
@Test
public void testNameSpace() {
    final TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMap("m").add("a", MinorType.INT).addMap("m").add("a", MinorType.INT).resumeMap().resumeSchema().buildSchema();
    final ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().readerSchema(schema).build();
    final ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    assertFalse(rsLoader.isProjectionEmpty());
    final RowSetLoader rootWriter = rsLoader.writer();
    rsLoader.startBatch();
    // Write a row the way that clients will do.
    final ScalarWriter a1Writer = rootWriter.scalar("a");
    final TupleWriter m1Writer = rootWriter.tuple("m");
    final ScalarWriter a2Writer = m1Writer.scalar("a");
    final TupleWriter m2Writer = m1Writer.tuple("m");
    final ScalarWriter a3Writer = m2Writer.scalar("a");
    rootWriter.start();
    a1Writer.setInt(11);
    a2Writer.setInt(12);
    a3Writer.setInt(13);
    rootWriter.save();
    rootWriter.start();
    a1Writer.setInt(21);
    a2Writer.setInt(22);
    a3Writer.setInt(23);
    rootWriter.save();
    // Try simplified test format
    rootWriter.addRow(31, mapValue(32, mapValue(33)));
    // Verify
    final RowSet actual = fixture.wrap(rsLoader.harvest());
    final SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(11, mapValue(12, mapValue(13))).addRow(21, mapValue(22, mapValue(23))).addRow(31, mapValue(32, mapValue(33))).build();
    RowSetUtilities.verify(expected, actual);
    rsLoader.close();
}