Github user liancheng commented on a diff in the pull request:
https://github.com/apache/spark/pull/4563#discussion_r24572328
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/parquet/newParquet.scala ---
@@ -413,18 +441,21 @@ case class ParquetRelation2
// When the data does not include the key and the key is requested
then we must fill it in
// based on information from the input split.
- if (!dataSchemaIncludesPartitionKeys &&
partitionKeyLocations.nonEmpty) {
+ if (!partitionKeysIncludedInDataSchema &&
partitionKeyLocations.nonEmpty) {
baseRDD.mapPartitionsWithInputSplit { case (split:
ParquetInputSplit, iterator) =>
val partValues = selectedPartitions.collectFirst {
case p if split.getPath.getParent.toString == p.path => p.values
}.get
+ val requiredPartOrdinal = partitionKeyLocations.keys.toSeq
+
iterator.map { pair =>
val row = pair._2.asInstanceOf[SpecificMutableRow]
var i = 0
- while (i < partValues.size) {
+ while (i < requiredPartOrdinal.size) {
// TODO Avoids boxing cost here!
- row.update(partitionKeyLocations(i), partValues(i))
+ val partOrdinal = requiredPartOrdinal(i)
+ row.update(partitionKeyLocations(partOrdinal),
partValues(partOrdinal))
--- End diff --
This is a bug fix, shouldn't always append *all* partition keys to the
result row.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
