use of io.trino.spi.PageBuilder in project trino by trinodb.
the class FunctionAssertions method getRetainedSizeOf.
private long getRetainedSizeOf(Object object) {
if (object instanceof PageBuilder) {
return ((PageBuilder) object).getRetainedSizeInBytes();
}
if (object instanceof Block) {
return ((Block) object).getRetainedSizeInBytes();
}
Class<?> type = object.getClass();
if (type.isArray()) {
if (type == int[].class) {
return sizeOf((int[]) object);
}
if (type == boolean[].class) {
return sizeOf((boolean[]) object);
}
if (type == byte[].class) {
return sizeOf((byte[]) object);
}
if (type == long[].class) {
return sizeOf((long[]) object);
}
if (type == short[].class) {
return sizeOf((short[]) object);
}
if (type == Block[].class) {
Object[] objects = (Object[]) object;
return Arrays.stream(objects).mapToLong(this::getRetainedSizeOf).sum();
}
throw new IllegalArgumentException(format("Unknown type encountered: %s", type));
}
long retainedSize = ClassLayout.parseClass(type).instanceSize();
Field[] fields = type.getDeclaredFields();
for (Field field : fields) {
try {
if (field.getType().isPrimitive() || Modifier.isStatic(field.getModifiers())) {
continue;
}
field.setAccessible(true);
retainedSize += getRetainedSizeOf(field.get(object));
} catch (IllegalAccessException t) {
throw new RuntimeException(t);
}
}
return retainedSize;
}
use of io.trino.spi.PageBuilder in project trino by trinodb.
the class TestJoinCompiler method testMultiChannel.
@Test(dataProvider = "hashEnabledValues")
public void testMultiChannel(boolean hashEnabled) {
// compile a single channel hash strategy
List<Type> types = ImmutableList.of(VARCHAR, VARCHAR, BIGINT, DOUBLE, BOOLEAN, VARCHAR);
List<Type> joinTypes = ImmutableList.of(VARCHAR, BIGINT, DOUBLE, BOOLEAN);
List<Type> outputTypes = ImmutableList.of(VARCHAR, BIGINT, DOUBLE, BOOLEAN, VARCHAR);
List<Integer> joinChannels = Ints.asList(1, 2, 3, 4);
List<Integer> outputChannels = Ints.asList(1, 2, 3, 4, 0);
// crate hash strategy with a single channel blocks -- make sure there is some overlap in values
List<Block> extraChannel = ImmutableList.of(BlockAssertions.createStringSequenceBlock(10, 20), BlockAssertions.createStringSequenceBlock(20, 30), BlockAssertions.createStringSequenceBlock(15, 25));
List<Block> varcharChannel = ImmutableList.of(BlockAssertions.createStringSequenceBlock(10, 20), BlockAssertions.createStringSequenceBlock(20, 30), BlockAssertions.createStringSequenceBlock(15, 25));
List<Block> longChannel = ImmutableList.of(BlockAssertions.createLongSequenceBlock(10, 20), BlockAssertions.createLongSequenceBlock(20, 30), BlockAssertions.createLongSequenceBlock(15, 25));
List<Block> doubleChannel = ImmutableList.of(BlockAssertions.createDoubleSequenceBlock(10, 20), BlockAssertions.createDoubleSequenceBlock(20, 30), BlockAssertions.createDoubleSequenceBlock(15, 25));
List<Block> booleanChannel = ImmutableList.of(BlockAssertions.createBooleanSequenceBlock(10, 20), BlockAssertions.createBooleanSequenceBlock(20, 30), BlockAssertions.createBooleanSequenceBlock(15, 25));
List<Block> extraUnusedChannel = ImmutableList.of(BlockAssertions.createBooleanSequenceBlock(10, 20), BlockAssertions.createBooleanSequenceBlock(20, 30), BlockAssertions.createBooleanSequenceBlock(15, 25));
OptionalInt hashChannel = OptionalInt.empty();
ImmutableList<List<Block>> channels = ImmutableList.of(extraChannel, varcharChannel, longChannel, doubleChannel, booleanChannel, extraUnusedChannel);
List<Block> precomputedHash = ImmutableList.of();
if (hashEnabled) {
ImmutableList.Builder<Block> hashChannelBuilder = ImmutableList.builder();
for (int i = 0; i < 3; i++) {
hashChannelBuilder.add(TypeTestUtils.getHashBlock(joinTypes, varcharChannel.get(i), longChannel.get(i), doubleChannel.get(i), booleanChannel.get(i)));
}
hashChannel = OptionalInt.of(6);
precomputedHash = hashChannelBuilder.build();
channels = ImmutableList.of(extraChannel, varcharChannel, longChannel, doubleChannel, booleanChannel, extraUnusedChannel, precomputedHash);
types = ImmutableList.of(VARCHAR, VARCHAR, BIGINT, DOUBLE, BOOLEAN, VARCHAR, BIGINT);
outputTypes = ImmutableList.of(VARCHAR, BIGINT, DOUBLE, BOOLEAN, VARCHAR, BIGINT);
outputChannels = Ints.asList(1, 2, 3, 4, 0, 6);
}
PagesHashStrategyFactory pagesHashStrategyFactory = joinCompiler.compilePagesHashStrategyFactory(types, joinChannels, Optional.of(outputChannels));
PagesHashStrategy hashStrategy = pagesHashStrategyFactory.createPagesHashStrategy(channels, hashChannel);
// todo add tests for filter function
PagesHashStrategy expectedHashStrategy = new SimplePagesHashStrategy(types, outputChannels, channels, joinChannels, hashChannel, Optional.empty(), blockTypeOperators);
// verify channel count
assertEquals(hashStrategy.getChannelCount(), outputChannels.size());
// verify size
int instanceSize = ClassLayout.parseClass(hashStrategy.getClass()).instanceSize();
long sizeInBytes = instanceSize + channels.stream().flatMap(List::stream).mapToLong(Block::getRetainedSizeInBytes).sum();
assertEquals(hashStrategy.getSizeInBytes(), sizeInBytes);
// verify hashStrategy is consistent with equals and hash code from block
for (int leftBlockIndex = 0; leftBlockIndex < varcharChannel.size(); leftBlockIndex++) {
PageBuilder pageBuilder = new PageBuilder(outputTypes);
Block[] leftBlocks = new Block[4];
leftBlocks[0] = varcharChannel.get(leftBlockIndex);
leftBlocks[1] = longChannel.get(leftBlockIndex);
leftBlocks[2] = doubleChannel.get(leftBlockIndex);
leftBlocks[3] = booleanChannel.get(leftBlockIndex);
int leftPositionCount = varcharChannel.get(leftBlockIndex).getPositionCount();
for (int leftBlockPosition = 0; leftBlockPosition < leftPositionCount; leftBlockPosition++) {
// hash code of position must match block hash
assertEquals(hashStrategy.hashPosition(leftBlockIndex, leftBlockPosition), expectedHashStrategy.hashPosition(leftBlockIndex, leftBlockPosition));
// position must be equal to itself
assertTrue(hashStrategy.positionEqualsPositionIgnoreNulls(leftBlockIndex, leftBlockPosition, leftBlockIndex, leftBlockPosition));
assertTrue(hashStrategy.positionEqualsPosition(leftBlockIndex, leftBlockPosition, leftBlockIndex, leftBlockPosition));
assertTrue(hashStrategy.positionNotDistinctFromPosition(leftBlockIndex, leftBlockPosition, leftBlockIndex, leftBlockPosition));
// check equality of every position against every other position in the block
for (int rightBlockIndex = 0; rightBlockIndex < varcharChannel.size(); rightBlockIndex++) {
Block rightBlock = varcharChannel.get(rightBlockIndex);
for (int rightBlockPosition = 0; rightBlockPosition < rightBlock.getPositionCount(); rightBlockPosition++) {
assertEquals(hashStrategy.positionEqualsPositionIgnoreNulls(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expectedHashStrategy.positionEqualsPositionIgnoreNulls(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition));
assertEquals(hashStrategy.positionEqualsPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expectedHashStrategy.positionEqualsPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition));
assertEquals(hashStrategy.positionNotDistinctFromPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expectedHashStrategy.positionNotDistinctFromPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition));
}
}
// check equality of every position against every other position in the block cursor
for (int rightBlockIndex = 0; rightBlockIndex < varcharChannel.size(); rightBlockIndex++) {
Block[] rightBlocks = new Block[4];
rightBlocks[0] = varcharChannel.get(rightBlockIndex);
rightBlocks[1] = longChannel.get(rightBlockIndex);
rightBlocks[2] = doubleChannel.get(rightBlockIndex);
rightBlocks[3] = booleanChannel.get(rightBlockIndex);
int rightPositionCount = varcharChannel.get(rightBlockIndex).getPositionCount();
for (int rightPosition = 0; rightPosition < rightPositionCount; rightPosition++) {
boolean expected = expectedHashStrategy.positionEqualsRow(leftBlockIndex, leftBlockPosition, rightPosition, new Page(rightBlocks));
boolean expectedNotDistinct = expectedHashStrategy.positionNotDistinctFromRow(leftBlockIndex, leftBlockPosition, rightPosition, new Page(rightBlocks));
assertEquals(hashStrategy.positionEqualsRow(leftBlockIndex, leftBlockPosition, rightPosition, new Page(rightBlocks)), expected);
assertEquals(hashStrategy.positionNotDistinctFromRow(leftBlockIndex, leftBlockPosition, rightPosition, new Page(rightBlocks)), expectedNotDistinct);
assertEquals(hashStrategy.rowEqualsRow(leftBlockPosition, new Page(leftBlocks), rightPosition, new Page(rightBlocks)), expected);
assertEquals(hashStrategy.rowNotDistinctFromRow(leftBlockPosition, new Page(leftBlocks), rightPosition, new Page(rightBlocks)), expectedNotDistinct);
assertEquals(hashStrategy.positionEqualsRowIgnoreNulls(leftBlockIndex, leftBlockPosition, rightPosition, new Page(rightBlocks)), expected);
}
}
// write position to output block
pageBuilder.declarePosition();
hashStrategy.appendTo(leftBlockIndex, leftBlockPosition, pageBuilder, 0);
}
// verify output block matches
Page page = pageBuilder.build();
if (hashEnabled) {
assertPageEquals(outputTypes, page, new Page(varcharChannel.get(leftBlockIndex), longChannel.get(leftBlockIndex), doubleChannel.get(leftBlockIndex), booleanChannel.get(leftBlockIndex), extraChannel.get(leftBlockIndex), precomputedHash.get(leftBlockIndex)));
} else {
assertPageEquals(outputTypes, page, new Page(varcharChannel.get(leftBlockIndex), longChannel.get(leftBlockIndex), doubleChannel.get(leftBlockIndex), booleanChannel.get(leftBlockIndex), extraChannel.get(leftBlockIndex)));
}
}
}
use of io.trino.spi.PageBuilder in project trino by trinodb.
the class TestJoinCompiler method testSingleChannel.
@Test(dataProvider = "hashEnabledValues")
public void testSingleChannel(boolean hashEnabled) {
List<Type> joinTypes = ImmutableList.of(VARCHAR);
List<Integer> joinChannels = Ints.asList(0);
// compile a single channel hash strategy
PagesHashStrategyFactory pagesHashStrategyFactory = joinCompiler.compilePagesHashStrategyFactory(joinTypes, joinChannels);
// create hash strategy with a single channel blocks -- make sure there is some overlap in values
List<Block> channel = ImmutableList.of(BlockAssertions.createStringSequenceBlock(10, 20), BlockAssertions.createStringSequenceBlock(20, 30), BlockAssertions.createStringSequenceBlock(15, 25));
OptionalInt hashChannel = OptionalInt.empty();
List<List<Block>> channels = ImmutableList.of(channel);
if (hashEnabled) {
ImmutableList.Builder<Block> hashChannelBuilder = ImmutableList.builder();
for (Block block : channel) {
hashChannelBuilder.add(TypeTestUtils.getHashBlock(joinTypes, block));
}
hashChannel = OptionalInt.of(1);
channels = ImmutableList.of(channel, hashChannelBuilder.build());
}
PagesHashStrategy hashStrategy = pagesHashStrategyFactory.createPagesHashStrategy(channels, hashChannel);
// verify channel count
assertEquals(hashStrategy.getChannelCount(), 1);
BlockTypeOperators blockTypeOperators = new BlockTypeOperators();
BlockPositionEqual equalOperator = blockTypeOperators.getEqualOperator(VARCHAR);
BlockPositionIsDistinctFrom distinctFromOperator = blockTypeOperators.getDistinctFromOperator(VARCHAR);
BlockPositionHashCode hashCodeOperator = blockTypeOperators.getHashCodeOperator(VARCHAR);
// verify hashStrategy is consistent with equals and hash code from block
for (int leftBlockIndex = 0; leftBlockIndex < channel.size(); leftBlockIndex++) {
Block leftBlock = channel.get(leftBlockIndex);
PageBuilder pageBuilder = new PageBuilder(ImmutableList.of(VARCHAR));
for (int leftBlockPosition = 0; leftBlockPosition < leftBlock.getPositionCount(); leftBlockPosition++) {
// hash code of position must match block hash
assertEquals(hashStrategy.hashPosition(leftBlockIndex, leftBlockPosition), hashCodeOperator.hashCodeNullSafe(leftBlock, leftBlockPosition));
// position must be equal to itself
assertTrue(hashStrategy.positionEqualsPositionIgnoreNulls(leftBlockIndex, leftBlockPosition, leftBlockIndex, leftBlockPosition));
// check equality of every position against every other position in the block
for (int rightBlockIndex = 0; rightBlockIndex < channel.size(); rightBlockIndex++) {
Block rightBlock = channel.get(rightBlockIndex);
for (int rightBlockPosition = 0; rightBlockPosition < rightBlock.getPositionCount(); rightBlockPosition++) {
boolean expected = equalOperator.equalNullSafe(leftBlock, leftBlockPosition, rightBlock, rightBlockPosition);
boolean expectedNotDistinct = !distinctFromOperator.isDistinctFrom(leftBlock, leftBlockPosition, rightBlock, rightBlockPosition);
assertEquals(hashStrategy.positionEqualsRow(leftBlockIndex, leftBlockPosition, rightBlockPosition, new Page(rightBlock)), expected);
assertEquals(hashStrategy.positionNotDistinctFromRow(leftBlockIndex, leftBlockPosition, rightBlockPosition, new Page(rightBlock)), expectedNotDistinct);
assertEquals(hashStrategy.rowEqualsRow(leftBlockPosition, new Page(leftBlock), rightBlockPosition, new Page(rightBlock)), expected);
assertEquals(hashStrategy.rowNotDistinctFromRow(leftBlockPosition, new Page(leftBlock), rightBlockPosition, new Page(rightBlock)), expectedNotDistinct);
assertEquals(hashStrategy.positionEqualsRowIgnoreNulls(leftBlockIndex, leftBlockPosition, rightBlockPosition, new Page(rightBlock)), expected);
assertEquals(hashStrategy.positionEqualsPositionIgnoreNulls(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expected);
assertEquals(hashStrategy.positionEqualsPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expected);
assertEquals(hashStrategy.positionNotDistinctFromPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expectedNotDistinct);
}
}
// check equality of every position against every other position in the block cursor
for (int rightBlockIndex = 0; rightBlockIndex < channel.size(); rightBlockIndex++) {
Block rightBlock = channel.get(rightBlockIndex);
for (int rightBlockPosition = 0; rightBlockPosition < rightBlock.getPositionCount(); rightBlockPosition++) {
boolean expected = equalOperator.equalNullSafe(leftBlock, leftBlockPosition, rightBlock, rightBlockPosition);
boolean expectedNotDistinct = !distinctFromOperator.isDistinctFrom(leftBlock, leftBlockPosition, rightBlock, rightBlockPosition);
assertEquals(hashStrategy.positionEqualsRow(leftBlockIndex, leftBlockPosition, rightBlockPosition, new Page(rightBlock)), expected);
assertEquals(hashStrategy.positionNotDistinctFromRow(leftBlockIndex, leftBlockPosition, rightBlockPosition, new Page(rightBlock)), expectedNotDistinct);
assertEquals(hashStrategy.rowEqualsRow(leftBlockPosition, new Page(leftBlock), rightBlockPosition, new Page(rightBlock)), expected);
assertEquals(hashStrategy.rowNotDistinctFromRow(leftBlockPosition, new Page(leftBlock), rightBlockPosition, new Page(rightBlock)), expectedNotDistinct);
assertEquals(hashStrategy.positionEqualsRowIgnoreNulls(leftBlockIndex, leftBlockPosition, rightBlockPosition, new Page(rightBlock)), expected);
assertEquals(hashStrategy.positionEqualsPositionIgnoreNulls(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expected);
assertEquals(hashStrategy.positionEqualsPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expected);
assertEquals(hashStrategy.positionNotDistinctFromPosition(leftBlockIndex, leftBlockPosition, rightBlockIndex, rightBlockPosition), expectedNotDistinct);
}
}
// write position to output block
pageBuilder.declarePosition();
hashStrategy.appendTo(leftBlockIndex, leftBlockPosition, pageBuilder, 0);
}
// verify output block matches
assertBlockEquals(VARCHAR, pageBuilder.build().getBlock(0), leftBlock);
}
}
use of io.trino.spi.PageBuilder in project trino by trinodb.
the class TestDeltaLakePageSink method testPageSinkStats.
@Test
public void testPageSinkStats() throws Exception {
File tempDir = Files.createTempDir();
try {
DeltaLakeWriterStats stats = new DeltaLakeWriterStats();
String tablePath = tempDir.getAbsolutePath() + "/test_table";
ConnectorPageSink pageSink = createPageSink(new Path(tablePath), stats);
List<LineItemColumn> columns = ImmutableList.copyOf(LineItemColumn.values());
List<Type> columnTypes = columns.stream().map(LineItemColumn::getType).map(TestDeltaLakePageSink::getTrinoType).collect(toList());
PageBuilder pageBuilder = new PageBuilder(columnTypes);
long rows = 0;
for (LineItem lineItem : new LineItemGenerator(0.01, 1, 1)) {
if (rows >= NUM_ROWS) {
break;
}
rows++;
pageBuilder.declarePosition();
for (int i = 0; i < columns.size(); i++) {
LineItemColumn column = columns.get(i);
BlockBuilder blockBuilder = pageBuilder.getBlockBuilder(i);
writeToBlock(blockBuilder, column, lineItem);
}
}
Page page = pageBuilder.build();
pageSink.appendPage(page);
JsonCodec<DataFileInfo> dataFileInfoCodec = new JsonCodecFactory().jsonCodec(DataFileInfo.class);
Collection<Slice> fragments = getFutureValue(pageSink.finish());
List<DataFileInfo> dataFileInfos = fragments.stream().map(Slice::getBytes).map(dataFileInfoCodec::fromJson).collect(toImmutableList());
assertEquals(dataFileInfos.size(), 1);
DataFileInfo dataFileInfo = dataFileInfos.get(0);
List<File> files = ImmutableList.copyOf(new File(tablePath).listFiles((dir, name) -> !name.endsWith(".crc")));
assertEquals(files.size(), 1);
File outputFile = files.get(0);
assertEquals(round(stats.getInputPageSizeInBytes().getAllTime().getMax()), page.getRetainedSizeInBytes());
assertEquals(dataFileInfo.getStatistics().getNumRecords(), Optional.of(rows));
assertEquals(dataFileInfo.getPartitionValues(), ImmutableList.of());
assertEquals(dataFileInfo.getSize(), outputFile.length());
assertEquals(dataFileInfo.getPath(), outputFile.getName());
Instant now = Instant.now();
assertTrue(dataFileInfo.getCreationTime() < now.toEpochMilli());
assertTrue(dataFileInfo.getCreationTime() > now.minus(1, MINUTES).toEpochMilli());
} finally {
deleteRecursively(tempDir.toPath(), ALLOW_INSECURE);
}
}
use of io.trino.spi.PageBuilder in project trino by trinodb.
the class JoinCompiler method generateAppendToMethod.
private static void generateAppendToMethod(ClassDefinition classDefinition, CallSiteBinder callSiteBinder, List<Type> types, List<Integer> outputChannels, List<FieldDefinition> channelFields) {
Parameter blockIndex = arg("blockIndex", int.class);
Parameter blockPosition = arg("blockPosition", int.class);
Parameter pageBuilder = arg("pageBuilder", PageBuilder.class);
Parameter outputChannelOffset = arg("outputChannelOffset", int.class);
MethodDefinition appendToMethod = classDefinition.declareMethod(a(PUBLIC), "appendTo", type(void.class), blockIndex, blockPosition, pageBuilder, outputChannelOffset);
Variable thisVariable = appendToMethod.getThis();
BytecodeBlock appendToBody = appendToMethod.getBody();
int pageBuilderOutputChannel = 0;
for (int outputChannel : outputChannels) {
Type type = types.get(outputChannel);
BytecodeExpression typeExpression = constantType(callSiteBinder, type);
BytecodeExpression block = thisVariable.getField(channelFields.get(outputChannel)).invoke("get", Object.class, blockIndex).cast(Block.class);
appendToBody.comment("%s.appendTo(channel_%s.get(outputChannel), blockPosition, pageBuilder.getBlockBuilder(outputChannelOffset + %s));", type.getClass(), outputChannel, pageBuilderOutputChannel).append(typeExpression).append(block).append(blockPosition).append(pageBuilder).append(outputChannelOffset).push(pageBuilderOutputChannel++).append(OpCode.IADD).invokeVirtual(PageBuilder.class, "getBlockBuilder", BlockBuilder.class, int.class).invokeInterface(Type.class, "appendTo", void.class, Block.class, int.class, BlockBuilder.class);
}
appendToBody.ret();
}
Aggregations