yihua commented on code in PR #10954: URL: https://github.com/apache/hudi/pull/10954#discussion_r1561624097
########## hudi-spark-datasource/hudi-spark3.2.x/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/Spark32HoodieParquetReader.scala: ########## @@ -0,0 +1,267 @@ +/* + * 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.spark.sql.execution.datasources.parquet + +import org.apache.hadoop.conf.Configuration +import org.apache.hadoop.fs.Path +import org.apache.hadoop.mapred.FileSplit +import org.apache.hadoop.mapreduce._ +import org.apache.hadoop.mapreduce.task.TaskAttemptContextImpl +import org.apache.parquet.filter2.compat.FilterCompat +import org.apache.parquet.filter2.predicate.FilterApi +import org.apache.parquet.format.converter.ParquetMetadataConverter.SKIP_ROW_GROUPS +import org.apache.parquet.hadoop._ +import org.apache.spark.TaskContext +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.JoinedRow +import org.apache.spark.sql.catalyst.expressions.codegen.GenerateUnsafeProjection +import org.apache.spark.sql.catalyst.util.DateTimeUtils +import org.apache.spark.sql.execution.datasources._ +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.sources._ +import org.apache.spark.sql.types._ + +import java.net.URI + +class Spark32HoodieParquetReader(enableVectorizedReader: Boolean, Review Comment: ```suggestion class Spark32ParquetReader(enableVectorizedReader: Boolean, ``` ########## hudi-spark-datasource/hudi-spark3.4.x/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/Spark34HoodieParquetReader.scala: ########## @@ -0,0 +1,277 @@ +/* + * 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.spark.sql.execution.datasources.parquet + +import org.apache.hadoop.conf.Configuration +import org.apache.hadoop.mapred.FileSplit +import org.apache.hadoop.mapreduce._ +import org.apache.hadoop.mapreduce.task.TaskAttemptContextImpl +import org.apache.parquet.filter2.compat.FilterCompat +import org.apache.parquet.filter2.predicate.FilterApi +import org.apache.parquet.format.converter.ParquetMetadataConverter.SKIP_ROW_GROUPS +import org.apache.parquet.hadoop._ +import org.apache.spark.TaskContext +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.JoinedRow +import org.apache.spark.sql.catalyst.expressions.codegen.GenerateUnsafeProjection +import org.apache.spark.sql.catalyst.util.DateTimeUtils +import org.apache.spark.sql.execution.datasources._ +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.sources._ +import org.apache.spark.sql.types._ + +class Spark34HoodieParquetReader(enableVectorizedReader: Boolean, Review Comment: ```suggestion class Spark34ParquetReader(enableVectorizedReader: Boolean, ``` ########## hudi-spark-datasource/hudi-spark3.3.x/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/Spark33HoodieParquetReader.scala: ########## @@ -0,0 +1,268 @@ +/* + * 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.spark.sql.execution.datasources.parquet + + +import org.apache.hadoop.conf.Configuration +import org.apache.hadoop.fs.Path +import org.apache.hadoop.mapred.FileSplit +import org.apache.hadoop.mapreduce._ +import org.apache.hadoop.mapreduce.task.TaskAttemptContextImpl +import org.apache.parquet.filter2.compat.FilterCompat +import org.apache.parquet.filter2.predicate.FilterApi +import org.apache.parquet.format.converter.ParquetMetadataConverter.SKIP_ROW_GROUPS +import org.apache.parquet.hadoop._ +import org.apache.spark.TaskContext +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.JoinedRow +import org.apache.spark.sql.catalyst.expressions.codegen.GenerateUnsafeProjection +import org.apache.spark.sql.catalyst.util.DateTimeUtils +import org.apache.spark.sql.execution.datasources._ +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.sources._ +import org.apache.spark.sql.types._ + +import java.net.URI + +class Spark33HoodieParquetReader(enableVectorizedReader: Boolean, Review Comment: ```suggestion class Spark33ParquetReader(enableVectorizedReader: Boolean, ``` ########## hudi-spark-datasource/hudi-spark3.4.x/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/Spark34HoodieParquetReader.scala: ########## @@ -0,0 +1,277 @@ +/* + * 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.spark.sql.execution.datasources.parquet + +import org.apache.hadoop.conf.Configuration +import org.apache.hadoop.mapred.FileSplit +import org.apache.hadoop.mapreduce._ +import org.apache.hadoop.mapreduce.task.TaskAttemptContextImpl +import org.apache.parquet.filter2.compat.FilterCompat +import org.apache.parquet.filter2.predicate.FilterApi +import org.apache.parquet.format.converter.ParquetMetadataConverter.SKIP_ROW_GROUPS +import org.apache.parquet.hadoop._ +import org.apache.spark.TaskContext +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.JoinedRow +import org.apache.spark.sql.catalyst.expressions.codegen.GenerateUnsafeProjection +import org.apache.spark.sql.catalyst.util.DateTimeUtils +import org.apache.spark.sql.execution.datasources._ +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.sources._ +import org.apache.spark.sql.types._ + +class Spark34HoodieParquetReader(enableVectorizedReader: Boolean, + datetimeRebaseModeInRead: String, + int96RebaseModeInRead: String, + enableParquetFilterPushDown: Boolean, + pushDownDate: Boolean, + pushDownTimestamp: Boolean, + pushDownDecimal: Boolean, + pushDownInFilterThreshold: Int, + pushDownStringPredicate: Boolean, + isCaseSensitive: Boolean, + timestampConversion: Boolean, + enableOffHeapColumnVector: Boolean, + capacity: Int, + returningBatch: Boolean, + enableRecordFilter: Boolean, + timeZoneId: Option[String]) extends SparkHoodieParquetReaderBase( + enableVectorizedReader = enableVectorizedReader, + enableParquetFilterPushDown = enableParquetFilterPushDown, + pushDownDate = pushDownDate, + pushDownTimestamp = pushDownTimestamp, + pushDownDecimal = pushDownDecimal, + pushDownInFilterThreshold = pushDownInFilterThreshold, + isCaseSensitive = isCaseSensitive, + timestampConversion = timestampConversion, + enableOffHeapColumnVector = enableOffHeapColumnVector, + capacity = capacity, + returningBatch = returningBatch, + enableRecordFilter = enableRecordFilter, + timeZoneId = timeZoneId) { + + /** + * Read an individual parquet file + * Code from ParquetFileFormat#buildReaderWithPartitionValues from Spark v3.4.2 adapted here + * + * @param file parquet file to read + * @param requiredSchema desired output schema of the data + * @param partitionSchema schema of the partition columns. Partition values will be appended to the end of every row + * @param filters filters for data skipping. Not guaranteed to be used; the spark plan will also apply the filters. + * @param sharedConf the hadoop conf + * @return iterator of rows read from the file output type says [[InternalRow]] but could be [[ColumnarBatch]] + */ + protected def doRead(file: PartitionedFile, + requiredSchema: StructType, + partitionSchema: StructType, + filters: Seq[Filter], + sharedConf: Configuration): Iterator[InternalRow] = { + assert(file.partitionValues.numFields == partitionSchema.size) + + val filePath = file.toPath + val split = new FileSplit(filePath, file.start, file.length, Array.empty[String]) + + + lazy val footerFileMetaData = + ParquetFooterReader.readFooter(sharedConf, filePath, SKIP_ROW_GROUPS).getFileMetaData + val datetimeRebaseSpec = DataSourceUtils.datetimeRebaseSpec( + footerFileMetaData.getKeyValueMetaData.get, + datetimeRebaseModeInRead) + // Try to push down filters when filter push-down is enabled. + val pushed = if (enableParquetFilterPushDown) { + val parquetSchema = footerFileMetaData.getSchema + val parquetFilters = new ParquetFilters( + parquetSchema, + pushDownDate, + pushDownTimestamp, + pushDownDecimal, + pushDownStringPredicate, + pushDownInFilterThreshold, + isCaseSensitive, + datetimeRebaseSpec) + filters + // Collects all converted Parquet filter predicates. Notice that not all predicates can be + // converted (`ParquetFilters.createFilter` returns an `Option`). That's why a `flatMap` + // is used here. + .flatMap(parquetFilters.createFilter(_)) + .reduceOption(FilterApi.and) + } else { + None + } + + // PARQUET_INT96_TIMESTAMP_CONVERSION says to apply timezone conversions to int96 timestamps' + // *only* if the file was created by something other than "parquet-mr", so check the actual + // writer here for this file. We have to do this per-file, as each file in the table may + // have different writers. + // Define isCreatedByParquetMr as function to avoid unnecessary parquet footer reads. + def isCreatedByParquetMr: Boolean = + footerFileMetaData.getCreatedBy().startsWith("parquet-mr") + + val convertTz = + if (timestampConversion && !isCreatedByParquetMr) { + Some(DateTimeUtils.getZoneId(sharedConf.get(SQLConf.SESSION_LOCAL_TIMEZONE.key))) + } else { + None + } + + val int96RebaseSpec = DataSourceUtils.int96RebaseSpec( + footerFileMetaData.getKeyValueMetaData.get, + int96RebaseModeInRead) + + val attemptId = new TaskAttemptID(new TaskID(new JobID(), TaskType.MAP, 0), 0) + val hadoopAttemptContext = + new TaskAttemptContextImpl(sharedConf, attemptId) + + // Try to push down filters when filter push-down is enabled. + // Notice: This push-down is RowGroups level, not individual records. + if (pushed.isDefined) { + ParquetInputFormat.setFilterPredicate(hadoopAttemptContext.getConfiguration, pushed.get) + } + val taskContext = Option(TaskContext.get()) + if (enableVectorizedReader) { + val vectorizedReader = new VectorizedParquetRecordReader( + convertTz.orNull, + datetimeRebaseSpec.mode.toString, + datetimeRebaseSpec.timeZone, + int96RebaseSpec.mode.toString, + int96RebaseSpec.timeZone, + enableOffHeapColumnVector && taskContext.isDefined, + capacity) + // SPARK-37089: We cannot register a task completion listener to close this iterator here + // because downstream exec nodes have already registered their listeners. Since listeners + // are executed in reverse order of registration, a listener registered here would close the + // iterator while downstream exec nodes are still running. When off-heap column vectors are + // enabled, this can cause a use-after-free bug leading to a segfault. + // + // Instead, we use FileScanRDD's task completion listener to close this iterator. + val iter = new RecordReaderIterator(vectorizedReader) + try { + vectorizedReader.initialize(split, hadoopAttemptContext) + vectorizedReader.initBatch(partitionSchema, file.partitionValues) + if (returningBatch) { + vectorizedReader.enableReturningBatches() + } + + // UnsafeRowParquetRecordReader appends the columns internally to avoid another copy. + iter.asInstanceOf[Iterator[InternalRow]] + } catch { + case e: Throwable => + // SPARK-23457: In case there is an exception in initialization, close the iterator to + // avoid leaking resources. + iter.close() + throw e + } + } else { + // ParquetRecordReader returns InternalRow + val readSupport = new ParquetReadSupport( + convertTz, + enableVectorizedReader = false, + datetimeRebaseSpec, + int96RebaseSpec) + val reader = if (pushed.isDefined && enableRecordFilter) { + val parquetFilter = FilterCompat.get(pushed.get, null) + new ParquetRecordReader[InternalRow](readSupport, parquetFilter) + } else { + new ParquetRecordReader[InternalRow](readSupport) + } + val readerWithRowIndexes = ParquetRowIndexUtil.addRowIndexToRecordReaderIfNeeded(reader, + requiredSchema) + val iter = new RecordReaderIterator[InternalRow](readerWithRowIndexes) + try { + readerWithRowIndexes.initialize(split, hadoopAttemptContext) + + val fullSchema = requiredSchema.toAttributes ++ partitionSchema.toAttributes + val unsafeProjection = GenerateUnsafeProjection.generate(fullSchema, fullSchema) + + if (partitionSchema.length == 0) { + // There is no partition columns + iter.map(unsafeProjection) + } else { + val joinedRow = new JoinedRow() + iter.map(d => unsafeProjection(joinedRow(d, file.partitionValues))) + } + } catch { + case e: Throwable => + // SPARK-23457: In case there is an exception in initialization, close the iterator to + // avoid leaking resources. + iter.close() + throw e + } + } + } +} + +object Spark34HoodieParquetReader extends SparkHoodieParquetReaderBuilder { Review Comment: Similar here and all other places ```suggestion object Spark34ParquetReader extends SparkHoodieParquetReaderBuilder { ``` ########## hudi-spark-datasource/hudi-spark3.5.x/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/Spark35HoodieParquetReader.scala: ########## @@ -0,0 +1,284 @@ +/* + * 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.spark.sql.execution.datasources.parquet + +import org.apache.hadoop.conf.Configuration +import org.apache.hadoop.mapred.FileSplit +import org.apache.hadoop.mapreduce.task.TaskAttemptContextImpl +import org.apache.hadoop.mapreduce.{JobID, TaskAttemptID, TaskID, TaskType} +import org.apache.parquet.filter2.compat.FilterCompat +import org.apache.parquet.filter2.predicate.FilterApi +import org.apache.parquet.hadoop.{ParquetInputFormat, ParquetRecordReader} +import org.apache.spark.TaskContext +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions.JoinedRow +import org.apache.spark.sql.catalyst.expressions.codegen.GenerateUnsafeProjection +import org.apache.spark.sql.catalyst.types.DataTypeUtils.toAttributes +import org.apache.spark.sql.catalyst.util.DateTimeUtils +import org.apache.spark.sql.execution.datasources.{DataSourceUtils, FileFormat, PartitionedFile, RecordReaderIterator} +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.sources._ +import org.apache.spark.sql.types.StructType + +class Spark35HoodieParquetReader(enableVectorizedReader: Boolean, Review Comment: ```suggestion class Spark35ParquetReader(enableVectorizedReader: Boolean, ``` -- 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]
