cloud-fan commented on code in PR #38939:
URL: https://github.com/apache/spark/pull/38939#discussion_r1042285999
##########
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/FileFormatWriter.scala:
##########
@@ -123,42 +116,106 @@ object FileFormatWriter extends Logging {
val outputWriterFactory =
fileFormat.prepareWrite(sparkSession, job, caseInsensitiveOptions,
dataSchema)
- val description = new WriteJobDescription(
+ new WriteJobDescription(
uuid = UUID.randomUUID.toString,
serializableHadoopConf = new
SerializableConfiguration(job.getConfiguration),
outputWriterFactory = outputWriterFactory,
- allColumns = finalOutputSpec.outputColumns,
+ allColumns = outputSpec.outputColumns,
dataColumns = dataColumns,
partitionColumns = partitionColumns,
bucketSpec = writerBucketSpec,
- path = finalOutputSpec.outputPath,
- customPartitionLocations = finalOutputSpec.customPartitionLocations,
+ path = outputSpec.outputPath,
+ customPartitionLocations = outputSpec.customPartitionLocations,
maxRecordsPerFile =
caseInsensitiveOptions.get("maxRecordsPerFile").map(_.toLong)
.getOrElse(sparkSession.sessionState.conf.maxRecordsPerFile),
timeZoneId = caseInsensitiveOptions.get(DateTimeUtils.TIMEZONE_OPTION)
.getOrElse(sparkSession.sessionState.conf.sessionLocalTimeZone),
statsTrackers = statsTrackers
)
+ }
+
+ private def createSortPlan(
+ plan: SparkPlan,
+ requiredOrdering: Seq[Expression],
+ outputSpec: OutputSpec): SortExec = {
+ // SPARK-21165: the `requiredOrdering` is based on the attributes from
analyzed plan, and
+ // the physical plan may have different attribute ids due to optimizer
removing some
+ // aliases. Here we bind the expression ahead to avoid potential attribute
ids mismatch.
+ val orderingExpr = bindReferences(
+ requiredOrdering.map(SortOrder(_, Ascending)), outputSpec.outputColumns)
+ SortExec(
+ orderingExpr,
+ global = false,
+ child = plan)
+ }
+
+ private def createConcurrentOutputWriterSpec(
+ sparkSession: SparkSession,
+ sortPlan: SortExec,
+ sortColumns: Seq[Attribute]): Option[ConcurrentOutputWriterSpec] = {
+ val maxWriters =
sparkSession.sessionState.conf.maxConcurrentOutputFileWriters
+ val concurrentWritersEnabled = maxWriters > 0 && sortColumns.isEmpty
+ if (concurrentWritersEnabled) {
+ Some(ConcurrentOutputWriterSpec(maxWriters, () =>
sortPlan.createSorter()))
+ } else {
+ None
+ }
+ }
+
+ /**
+ * Write files using [[SparkPlan.executeWrite]]
+ */
+ private def executeWrite(
+ session: SparkSession,
+ planForWrites: SparkPlan,
+ writeFilesSpec: WriteFilesSpec,
+ job: Job): Set[String] = {
+ val committer = writeFilesSpec.committer
+ val description = writeFilesSpec.description
+
+ writeAndCommit(session, job, description, committer) {
+ val rdd = planForWrites.executeWrite(writeFilesSpec)
+ val ret = new Array[WriteTaskResult](rdd.partitions.length)
+ session.sparkContext.runJob(
+ rdd,
+ (context: TaskContext, iter: Iterator[WriterCommitMessage]) => {
+ assert(iter.hasNext)
Review Comment:
we should make sure this iterator only have one element
--
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]
