Skip to content

fix bug: requested file schema no change in stream task #226

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
xuchen-plus merged 1 commit into from
May 22, 2023
Merged

fix bug: requested file schema no change in stream task #226

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
90 changes: 57 additions & 33 deletions ...src/main/scala/org/apache/spark/sql/execution/datasources/v2/merge/MergeParquetScan.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -128,21 +128,8 @@ abstract class MergeDeltaParquetScan(sparkSession: SparkSession,
val readDataSchemaAsJson = readDataSchema.json

val requestedFields = readDataSchema.fieldNames
val requestFilesSchema =
fileInfo
.groupBy(_.range_version)
.map(m => {
val fileExistCols = m._2.head.file_exist_cols.split(",")
m._1 + "->" + StructType(
requestedFields.filter(f => fileExistCols.contains(f) || tableInfo.hash_partition_columns.contains(f))
.map(c => tableInfo.schema(c))
).json
}).mkString("|")

hadoopConf.set(ParquetInputFormat.READ_SUPPORT_CLASS, classOf[ParquetReadSupport].getName)
hadoopConf.set(
ParquetReadSupport.SPARK_ROW_REQUESTED_SCHEMA,
requestFilesSchema)
hadoopConf.set(
ParquetWriteSupport.SPARK_ROW_SCHEMA,
readDataSchemaAsJson)
Expand Down Expand Up @@ -247,15 +234,52 @@ abstract class MergeDeltaParquetScan(sparkSession: SparkSession,

// produce requested schema
val requestedFields = readDataSchema.fieldNames
val requestFilesSchemaMap = fileInfo
.groupBy(_.range_version)
.map(m => {
val fileExistCols = m._2.head.file_exist_cols.split(",")
(m._1, StructType(
requestedFields.filter(f => fileExistCols.contains(f) || tableInfo.hash_partition_columns.contains(f))
.map(c => tableInfo.schema(c))
))
})

val requestFilesSchemaMap = if (isStreaming) {
val requestFilesSchema = newFileIndex.getFileInfoForPartitionVersion()
.groupBy(_.range_version)
.map(m => {
val fileExistCols = m._2.head.file_exist_cols.split(",")
m._1 + "->" + StructType(
requestedFields.filter(f => fileExistCols.contains(f) || tableInfo.hash_partition_columns.contains(f))
.map(c => tableInfo.schema(c))
).json
}).mkString("|")
hadoopConf.set(ParquetReadSupport.SPARK_ROW_REQUESTED_SCHEMA, requestFilesSchema)

newFileIndex.getFileInfoForPartitionVersion()
.groupBy(_.range_version)
.map(m => {
val fileExistCols = m._2.head.file_exist_cols.split(",")
(m._1, StructType(
requestedFields.filter(f => fileExistCols.contains(f) || tableInfo.hash_partition_columns.contains(f))
.map(c => tableInfo.schema(c))
))
})
} else {

val requestFilesSchema =
fileInfo
.groupBy(_.range_version)
.map(m => {
val fileExistCols = m._2.head.file_exist_cols.split(",")
m._1 + "->" + StructType(
requestedFields.filter(f => fileExistCols.contains(f) || tableInfo.hash_partition_columns.contains(f))
.map(c => tableInfo.schema(c))
).json
}).mkString("|")

hadoopConf.set(ParquetReadSupport.SPARK_ROW_REQUESTED_SCHEMA, requestFilesSchema)
fileInfo
.groupBy(_.range_version)
.map(m => {
val fileExistCols = m._2.head.file_exist_cols.split(",")
(m._1, StructType(
requestedFields.filter(f => fileExistCols.contains(f) || tableInfo.hash_partition_columns.contains(f))
.map(c => tableInfo.schema(c))
))
})
}

partition.files.flatMap { file =>
val filePath = file.getPath
Expand All @@ -270,13 +294,12 @@ abstract class MergeDeltaParquetScan(sparkSession: SparkSession,
requestFilesSchemaMap,
readDataSchema,
readPartitionSchema.fieldNames)
}.toSeq
//.sortBy(_.length)(implicitly[Ordering[Long]].reverse)
}
}

if (splitFiles.length == 1) {
val path = new Path(splitFiles(0).filePath)
if (!isSplittable(path) && splitFiles(0).length >
val path = new Path(splitFiles.head.filePath)
if (!isSplittable(path) && splitFiles.head.length >
sparkSession.sparkContext.getConf.get(IO_WARNING_LARGEFILETHRESHOLD)) {
logWarning(s"Loading one large unsplittable file ${path.toString} with only one " +
s"partition, the reason is: ${getFileUnSplittableReason(path)}")
Expand Down Expand Up @@ -399,10 +422,10 @@ case class OnePartitionMergeBucketScan(sparkSession: SparkSession,

Seq.tabulate(bucketNum) { bucketId =>
var files = fileWithBucketId.getOrElse(bucketId, Array.empty)
val isSingleFile = files.size == 1
val isSingleFile = files.length == 1

if (!isSingleFile) {
val versionFiles = for (version <- 0 to files.size - 1) yield files(version).copy(writeVersion = version + 1)
val versionFiles = for (version <- files.indices) yield files(version).copy(writeVersion = version + 1)
files = versionFiles.toArray
}
MergeFilePartition(bucketId, Array(files), isSingleFile)
Expand Down Expand Up @@ -456,10 +479,11 @@ case class MultiPartitionMergeBucketScan(sparkSession: SparkSession,

var allPartitionIsSingleFile = true
var isSingleFile = false
for (index <- 0 to files.size - 1) {
isSingleFile = files(index).size == 1

for (index <- files.indices) {
isSingleFile = files(index).length == 1
if (!isSingleFile) {
val versionFiles = for (elem <- 0 to files(index).size - 1) yield files(index)(elem).copy(writeVersion = elem)
val versionFiles = for (elem <- files(index).indices) yield files(index)(elem).copy(writeVersion = elem)
files(index) = versionFiles.toArray
allPartitionIsSingleFile = false
}
Expand Down Expand Up @@ -514,9 +538,9 @@ case class MultiPartitionMergeScan(sparkSession: SparkSession,
groupByPartition.foreach(p => {
p._2.groupBy(_.fileBucketId).foreach(g => {
var files = g._2.toArray
val isSingleFile = files.size == 1
val isSingleFile = files.length == 1
if (!isSingleFile) {
val versionFiles = for (version <- 0 to files.size - 1) yield files(version).copy(writeVersion = version)
val versionFiles = for (version <- files.indices) yield files(version).copy(writeVersion = version)
files = versionFiles.toArray
}
partitions += MergeFilePartition(i, Array(files), isSingleFile)
Expand Down