smiklosovic commented on code in PR #4373: URL: https://github.com/apache/cassandra/pull/4373#discussion_r2341547735
########## src/java/org/apache/cassandra/db/virtual/AbstractLazyVirtualTable.java: ########## @@ -0,0 +1,653 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package org.apache.cassandra.db.virtual; + +import java.nio.ByteBuffer; +import java.util.Arrays; +import java.util.Comparator; +import java.util.HashMap; +import java.util.Iterator; +import java.util.Map; +import java.util.NavigableMap; +import java.util.TreeMap; +import java.util.concurrent.TimeUnit; +import java.util.function.Consumer; +import java.util.function.Function; +import java.util.function.UnaryOperator; + +import javax.annotation.Nullable; + +import accord.utils.Invariants; +import org.apache.cassandra.config.DatabaseDescriptor; +import org.apache.cassandra.db.Clusterable; +import org.apache.cassandra.db.Clustering; +import org.apache.cassandra.db.ConsistencyLevel; +import org.apache.cassandra.db.DataRange; +import org.apache.cassandra.db.DecoratedKey; +import org.apache.cassandra.db.DeletionTime; +import org.apache.cassandra.db.LivenessInfo; +import org.apache.cassandra.db.RegularAndStaticColumns; +import org.apache.cassandra.db.filter.ClusteringIndexFilter; +import org.apache.cassandra.db.filter.ColumnFilter; +import org.apache.cassandra.db.filter.DataLimits; +import org.apache.cassandra.db.filter.RowFilter; +import org.apache.cassandra.db.marshal.AbstractType; +import org.apache.cassandra.db.marshal.CompositeType; +import org.apache.cassandra.db.partitions.PartitionUpdate; +import org.apache.cassandra.db.partitions.UnfilteredPartitionIterator; +import org.apache.cassandra.db.rows.BTreeRow; +import org.apache.cassandra.db.rows.BufferCell; +import org.apache.cassandra.db.rows.ColumnData; +import org.apache.cassandra.db.rows.EncodingStats; +import org.apache.cassandra.db.rows.Row; +import org.apache.cassandra.db.rows.Unfiltered; +import org.apache.cassandra.db.rows.UnfilteredRowIterator; +import org.apache.cassandra.dht.AbstractBounds; +import org.apache.cassandra.dht.Bounds; +import org.apache.cassandra.exceptions.InvalidRequestException; +import org.apache.cassandra.exceptions.ReadTimeoutException; +import org.apache.cassandra.schema.ColumnMetadata; +import org.apache.cassandra.schema.TableMetadata; +import org.apache.cassandra.utils.BulkIterator; +import org.apache.cassandra.utils.Clock; +import org.apache.cassandra.utils.FBUtilities; +import org.apache.cassandra.utils.btree.BTree; +import org.apache.cassandra.utils.btree.UpdateFunction; + +import static org.apache.cassandra.db.ClusteringPrefix.Kind.STATIC_CLUSTERING; +import static org.apache.cassandra.utils.Clock.Global.nanoTime; + +/** + * An abstract virtual table implementation that builds the resultset on demand. + */ +public abstract class AbstractLazyVirtualTable implements VirtualTable +{ + // in the special case where we know we have enough rows in the collector, throw this exception to terminate early + static class InternalDoneException extends RuntimeException {} + + public static class FilterRange<V> + { + final V min, max; + public FilterRange(V min, V max) + { + this.min = min; + this.max = max; + } + } + + public interface PartitionsCollector + { + DataRange dataRange(); + RowFilter rowFilter(); + RowCollector row(Object... primaryKeys); + PartitionCollector partition(Object... partitionKeys); + UnfilteredPartitionIterator finish(); + + default @Nullable DecoratedKey singleKey() + { + AbstractBounds<?> bounds = dataRange().keyRange(); + if (!bounds.isStartInclusive() || !bounds.isEndInclusive() || !bounds.left.equals(bounds.right) || !(bounds.left instanceof DecoratedKey)) + return null; + + return (DecoratedKey) bounds.left; + } + + <I, O> FilterRange<O> filters(String column, Function<I, O> translate, UnaryOperator<O> increment, UnaryOperator<O> decrement); + } + + public interface PartitionCollector + { + RowCollector row(Object... clusteringKeys); + } + + public interface RowCollector + { + default PartitionsCollector lazyAdd(Consumer<ColumnsCollector> addToIfNeeded) { return eagerAdd(addToIfNeeded); } + PartitionsCollector eagerAdd(Consumer<ColumnsCollector> addToNow); + } + + public interface ColumnsCollector + { + <V> ColumnsCollector add(String columnName, V value, Function<V, Object> transform); + default ColumnsCollector add(String columnName, Object value) { return add(columnName, value, Function.identity()); } + } + + public static class SimplePartitionsCollector implements PartitionsCollector + { + final TableMetadata metadata; + final boolean isSorted; + + final Map<String, ColumnMetadata> columnLookup = new HashMap<>(); + final NavigableMap<DecoratedKey, SimplePartition> partitions; + + final DataRange dataRange; + final ColumnFilter columnFilter; + final RowFilter rowFilter; + final DataLimits limits; + + final long startedAt = Clock.Global.nanoTime(); + final long timeoutAt; + + final long nowInSeconds = Clock.Global.nowInSeconds(); + final long timestamp; + + int totalRowCount; + int lastFilteredTotalRowCount; + + @Override public DataRange dataRange() { return dataRange; } + @Override public RowFilter rowFilter() { return rowFilter; } + public ColumnFilter columnFilter() { return columnFilter; } + + public SimplePartitionsCollector(TableMetadata metadata, boolean isSorted, DataRange dataRange, ColumnFilter columnFilter, RowFilter rowFilter, DataLimits limits) + { + this.metadata = metadata; + this.isSorted = isSorted; + this.dataRange = dataRange; + this.columnFilter = columnFilter; + this.rowFilter = rowFilter; + this.limits = limits; + this.timestamp = FBUtilities.timestampMicros(); + this.timeoutAt = startedAt + DatabaseDescriptor.getReadRpcTimeout(TimeUnit.NANOSECONDS); + this.partitions = new TreeMap<>(dataRange.isReversed() ? DecoratedKey.comparator.reversed() : DecoratedKey.comparator); + for (ColumnMetadata cm : metadata.columns()) + columnLookup.put(cm.name.toString(), cm); + } + + @Override + public PartitionCollector partition(Object ... partitionKeys) + { + int pkSize = metadata.partitionKeyColumns().size(); + if (pkSize != partitionKeys.length) + throw new IllegalArgumentException(); + + DecoratedKey partitionKey = makeDecoratedKey(partitionKeys); + if (!dataRange.contains(partitionKey)) + return dropCks -> dropRow -> this; + + return partitions.computeIfAbsent(partitionKey, SimplePartition::new); + } + + @Override + public UnfilteredPartitionIterator finish() + { + final Iterator<SimplePartition> partitions = this.partitions.values().iterator(); + return new UnfilteredPartitionIterator() + { + @Override public TableMetadata metadata() { return metadata; } + @Override public void close() {} + + @Override + public boolean hasNext() + { + return partitions.hasNext(); + } + + @Override + public UnfilteredRowIterator next() + { + SimplePartition partition = partitions.next(); + Iterator<Row> rows = partition.rows(); + + return new UnfilteredRowIterator() + { + @Override public TableMetadata metadata() { return metadata; } + @Override public boolean isReverseOrder() { return dataRange.isReversed(); } + @Override public RegularAndStaticColumns columns() { return columnFilter.fetchedColumns(); } + @Override public DecoratedKey partitionKey() { return partition.key; } + + @Override public Row staticRow() { return partition.staticRow(); } + @Override public boolean hasNext() { return rows.hasNext(); } + @Override public Unfiltered next() { return rows.next(); } + + @Override public void close() {} + @Override public DeletionTime partitionLevelDeletion() { return DeletionTime.LIVE; } + @Override public EncodingStats stats() { return EncodingStats.NO_STATS; } + }; + } + }; + } + + @Override + @Nullable + public <I, O> FilterRange<O> filters(String columnName, Function<I, O> translate, UnaryOperator<O> increment, UnaryOperator<O> decrement) + { + ColumnMetadata column = columnLookup.get(columnName); + O min = null, max = null; + for (RowFilter.Expression expression : rowFilter().getExpressions()) + { + if (!expression.column().equals(column)) + continue; + + if (expression.isCustom()) + continue; + + O bound = translate.apply((I)column.type.compose(expression.getIndexValue())); + switch (expression.operator()) + { + default: continue; + case EQ: min = max = bound; break; + case LTE: max = bound; break; + case LT: max = decrement.apply(bound); break; + case GTE: min = bound; break; + case GT: min = increment.apply(bound); break; + } + } + + return new FilterRange<>(min, max); + } + + @Override + public RowCollector row(Object... primaryKeys) + { + int pkSize = metadata.partitionKeyColumns().size(); + int ckSize = metadata.clusteringColumns().size(); + if (pkSize + ckSize != primaryKeys.length) + throw new IllegalArgumentException(); + + Object[] partitionKeyValues = new Object[pkSize]; + Object[] clusteringValues = new Object[ckSize]; + + System.arraycopy(primaryKeys, 0, partitionKeyValues, 0, pkSize); + System.arraycopy(primaryKeys, pkSize, clusteringValues, 0, ckSize); + + DecoratedKey partitionKey = makeDecoratedKey(partitionKeyValues); + Clustering<?> clustering = makeClustering(clusteringValues); + + if (!dataRange.contains(partitionKey) || !dataRange.clusteringIndexFilter(partitionKey).selects(clustering)) + return drop -> this; + + return partitions.computeIfAbsent(partitionKey, SimplePartition::new).row(); + } + + private DecoratedKey makeDecoratedKey(Object... partitionKeyValues) + { + ByteBuffer partitionKey = partitionKeyValues.length == 1 + ? decompose(metadata.partitionKeyType, partitionKeyValues[0]) + : ((CompositeType) metadata.partitionKeyType).decompose(partitionKeyValues); + return metadata.partitioner.decorateKey(partitionKey); + } + + private Clustering<?> makeClustering(Object... clusteringValues) + { + if (clusteringValues.length == 0) + return Clustering.EMPTY; + + ByteBuffer[] clusteringByteBuffers = new ByteBuffer[clusteringValues.length]; + for (int i = 0; i < clusteringValues.length; i++) + clusteringByteBuffers[i] = decompose(metadata.clusteringColumns().get(i).type, clusteringValues[i]); + return Clustering.make(clusteringByteBuffers); + } + + private final class SimplePartition implements PartitionCollector + { + private final DecoratedKey key; + // we assume no duplicate rows, and impose the condition lazily + private SimpleRow[] rows; + private int rowCount; + private SimpleRow staticRow; + private boolean dropRows; + + private SimplePartition(DecoratedKey key) + { + this.key = key; + this.rows = new SimpleRow[1]; + } + + @Override + public RowCollector row(Object... clusteringKeys) + { + int ckSize = metadata.clusteringColumns().size(); + if (ckSize != clusteringKeys.length) + throw new IllegalArgumentException(); + + return row(makeClustering(clusteringKeys)); + } + + RowCollector row(Clustering<?> clustering) + { + if (nanoTime() > timeoutAt) + throw new ReadTimeoutException(ConsistencyLevel.ONE, 0, 1, false); + + if (dropRows || !dataRange.clusteringIndexFilter(key).selects(clustering)) + return drop -> SimplePartitionsCollector.this; + + if (totalRowCount >= limits.count()) + { + boolean filter; + if (!isSorted) filter = totalRowCount / 2 >= Math.max(1024, limits.count()); + else + { + int rowsAddedSinceLastFiltered = totalRowCount - lastFilteredTotalRowCount; + int threshold = Math.max(32, Math.min(1024, lastFilteredTotalRowCount / 2)); + filter = lastFilteredTotalRowCount == 0 || rowsAddedSinceLastFiltered >= threshold; + } + + if (filter) + { + // first filter within each partition + for (SimplePartition partition : partitions.values()) + { + int curCount = partition.rowCount; + int newCount = Math.min(curCount, limits.perPartitionCount()); + newCount = partition.filterAndSortAndTruncate(newCount); + totalRowCount -= curCount - newCount; + } + + // then drop any partitions that completely fall outside our limit + Iterator<SimplePartition> iter = partitions.descendingMap().values().iterator(); + SimplePartition last; + while (true) + { + SimplePartition next = last = iter.next(); + if (totalRowCount - next.rowCount < limits.count()) + break; + + iter.remove(); + totalRowCount -= next.rowCount; + if (next == this) + dropRows = true; + } + + // possibly truncate the last partition if it partially falls outside the limit + int overflow = Math.max(0, totalRowCount - limits.count()); + int curCount = last.rowCount; + int newCount = curCount - overflow; + newCount = last.filterAndSortAndTruncate(newCount); + totalRowCount -= curCount - newCount; + lastFilteredTotalRowCount = totalRowCount; + + if (isSorted && totalRowCount >= limits.count()) + throw new InternalDoneException(); + + if (dropRows) + return drop -> SimplePartitionsCollector.this; + } + } + + SimpleRow result = new SimpleRow(clustering); + if (clustering.kind() == STATIC_CLUSTERING) + { + Invariants.require(staticRow == null); + staticRow = result; + } + else + { + totalRowCount++; + if (rowCount == rows.length) + rows = Arrays.copyOf(rows, Math.max(8, rowCount * 2)); + rows[rowCount++] = result; + } + return result; + } + + void filterAndSort() + { + int newCount = 0; + for (int i = 0 ; i < rowCount; ++i) + { + if (rows[i].rowFilterIncludes()) + { + if (newCount != i) + rows[newCount] = rows[i]; + newCount++; + } + } + totalRowCount -= (rowCount - newCount); + Arrays.fill(rows, newCount, rowCount, null); + rowCount = newCount; + Arrays.sort(rows, 0, newCount, rowComparator()); + } + + int filterAndSortAndTruncate(int newCount) + { + Invariants.requireArgument(newCount <= rowCount); + filterAndSort(); + if (rowCount < newCount) + return rowCount; + + Arrays.fill(rows, newCount, rowCount, null); + rowCount = newCount; + return newCount; + } + + private Comparator<SimpleRow> rowComparator() + { + Comparator<Clusterable> cmp = dataRange.isReversed() ? metadata.comparator.reversed() : metadata.comparator; + return (a, b) -> cmp.compare(a.clustering, b.clustering); + } + + Row staticRow() + { + if (staticRow == null) + return null; + + return staticRow.materialiseAndFilter(); + } + + Iterator<Row> rows() + { + filterAndSort(); + return Arrays.stream(rows, 0, rowCount).map(SimpleRow::materialiseAndFilter).iterator(); + } + + private final class SimpleRow implements RowCollector + { + final Clustering<?> clustering; + SomeColumns state; + + private SimpleRow(Clustering<?> clustering) + { + this.clustering = clustering; + } + + @Override + public PartitionsCollector lazyAdd(Consumer<ColumnsCollector> addToIfNeeded) + { + Invariants.require(state == null); + state = new LazyColumnsCollector(addToIfNeeded); + return SimplePartitionsCollector.this; + } + + @Override + public PartitionsCollector eagerAdd(Consumer<ColumnsCollector> addToNow) + { + Invariants.require(state == null); + state = new EagerColumnsCollector(addToNow); + return SimplePartitionsCollector.this; + } + + boolean rowFilterIncludes() + { + return null != materialiseAndFilter(); + } + + Row materialiseAndFilter() + { + if (state == null) + return null; + + FilteredRow filtered = state.materialiseAndFilter(this); + state = filtered; + return filtered == null ? null : filtered.row; + } + + DecoratedKey partitionKey() + { + return SimplePartition.this.key; + } + + SimplePartitionsCollector collector() + { + return SimplePartitionsCollector.this; + } + } + } + + static abstract class SomeColumns + { + abstract FilteredRow materialiseAndFilter(SimplePartition.SimpleRow parent); + } + + static class LazyColumnsCollector extends SomeColumns + { + final Consumer<ColumnsCollector> lazy; + LazyColumnsCollector(Consumer<ColumnsCollector> lazy) + { + this.lazy = lazy; + } + + @Override + FilteredRow materialiseAndFilter(SimplePartition.SimpleRow parent) + { + return parent.collector().new EagerColumnsCollector(lazy).materialiseAndFilter(parent); + } + } + + class EagerColumnsCollector extends SomeColumns implements ColumnsCollector + { + Object[] columns = new Object[4]; + int columnCount; + + public EagerColumnsCollector(Consumer<ColumnsCollector> add) + { + add.accept(this); + } + + @Override + public <V> ColumnsCollector add(String name, V input, Function<V, Object> f) + { + ColumnMetadata cm = columnLookup.get(name); + if (!columnFilter.fetches(cm)) Review Comment: _So the equivalent here would be calling columns.add("ballot", txn.ballot(), AccordDebugKeyspace::toStringOrNull)_ Aha right, I see ... basically the "callable" in my comment is conceptually the same as the transformation function itself. Yeah ... While it might be theoretically expensive to even call `txn.ballot()` (replace with any other arbitrary object / method which needs to be passed / called), I can see that the current code is just fine for the common usage. These columns just hold simple objects / strings / numbers etc. so passing them is basically for free. I leave your suggestion in the last sentence to your judgement. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
