anuragmantri commented on code in PR #14948: URL: https://github.com/apache/iceberg/pull/14948#discussion_r4201225230
########## spark/v4.1/spark/src/main/java/org/apache/iceberg/spark/source/MergingSortedRowDataReader.java: ########## @@ -0,0 +1,361 @@ +/* + * 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.iceberg.spark.source; + +import java.io.IOException; +import java.io.UncheckedIOException; +import java.util.Comparator; +import java.util.List; +import java.util.Objects; +import java.util.Set; +import org.apache.iceberg.BaseScanTaskGroup; +import org.apache.iceberg.FileScanTask; +import org.apache.iceberg.ScanTaskGroup; +import org.apache.iceberg.Schema; +import org.apache.iceberg.SortField; +import org.apache.iceberg.SortOrder; +import org.apache.iceberg.SortOrderComparators; +import org.apache.iceberg.StructLike; +import org.apache.iceberg.Table; +import org.apache.iceberg.io.CloseableGroup; +import org.apache.iceberg.io.CloseableIterable; +import org.apache.iceberg.io.CloseableIterator; +import org.apache.iceberg.io.FileIO; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.spark.SparkSchemaUtil; +import org.apache.iceberg.spark.source.metrics.TaskNumDeletes; +import org.apache.iceberg.spark.source.metrics.TaskNumSplits; +import org.apache.iceberg.types.Type; +import org.apache.iceberg.types.TypeUtil; +import org.apache.iceberg.types.Types; +import org.apache.iceberg.util.Exceptions; +import org.apache.iceberg.util.SortedMerge; +import org.apache.spark.rdd.InputFileBlockHolder; +import org.apache.spark.sql.catalyst.InternalRow; +import org.apache.spark.sql.catalyst.ProjectingInternalRow; +import org.apache.spark.sql.catalyst.expressions.UnsafeProjection; +import org.apache.spark.sql.connector.metric.CustomTaskMetric; +import org.apache.spark.sql.connector.read.PartitionReader; +import org.apache.spark.sql.types.StructType; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; +import scala.collection.immutable.Range; + +/** + * Returns the rows of a task group's files as one stream in the files' shared sort order. + * + * <p>All files must have the same sort order. Sort orders with a UUID key, a floating point key + * followed by another key, or a nested key that is not projected are rejected. All files are open + * at once and each loads its own deletes. + */ +class MergingSortedRowDataReader implements PartitionReader<InternalRow> { + private static final Logger LOG = LoggerFactory.getLogger(MergingSortedRowDataReader.class); + + private final CloseableGroup resources; + private final CloseableIterator<TaggedRow> mergedIterator; + private final List<RowDataReader> fileReaders; + private final ProjectingInternalRow projectingRow; + private InternalRow current; + private FileBlock currentBlock; + + MergingSortedRowDataReader(SparkInputPartition partition) { + this( + partition.table(), + partition.io(), + partition.taskGroup(), + partition.projection(), + partition.isCaseSensitive(), + partition.cacheDeleteFilesOnExecutors()); + } + + MergingSortedRowDataReader( + Table table, + FileIO io, + ScanTaskGroup<FileScanTask> taskGroup, + Schema projection, + boolean caseSensitive, + boolean cacheDeleteFilesOnExecutors) { + List<FileScanTask> tasks = Lists.newArrayList(taskGroup.tasks()); + int numTasks = tasks.size(); + + Preconditions.checkArgument( + numTasks > 1, "Merging reader requires at least two tasks, got %s", numTasks); + + Integer expectedOrderId = tasks.get(0).file().sortOrderId(); + Preconditions.checkArgument( + expectedOrderId != null && expectedOrderId != SortOrder.unsorted().orderId(), + "Merging reader requires sorted files, got sort order %s", + expectedOrderId); + for (FileScanTask task : tasks) { + Preconditions.checkArgument( + Objects.equals(task.file().sortOrderId(), expectedOrderId), + "Cannot merge files with different sort orders: %s has sort order %s, expected %s", + task.file().location(), + task.file().sortOrderId(), + expectedOrderId); + } + + SortOrder sortOrder = table.sortOrders().get(expectedOrderId); Review Comment: Added the precondition back, along with a test for it. ########## spark/v4.1/spark/src/main/java/org/apache/iceberg/spark/source/MergingSortedRowDataReader.java: ########## @@ -0,0 +1,361 @@ +/* + * 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.iceberg.spark.source; + +import java.io.IOException; +import java.io.UncheckedIOException; +import java.util.Comparator; +import java.util.List; +import java.util.Objects; +import java.util.Set; +import org.apache.iceberg.BaseScanTaskGroup; +import org.apache.iceberg.FileScanTask; +import org.apache.iceberg.ScanTaskGroup; +import org.apache.iceberg.Schema; +import org.apache.iceberg.SortField; +import org.apache.iceberg.SortOrder; +import org.apache.iceberg.SortOrderComparators; +import org.apache.iceberg.StructLike; +import org.apache.iceberg.Table; +import org.apache.iceberg.io.CloseableGroup; +import org.apache.iceberg.io.CloseableIterable; +import org.apache.iceberg.io.CloseableIterator; +import org.apache.iceberg.io.FileIO; +import org.apache.iceberg.relocated.com.google.common.base.Preconditions; +import org.apache.iceberg.relocated.com.google.common.collect.ImmutableList; +import org.apache.iceberg.relocated.com.google.common.collect.Lists; +import org.apache.iceberg.relocated.com.google.common.collect.Sets; +import org.apache.iceberg.spark.SparkSchemaUtil; +import org.apache.iceberg.spark.source.metrics.TaskNumDeletes; +import org.apache.iceberg.spark.source.metrics.TaskNumSplits; +import org.apache.iceberg.types.Type; +import org.apache.iceberg.types.TypeUtil; +import org.apache.iceberg.types.Types; +import org.apache.iceberg.util.Exceptions; +import org.apache.iceberg.util.SortedMerge; +import org.apache.spark.rdd.InputFileBlockHolder; +import org.apache.spark.sql.catalyst.InternalRow; +import org.apache.spark.sql.catalyst.ProjectingInternalRow; +import org.apache.spark.sql.catalyst.expressions.UnsafeProjection; +import org.apache.spark.sql.connector.metric.CustomTaskMetric; +import org.apache.spark.sql.connector.read.PartitionReader; +import org.apache.spark.sql.types.StructType; +import org.slf4j.Logger; +import org.slf4j.LoggerFactory; +import scala.collection.immutable.Range; + +/** + * Returns the rows of a task group's files as one stream in the files' shared sort order. + * + * <p>All files must have the same sort order. Sort orders with a UUID key, a floating point key + * followed by another key, or a nested key that is not projected are rejected. All files are open + * at once and each loads its own deletes. + */ +class MergingSortedRowDataReader implements PartitionReader<InternalRow> { + private static final Logger LOG = LoggerFactory.getLogger(MergingSortedRowDataReader.class); + + private final CloseableGroup resources; + private final CloseableIterator<TaggedRow> mergedIterator; + private final List<RowDataReader> fileReaders; + private final ProjectingInternalRow projectingRow; + private InternalRow current; + private FileBlock currentBlock; + + MergingSortedRowDataReader(SparkInputPartition partition) { + this( + partition.table(), + partition.io(), + partition.taskGroup(), + partition.projection(), + partition.isCaseSensitive(), + partition.cacheDeleteFilesOnExecutors()); + } + + MergingSortedRowDataReader( + Table table, + FileIO io, + ScanTaskGroup<FileScanTask> taskGroup, + Schema projection, + boolean caseSensitive, + boolean cacheDeleteFilesOnExecutors) { + List<FileScanTask> tasks = Lists.newArrayList(taskGroup.tasks()); + int numTasks = tasks.size(); + + Preconditions.checkArgument( + numTasks > 1, "Merging reader requires at least two tasks, got %s", numTasks); + + Integer expectedOrderId = tasks.get(0).file().sortOrderId(); + Preconditions.checkArgument( + expectedOrderId != null && expectedOrderId != SortOrder.unsorted().orderId(), + "Merging reader requires sorted files, got sort order %s", + expectedOrderId); + for (FileScanTask task : tasks) { + Preconditions.checkArgument( + Objects.equals(task.file().sortOrderId(), expectedOrderId), + "Cannot merge files with different sort orders: %s has sort order %s, expected %s", + task.file().location(), + task.file().sortOrderId(), + expectedOrderId); + } + + SortOrder sortOrder = table.sortOrders().get(expectedOrderId); + + LOG.debug( + "Creating merging reader for {} tasks with sort order {} in table {}", + numTasks, + expectedOrderId, + table.name()); + + Schema mergeReadSchema = mergeReadSchema(projection, sortOrder, table); + this.projectingRow = buildProjectingRow(projection); + UnsafeProjection deepCopyProjection = + UnsafeProjection.create(SparkSchemaUtil.convert(mergeReadSchema)); + + this.resources = new CloseableGroup(); + resources.setSuppressCloseFailure(true); + this.fileReaders = + tasks.stream() + .map( + task -> + new RowDataReader( + table, + io, + new BaseScanTaskGroup<>(ImmutableList.of(task)), + mergeReadSchema, + caseSensitive, + cacheDeleteFilesOnExecutors)) + .toList(); + fileReaders.forEach(resources::addCloseable); + + List<CloseableIterable<TaggedRow>> fileIterables = Lists.newArrayListWithCapacity(tasks.size()); + for (int i = 0; i < tasks.size(); i++) { + fileIterables.add( + new TaggedRowIterable(fileReaders.get(i), tasks.get(i), deepCopyProjection)); + } + Comparator<InternalRow> rowComparator = buildComparator(mergeReadSchema, sortOrder); + SortedMerge<TaggedRow> sortedMerge = + new SortedMerge<>((a, b) -> rowComparator.compare(a.row(), b.row()), fileIterables); + resources.addCloseable(sortedMerge); + boolean threw = true; + try { + this.mergedIterator = sortedMerge.iterator(); + threw = false; + } finally { + if (threw) { + Exceptions.close(resources, true); + } + } + } + + private static class TaggedRowIterable implements CloseableIterable<TaggedRow> { + private final RowDataReader reader; + private final UnsafeProjection deepCopyProjection; + private final FileBlock block; + + private TaggedRowIterable( + RowDataReader reader, FileScanTask task, UnsafeProjection deepCopyProjection) { + this.reader = reader; + this.deepCopyProjection = deepCopyProjection; + this.block = new FileBlock(task.file().location(), task.start(), task.length()); + } + + @Override + public CloseableIterator<TaggedRow> iterator() { + return new TaggedRowIterator(reader, deepCopyProjection, block); + } + + @Override + public void close() {} + } + + private static class TaggedRowIterator implements CloseableIterator<TaggedRow> { + private final RowDataReader reader; + private final UnsafeProjection deepCopyProjection; + private final FileBlock block; + private boolean advanced = false; + private boolean hasNext = false; + + private TaggedRowIterator( + RowDataReader reader, UnsafeProjection deepCopyProjection, FileBlock block) { + this.reader = reader; + this.deepCopyProjection = deepCopyProjection; + this.block = block; + } + + @Override + public boolean hasNext() { + if (!advanced) { + try { + hasNext = reader.next(); + advanced = true; + } catch (IOException e) { + throw new UncheckedIOException("Failed to advance reader", e); + } + } + return hasNext; + } + + @Override + public TaggedRow next() { + if (!advanced) { Review Comment: Good point. `next()` now just calls `hasNext()` and throws `NoSuchElementException` when nothing is left, so it also no longer returns a stale row if it is called past the end. -- 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]
