Github user HyukjinKwon commented on a diff in the pull request:
https://github.com/apache/spark/pull/21477#discussion_r195622516
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/execution/python/PythonForeachWriter.scala
---
@@ -0,0 +1,161 @@
+/*
+ * 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.python
+
+import java.io.File
+import java.util.concurrent.TimeUnit
+import java.util.concurrent.locks.ReentrantLock
+
+import org.apache.spark.{SparkEnv, TaskContext}
+import org.apache.spark.api.python._
+import org.apache.spark.internal.Logging
+import org.apache.spark.memory.TaskMemoryManager
+import org.apache.spark.sql.ForeachWriter
+import org.apache.spark.sql.catalyst.expressions.UnsafeRow
+import org.apache.spark.sql.types.StructType
+import org.apache.spark.util.{NextIterator, Utils}
+
+class PythonForeachWriter(func: PythonFunction, schema: StructType)
+ extends ForeachWriter[UnsafeRow] {
+
+ private lazy val context = TaskContext.get()
+ private lazy val buffer = new PythonForeachWriter.UnsafeRowBuffer(
+ context.taskMemoryManager, new
File(Utils.getLocalDir(SparkEnv.get.conf)), schema.fields.length)
+ private lazy val inputRowIterator = buffer.iterator
+
+ private lazy val inputByteIterator = {
+ EvaluatePython.registerPicklers()
+ val objIterator = inputRowIterator.map { row =>
EvaluatePython.toJava(row, schema) }
+ new SerDeUtil.AutoBatchedPickler(objIterator)
+ }
+
+ private lazy val pythonRunner = {
+ val conf = SparkEnv.get.conf
+ val bufferSize = conf.getInt("spark.buffer.size", 65536)
+ val reuseWorker = conf.getBoolean("spark.python.worker.reuse", true)
+ PythonRunner(func, bufferSize, reuseWorker)
+ }
+
+ private lazy val outputIterator =
+ pythonRunner.compute(inputByteIterator, context.partitionId(), context)
+
+ override def open(partitionId: Long, version: Long): Boolean = {
+ outputIterator // initialize everything
+ TaskContext.get.addTaskCompletionListener { _ => buffer.close() }
+ true
+ }
+
+ override def process(value: UnsafeRow): Unit = {
+ buffer.add(value)
+ }
+
+ override def close(errorOrNull: Throwable): Unit = {
+ buffer.allRowsAdded()
+ if (outputIterator.hasNext) outputIterator.next() // to throw python
exception if there was one
+ }
+}
+
+object PythonForeachWriter {
+
+ /**
+ * A buffer that is designed for the sole purpose of buffering
UnsafeRows in PythonForeachWriter.
+ * It is designed to be used with only 1 writer thread (i.e. JVM task
thread) and only 1 reader
+ * thread (i.e. PythonRunner writing thread that reads from the buffer
and writes to the Python
+ * worker stdin). Adds to the buffer are non-blocking, and reads through
the buffer's iterator
+ * are blocking, that is, it blocks until new data is available or all
data has been added.
+ *
+ * Internally, it uses a [[HybridRowQueue]] to buffer the rows in a
practically unlimited queue
+ * across memory and local disk. However, HybridRowQueue is designed to
be used only with
+ * EvalPythonExec where the reader is always behind the the writer, that
is, the reader does not
+ * try to read n+1 rows if the writer has only written n rows at any
point of time. This
+ * assumption is not true for PythonForeachWriter where rows may be
added at a different rate as
+ * they are consumed by the python worker. Hence, to maintain the
invariant of the reader being
+ * behind the writer while using HybridRowQueue, the buffer does the
following
+ * - Keeps a count of the rows in the HybridRowQueue
+ * - Blocks the buffer's consuming iterator when the count is 0 so that
the reader does not
+ * try to read more rows than what has been written.
+ *
+ * The implementation of the blocking iterator (ReentrantLock,
Condition, etc.) has been borrowed
+ * from that of ArrayBlockingQueue.
+ */
+ class UnsafeRowBuffer(taskMemoryManager: TaskMemoryManager, tempDir:
File, numFields: Int)
+ extends Logging {
+ private val queue = HybridRowQueue(taskMemoryManager, tempDir,
numFields)
+ private val lock = new ReentrantLock()
+ private val unblockRemove = lock.newCondition()
+
+ // All of these are guarded by `lock`
+ private var count = 0L
+ private var allAdded = false
+ private var exception: Throwable = null
+
+ val iterator = new NextIterator[UnsafeRow] {
+ override protected def getNext(): UnsafeRow = {
+ val row = remove()
+ if (row == null) finished = true
+ row
+ }
+ override protected def close(): Unit = { }
+ }
+
+ def add(row: UnsafeRow): Unit = withLock {
+ assert(queue.add(row), s"Failed to add row to HybridRowQueue while
sending data to Python" +
+ s"[count = $count, allAdded = $allAdded, exception = $exception]")
+ count += 1
+ unblockRemove.signal()
+ logTrace(s"Added $row, $count left")
+ }
+
+ private def remove(): UnsafeRow = withLock {
+ while (count == 0 && !allAdded && exception == null) {
+ unblockRemove.await(100, TimeUnit.MILLISECONDS)
+ }
+
+ // If there was any error in the adding thread, then rethrow it in
the removing thread
+ if (exception != null) throw exception
+
+ if (count > 0) {
+ val row = queue.remove()
+ assert(row != null, s"HybridRowQueue.remove() returned null " +
--- End diff --
looks `s` can be removed
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
