use of org.apache.drill.exec.vector.accessor.ArrayReader in project drill by apache.
the class RowSetTest method TestTopScalarArray.
@Test
public void TestTopScalarArray() {
BatchSchema batchSchema = new SchemaBuilder().add("c", MinorType.INT).addArray("a", MinorType.INT).build();
ExtendableRowSet rs1 = fixture.rowSet(batchSchema);
RowSetWriter writer = rs1.writer();
writer.column(0).setInt(10);
ArrayWriter array = writer.column(1).array();
array.setInt(100);
array.setInt(110);
writer.save();
writer.column(0).setInt(20);
array = writer.column(1).array();
array.setInt(200);
array.setInt(120);
array.setInt(220);
writer.save();
writer.column(0).setInt(30);
writer.save();
writer.done();
RowSetReader reader = rs1.reader();
assertTrue(reader.next());
assertEquals(10, reader.column(0).getInt());
ArrayReader arrayReader = reader.column(1).array();
assertEquals(2, arrayReader.size());
assertEquals(100, arrayReader.getInt(0));
assertEquals(110, arrayReader.getInt(1));
assertTrue(reader.next());
assertEquals(20, reader.column(0).getInt());
arrayReader = reader.column(1).array();
assertEquals(3, arrayReader.size());
assertEquals(200, arrayReader.getInt(0));
assertEquals(120, arrayReader.getInt(1));
assertEquals(220, arrayReader.getInt(2));
assertTrue(reader.next());
assertEquals(30, reader.column(0).getInt());
arrayReader = reader.column(1).array();
assertEquals(0, arrayReader.size());
assertFalse(reader.next());
SingleRowSet rs2 = fixture.rowSetBuilder(batchSchema).add(10, new int[] { 100, 110 }).add(20, new int[] { 200, 120, 220 }).add(30, null).build();
new RowSetComparison(rs1).verifyAndClear(rs2);
}
use of org.apache.drill.exec.vector.accessor.ArrayReader in project drill by axbaretto.
the class TestResultSetLoaderMapArray method testOverwriteRow.
/**
* Version of the {#link TestResultSetLoaderProtocol#testOverwriteRow()} test
* that uses nested columns inside an array of maps. Here we must call
* <tt>start()</tt> to reset the array back to the initial start position after
* each "discard."
*/
@Test
public void testOverwriteRow() {
TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMapArray("m").add("b", MinorType.INT).add("c", 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();
// Can't use the shortcut to populate rows when doing overwrites.
ScalarWriter aWriter = rootWriter.scalar("a");
ArrayWriter maWriter = rootWriter.array("m");
TupleWriter mWriter = maWriter.tuple();
ScalarWriter bWriter = mWriter.scalar("b");
ScalarWriter cWriter = mWriter.scalar("c");
// Write 100,000 rows, overwriting 99% of them. This will cause vector
// overflow and data corruption if overwrite does not work; but will happily
// produce the correct result if everything works as it should.
byte[] value = new byte[512];
Arrays.fill(value, (byte) 'X');
int count = 0;
rsLoader.startBatch();
while (count < 10_000) {
rootWriter.start();
count++;
aWriter.setInt(count);
for (int i = 0; i < 10; i++) {
bWriter.setInt(count * 10 + i);
cWriter.setBytes(value, value.length);
maWriter.save();
}
if (count % 100 == 0) {
rootWriter.save();
}
}
// Verify using a reader.
RowSet result = fixture.wrap(rsLoader.harvest());
assertEquals(count / 100, result.rowCount());
RowSetReader reader = result.reader();
ArrayReader maReader = reader.array("m");
TupleReader mReader = maReader.tuple();
int rowId = 1;
while (reader.next()) {
assertEquals(rowId * 100, reader.scalar("a").getInt());
assertEquals(10, maReader.size());
for (int i = 0; i < 10; i++) {
maReader.setPosn(i);
assertEquals(rowId * 1000 + i, mReader.scalar("b").getInt());
assertTrue(Arrays.equals(value, mReader.scalar("c").getBytes()));
}
rowId++;
}
result.clear();
rsLoader.close();
}
use of org.apache.drill.exec.vector.accessor.ArrayReader in project drill by axbaretto.
the class RowSetTest method testRepeatedMapStructure.
@Test
public void testRepeatedMapStructure() {
TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMapArray("m").add("b", MinorType.INT).add("c", MinorType.INT).resumeSchema().buildSchema();
ExtendableRowSet rowSet = fixture.rowSet(schema);
RowSetWriter writer = rowSet.writer();
// Map and Int
// Pick out components and lightly test. (Assumes structure
// tested earlier is still valid, so no need to exhaustively
// test again.)
assertEquals(ObjectType.SCALAR, writer.column("a").type());
assertEquals(ObjectType.ARRAY, writer.column("m").type());
ArrayWriter maWriter = writer.column(1).array();
assertEquals(ObjectType.TUPLE, maWriter.entryType());
TupleWriter mapWriter = maWriter.tuple();
assertEquals(ObjectType.SCALAR, mapWriter.column("b").type());
assertEquals(ObjectType.SCALAR, mapWriter.column("c").type());
ScalarWriter aWriter = writer.column("a").scalar();
ScalarWriter bWriter = mapWriter.scalar("b");
ScalarWriter cWriter = mapWriter.scalar("c");
assertEquals(ValueType.INTEGER, aWriter.valueType());
assertEquals(ValueType.INTEGER, bWriter.valueType());
assertEquals(ValueType.INTEGER, cWriter.valueType());
// Write data
aWriter.setInt(10);
bWriter.setInt(101);
cWriter.setInt(102);
// Advance to next array position
maWriter.save();
bWriter.setInt(111);
cWriter.setInt(112);
maWriter.save();
writer.save();
aWriter.setInt(20);
bWriter.setInt(201);
cWriter.setInt(202);
maWriter.save();
bWriter.setInt(211);
cWriter.setInt(212);
maWriter.save();
writer.save();
aWriter.setInt(30);
bWriter.setInt(301);
cWriter.setInt(302);
maWriter.save();
bWriter.setInt(311);
cWriter.setInt(312);
maWriter.save();
writer.save();
// Finish the row set and get a reader.
SingleRowSet actual = writer.done();
RowSetReader reader = actual.reader();
// Verify reader structure
assertEquals(ObjectType.SCALAR, reader.column("a").type());
assertEquals(ObjectType.ARRAY, reader.column("m").type());
ArrayReader maReader = reader.column(1).array();
assertEquals(ObjectType.TUPLE, maReader.entryType());
TupleReader mapReader = maReader.tuple();
assertEquals(ObjectType.SCALAR, mapReader.column("b").type());
assertEquals(ObjectType.SCALAR, mapReader.column("c").type());
ScalarReader aReader = reader.column("a").scalar();
ScalarReader bReader = mapReader.scalar("b");
ScalarReader cReader = mapReader.scalar("c");
assertEquals(ValueType.INTEGER, aReader.valueType());
assertEquals(ValueType.INTEGER, bReader.valueType());
assertEquals(ValueType.INTEGER, cReader.valueType());
// Row 1: use index accessors
assertTrue(reader.next());
assertEquals(10, aReader.getInt());
TupleReader ixReader = maReader.tuple(0);
assertEquals(101, ixReader.scalar(0).getInt());
assertEquals(102, ixReader.scalar(1).getInt());
ixReader = maReader.tuple(1);
assertEquals(111, ixReader.scalar(0).getInt());
assertEquals(112, ixReader.scalar(1).getInt());
// Row 2: use common accessor with explicit positioning,
// but access scalars through the map reader.
assertTrue(reader.next());
assertEquals(20, aReader.getInt());
maReader.setPosn(0);
assertEquals(201, mapReader.scalar(0).getInt());
assertEquals(202, mapReader.scalar(1).getInt());
maReader.setPosn(1);
assertEquals(211, mapReader.scalar(0).getInt());
assertEquals(212, mapReader.scalar(1).getInt());
// Row 3: use common accessor for scalars
assertTrue(reader.next());
assertEquals(30, aReader.getInt());
maReader.setPosn(0);
assertEquals(301, bReader.getInt());
assertEquals(302, cReader.getInt());
maReader.setPosn(1);
assertEquals(311, bReader.getInt());
assertEquals(312, cReader.getInt());
assertFalse(reader.next());
// Verify that the map accessor's value count was set.
@SuppressWarnings("resource") RepeatedMapVector mapVector = (RepeatedMapVector) actual.container().getValueVector(1).getValueVector();
assertEquals(3, mapVector.getAccessor().getValueCount());
// Verify the readers and writers again using the testing tools.
SingleRowSet expected = fixture.rowSetBuilder(schema).addRow(10, objArray(objArray(101, 102), objArray(111, 112))).addRow(20, objArray(objArray(201, 202), objArray(211, 212))).addRow(30, objArray(objArray(301, 302), objArray(311, 312))).build();
new RowSetComparison(expected).verifyAndClearAll(actual);
}
use of org.apache.drill.exec.vector.accessor.ArrayReader in project drill by apache.
the class TestResultSetLoaderRepeatedList method test2DOverflow.
@Test
public void test2DOverflow() {
final TupleMetadata schema = new SchemaBuilder().add("id", MinorType.INT).addRepeatedList("list2").addArray(MinorType.VARCHAR).resumeSchema().buildSchema();
final ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().rowCountLimit(ValueVector.MAX_ROW_COUNT).readerSchema(schema).build();
final ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
final RowSetLoader writer = rsLoader.writer();
// Fill the batch with enough data to cause overflow.
// Data must be large enough to cause overflow before 64K rows
// Make a bit bigger to overflow early.
final int outerSize = 7;
final int innerSize = 5;
final int strLength = ValueVector.MAX_BUFFER_SIZE / ValueVector.MAX_ROW_COUNT / outerSize / innerSize + 20;
final byte[] value = new byte[strLength - 6];
Arrays.fill(value, (byte) 'X');
final String strValue = new String(value, Charsets.UTF_8);
int rowCount = 0;
int elementCount = 0;
final ArrayWriter outerWriter = writer.array(1);
final ArrayWriter innerWriter = outerWriter.array();
final ScalarWriter elementWriter = innerWriter.scalar();
rsLoader.startBatch();
while (!writer.isFull()) {
writer.start();
writer.scalar(0).setInt(rowCount);
for (int j = 0; j < outerSize; j++) {
for (int k = 0; k < innerSize; k++) {
elementWriter.setString(String.format("%s%06d", strValue, elementCount));
elementCount++;
}
outerWriter.save();
}
writer.save();
rowCount++;
}
// Number of rows should be driven by vector size.
// Our row count should include the overflow row
final int expectedCount = ValueVector.MAX_BUFFER_SIZE / (strLength * innerSize * outerSize);
assertEquals(expectedCount + 1, rowCount);
// Loader's row count should include only "visible" rows
assertEquals(expectedCount, writer.rowCount());
// Total count should include invisible and look-ahead rows.
assertEquals(expectedCount + 1, rsLoader.totalRowCount());
// Result should exclude the overflow row
RowSet result = fixture.wrap(rsLoader.harvest());
assertEquals(expectedCount, result.rowCount());
// Verify the data.
RowSetReader reader = result.reader();
ArrayReader outerReader = reader.array(1);
ArrayReader innerReader = outerReader.array();
ScalarReader strReader = innerReader.scalar();
int readRowCount = 0;
int readElementCount = 0;
while (reader.next()) {
assertEquals(readRowCount, reader.scalar(0).getInt());
for (int i = 0; i < outerSize; i++) {
assertTrue(outerReader.next());
for (int j = 0; j < innerSize; j++) {
assertTrue(innerReader.next());
assertEquals(String.format("%s%06d", strValue, readElementCount), strReader.getString());
readElementCount++;
}
assertFalse(innerReader.next());
}
assertFalse(outerReader.next());
readRowCount++;
}
assertEquals(readRowCount, result.rowCount());
result.clear();
// Write a few more rows to verify the overflow row.
rsLoader.startBatch();
for (int i = 0; i < 1000; i++) {
writer.start();
writer.scalar(0).setInt(rowCount);
for (int j = 0; j < outerSize; j++) {
for (int k = 0; k < innerSize; k++) {
elementWriter.setString(String.format("%s%06d", strValue, elementCount));
elementCount++;
}
outerWriter.save();
}
writer.save();
rowCount++;
}
result = fixture.wrap(rsLoader.harvest());
assertEquals(1001, result.rowCount());
final int startCount = readRowCount;
reader = result.reader();
outerReader = reader.array(1);
innerReader = outerReader.array();
strReader = innerReader.scalar();
while (reader.next()) {
assertEquals(readRowCount, reader.scalar(0).getInt());
for (int i = 0; i < outerSize; i++) {
assertTrue(outerReader.next());
for (int j = 0; j < innerSize; j++) {
assertTrue(innerReader.next());
elementWriter.setString(String.format("%s%06d", strValue, readElementCount));
assertEquals(String.format("%s%06d", strValue, readElementCount), strReader.getString());
readElementCount++;
}
assertFalse(innerReader.next());
}
assertFalse(outerReader.next());
readRowCount++;
}
assertEquals(readRowCount - startCount, result.rowCount());
result.clear();
rsLoader.close();
}
use of org.apache.drill.exec.vector.accessor.ArrayReader in project drill by apache.
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 ResultSetOptionBuilder().readerSchema(schema).rowCountLimit(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++) {
assertTrue(maReader.next());
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();
}
Aggregations