yma11 commented on code in PR #5462: URL: https://github.com/apache/incubator-gluten/pull/5462#discussion_r1575699626
########## backends-velox/src/main/scala/org/apache/gluten/execution/python/ColumnarArrowEvalPythonExec.scala: ########## @@ -0,0 +1,365 @@ +/* + * 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.api.python + +import org.apache.gluten.columnarbatch.ColumnarBatches +import org.apache.gluten.extension.GlutenPlan +import org.apache.gluten.memory.arrowalloc.ArrowBufferAllocators +import org.apache.gluten.utils.Iterators +import org.apache.gluten.vectorized.ArrowWritableColumnVector + +import org.apache.spark.{ContextAwareIterator, SparkEnv, TaskContext} +import org.apache.spark.internal.Logging +import org.apache.spark.rdd.RDD +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.execution.SparkPlan +import org.apache.spark.sql.execution.python.{BasePythonRunnerShim, EvalPythonExec, PythonUDFRunner} +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.types.{DataType, StructField, StructType} +import org.apache.spark.sql.utils.{SparkArrowUtil, SparkSchemaUtil, SparkVectorUtil} +import org.apache.spark.sql.vectorized.{ColumnarBatch, ColumnVector} +import org.apache.spark.util.Utils + +import org.apache.arrow.vector.{VectorLoader, VectorSchemaRoot} +import org.apache.arrow.vector.ipc.{ArrowStreamReader, ArrowStreamWriter} + +import java.io.{DataInputStream, DataOutputStream} +import java.net.Socket +import java.util.concurrent.atomic.AtomicBoolean + +import scala.collection.mutable.ArrayBuffer + +class ColumnarArrowPythonRunner( + funcs: Seq[ChainedPythonFunctions], + evalType: Int, + argOffsets: Array[Array[Int]], + schema: StructType, + timeZoneId: String, + conf: Map[String, String]) + extends BasePythonRunnerShim(funcs, evalType, argOffsets) { + + override val simplifiedTraceback: Boolean = SQLConf.get.pysparkSimplifiedTraceback + + override val bufferSize: Int = SQLConf.get.pandasUDFBufferSize + require( + bufferSize >= 4, + "Pandas execution requires more than 4 bytes. Please set higher buffer. " + + s"Please change '${SQLConf.PANDAS_UDF_BUFFER_SIZE.key}'.") + + protected def newReaderIterator( + stream: DataInputStream, + writerThread: WriterThread, + startTime: Long, + env: SparkEnv, + worker: Socket, + pid: scala.Option[scala.Int], + releasedOrClosed: AtomicBoolean, + context: TaskContext): Iterator[ColumnarBatch] = { + + new ReaderIterator( + stream, + writerThread, + startTime, + env, + worker, + None, + releasedOrClosed, + context) { + private val allocator = ArrowBufferAllocators.contextInstance() + + private var reader: ArrowStreamReader = _ + private var root: VectorSchemaRoot = _ + private var schema: StructType = _ + private var vectors: Array[ColumnVector] = _ + + context.addTaskCompletionListener[Unit] { + _ => + if (reader != null) { + reader.close(false) + } + if (root != null) { + root.close() + } + } + + private var batchLoaded = true + + override protected def read(): ColumnarBatch = { + if (writerThread.exception.isDefined) { + throw writerThread.exception.get + } + try { + if (reader != null && batchLoaded) { + batchLoaded = reader.loadNextBatch() + if (batchLoaded) { + val batch = new ColumnarBatch(vectors) + batch.setNumRows(root.getRowCount) + batch + } else { + reader.close(false) Review Comment: This is following vanilla spark's behavior as we need, refer [code](https://github.com/apache/spark/blob/branch-3.3/sql/core/src/main/scala/org/apache/spark/sql/execution/python/PythonArrowOutput.scala#L82). ########## backends-velox/src/main/scala/org/apache/gluten/execution/python/ColumnarArrowEvalPythonExec.scala: ########## @@ -0,0 +1,365 @@ +/* + * 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.api.python + +import org.apache.gluten.columnarbatch.ColumnarBatches +import org.apache.gluten.extension.GlutenPlan +import org.apache.gluten.memory.arrowalloc.ArrowBufferAllocators +import org.apache.gluten.utils.Iterators +import org.apache.gluten.vectorized.ArrowWritableColumnVector + +import org.apache.spark.{ContextAwareIterator, SparkEnv, TaskContext} +import org.apache.spark.internal.Logging +import org.apache.spark.rdd.RDD +import org.apache.spark.sql.catalyst.InternalRow +import org.apache.spark.sql.catalyst.expressions._ +import org.apache.spark.sql.execution.SparkPlan +import org.apache.spark.sql.execution.python.{BasePythonRunnerShim, EvalPythonExec, PythonUDFRunner} +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.types.{DataType, StructField, StructType} +import org.apache.spark.sql.utils.{SparkArrowUtil, SparkSchemaUtil, SparkVectorUtil} +import org.apache.spark.sql.vectorized.{ColumnarBatch, ColumnVector} +import org.apache.spark.util.Utils + +import org.apache.arrow.vector.{VectorLoader, VectorSchemaRoot} +import org.apache.arrow.vector.ipc.{ArrowStreamReader, ArrowStreamWriter} + +import java.io.{DataInputStream, DataOutputStream} +import java.net.Socket +import java.util.concurrent.atomic.AtomicBoolean + +import scala.collection.mutable.ArrayBuffer + +class ColumnarArrowPythonRunner( + funcs: Seq[ChainedPythonFunctions], + evalType: Int, + argOffsets: Array[Array[Int]], + schema: StructType, + timeZoneId: String, + conf: Map[String, String]) + extends BasePythonRunnerShim(funcs, evalType, argOffsets) { + + override val simplifiedTraceback: Boolean = SQLConf.get.pysparkSimplifiedTraceback + + override val bufferSize: Int = SQLConf.get.pandasUDFBufferSize + require( + bufferSize >= 4, + "Pandas execution requires more than 4 bytes. Please set higher buffer. " + + s"Please change '${SQLConf.PANDAS_UDF_BUFFER_SIZE.key}'.") + + protected def newReaderIterator( + stream: DataInputStream, + writerThread: WriterThread, + startTime: Long, + env: SparkEnv, + worker: Socket, + pid: scala.Option[scala.Int], + releasedOrClosed: AtomicBoolean, + context: TaskContext): Iterator[ColumnarBatch] = { + + new ReaderIterator( + stream, + writerThread, + startTime, + env, + worker, + None, + releasedOrClosed, + context) { + private val allocator = ArrowBufferAllocators.contextInstance() + + private var reader: ArrowStreamReader = _ + private var root: VectorSchemaRoot = _ + private var schema: StructType = _ + private var vectors: Array[ColumnVector] = _ + + context.addTaskCompletionListener[Unit] { + _ => + if (reader != null) { + reader.close(false) + } + if (root != null) { + root.close() + } + } + + private var batchLoaded = true + + override protected def read(): ColumnarBatch = { + if (writerThread.exception.isDefined) { + throw writerThread.exception.get + } + try { + if (reader != null && batchLoaded) { + batchLoaded = reader.loadNextBatch() + if (batchLoaded) { + val batch = new ColumnarBatch(vectors) + batch.setNumRows(root.getRowCount) + batch + } else { + reader.close(false) + // Reach end of stream. Call `read()` again to read control data. + read() + } + } else { + stream.readInt() match { + case SpecialLengths.START_ARROW_STREAM => + reader = new ArrowStreamReader(stream, allocator) + root = reader.getVectorSchemaRoot() + schema = SparkArrowUtil.fromArrowSchema(root.getSchema()) + vectors = ArrowWritableColumnVector + .loadColumns(root.getRowCount, root.getFieldVectors) + .toArray[ColumnVector] + read() + case SpecialLengths.TIMING_DATA => + handleTimingData() + read() + case SpecialLengths.PYTHON_EXCEPTION_THROWN => + throw handlePythonException() + case SpecialLengths.END_OF_DATA_SECTION => + handleEndOfDataSection() + null + } + } + } catch handleException + } + } + } + + override protected def newWriterThread( + env: SparkEnv, + worker: Socket, + inputIterator: Iterator[ColumnarBatch], + partitionIndex: Int, + context: TaskContext): WriterThread = { + new WriterThread(env, worker, inputIterator, partitionIndex, context) { + override protected def writeCommand(dataOut: DataOutputStream): Unit = { + // Write config for the worker as a number of key -> value pairs of strings + dataOut.writeInt(conf.size) + for ((k, v) <- conf) { + PythonRDD.writeUTF(k, dataOut) + PythonRDD.writeUTF(v, dataOut) + } + PythonUDFRunner.writeUDFs(dataOut, funcs, argOffsets) + } + + override protected def writeIteratorToStream(dataOut: DataOutputStream): Unit = { + var numRows: Long = 0 + val arrowSchema = SparkSchemaUtil.toArrowSchema(schema, timeZoneId) + val allocator = ArrowBufferAllocators.contextInstance() + val root = VectorSchemaRoot.create(arrowSchema, allocator) + + Utils.tryWithSafeFinally { + val loader = new VectorLoader(root) + val writer = new ArrowStreamWriter(root, null, dataOut) + writer.start() + while (inputIterator.hasNext) { + val nextBatch = inputIterator.next() + numRows += nextBatch.numRows + + val cols = (0 until nextBatch.numCols).toList.map( + i => + nextBatch + .asInstanceOf[ColumnarBatch] + .column(i) + .asInstanceOf[ArrowWritableColumnVector] + .getValueVector) + val nextRecordBatch = + SparkVectorUtil.toArrowRecordBatch(nextBatch.numRows, cols) + loader.load(nextRecordBatch) + writer.writeBatch() + if (nextRecordBatch != null) { + nextRecordBatch.close() + } + } + // end writes footer to the output stream and doesn't clean any resources. + // It could throw exception if the output stream is closed, so it should be + // in the try block. + writer.end() + } { + root.close() + // allocator can't close now or the data will loss + // allocator.close() + } + } + } + } +} + +class CloseableColumnBatchIterator(itr: Iterator[ColumnarBatch]) + extends Iterator[ColumnarBatch] + with Logging { + var cb: ColumnarBatch = null + + private def closeCurrentBatch(): Unit = { + if (cb != null) { + cb.close + cb = null + } + } + + override def hasNext: Boolean = { + itr.hasNext + } + + override def next(): ColumnarBatch = { + closeCurrentBatch() + cb = itr.next() + cb + } +} + +case class ColumnarArrowEvalPythonExec( + udfs: Seq[PythonUDF], + resultAttrs: Seq[Attribute], + child: SparkPlan, + evalType: Int) + extends EvalPythonExec + with GlutenPlan { + override def supportsColumnar: Boolean = true + // TODO: add additional projection support by pre-project + // FIXME: incorrect metrics updater + + override protected def evaluate( + funcs: Seq[ChainedPythonFunctions], + argOffsets: Array[Array[Int]], + iter: Iterator[InternalRow], + schema: StructType, + context: TaskContext): Iterator[InternalRow] = { + throw new IllegalStateException( + "ColumnarArrowEvalPythonExec doesn't support evaluate InternalRow.") + } + + private val sessionLocalTimeZone = conf.sessionLocalTimeZone + private def getPythonRunnerConfMap(conf: SQLConf): Map[String, String] = { + val timeZoneConf = Seq(SQLConf.SESSION_LOCAL_TIMEZONE.key -> conf.sessionLocalTimeZone) + val pandasColsByName = Seq( + SQLConf.PANDAS_GROUPED_MAP_ASSIGN_COLUMNS_BY_NAME.key -> + conf.pandasGroupedMapAssignColumnsByName.toString) + val arrowSafeTypeCheck = Seq( + SQLConf.PANDAS_ARROW_SAFE_TYPE_CONVERSION.key -> + conf.arrowSafeTypeConversion.toString) + Map(timeZoneConf ++ pandasColsByName ++ arrowSafeTypeCheck: _*) + } + private val pythonRunnerConf = getPythonRunnerConfMap(conf) + + protected def evaluateColumnar( + funcs: Seq[ChainedPythonFunctions], + argOffsets: Array[Array[Int]], + iter: Iterator[ColumnarBatch], + schema: StructType, + context: TaskContext): Iterator[ColumnarBatch] = { + + val outputTypes = output.drop(child.output.length).map(_.dataType) + + val columnarBatchIter = new ColumnarArrowPythonRunner( + funcs, + evalType, + argOffsets, + schema, + sessionLocalTimeZone, + pythonRunnerConf).compute(iter, context.partitionId(), context) + + columnarBatchIter.map { + batch => + val actualDataTypes = (0 until batch.numCols()).map(i => batch.column(i).dataType()) + assert( + outputTypes == actualDataTypes, + "Invalid schema from arrow_udf: " + + s"expected ${outputTypes.mkString(", ")}, got ${actualDataTypes.mkString(", ")}") + batch + } + } + + private def collectFunctions(udf: PythonUDF): (ChainedPythonFunctions, Seq[Expression]) = { + udf.children match { + case Seq(u: PythonUDF) => + val (chained, children) = collectFunctions(u) + (ChainedPythonFunctions(chained.funcs ++ Seq(udf.func)), children) + case children => + // There should not be any other UDFs, or the children can't be evaluated directly. + assert(children.forall(_.find(_.isInstanceOf[PythonUDF]).isEmpty)) + (ChainedPythonFunctions(Seq(udf.func)), udf.children) + } + } + + override protected def doExecuteColumnar(): RDD[ColumnarBatch] = { + val inputRDD = child.executeColumnar() + inputRDD.mapPartitions { + iter => + val context = TaskContext.get() + val (pyFuncs, inputs) = udfs.map(collectFunctions).unzip + // flatten all the arguments + val allInputs = new ArrayBuffer[Expression] + val dataTypes = new ArrayBuffer[DataType] + val argOffsets = inputs.map { + input => + input.map { + e => + if (allInputs.exists(_.semanticEquals(e))) { + allInputs.indexWhere(_.semanticEquals(e)) + } else { + allInputs += e + dataTypes += e.dataType + allInputs.length - 1 + } + }.toArray + }.toArray + val schema = StructType(dataTypes.zipWithIndex.map { + case (dt, i) => + StructField(s"_$i", dt) + }.toSeq) + val contextAwareIterator = new ContextAwareIterator(context, iter) + val input_cb_cache = new ArrayBuffer[ColumnarBatch]() + val inputBatchIter = contextAwareIterator.map { + input_cb => + ColumnarBatches.ensureLoaded(ArrowBufferAllocators.contextInstance, input_cb) + // 0. cache input for later merge + ColumnarBatches.retain(input_cb) + input_cb_cache += input_cb + input_cb + } + + val outputColumnarBatchIterator = + evaluateColumnar(pyFuncs, argOffsets, inputBatchIter, schema, context) + val res = new CloseableColumnBatchIterator( + outputColumnarBatchIterator.zipWithIndex.map { + case (output_cb, batchId) => Review Comment: updated. -- 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]
