Github user JoshRosen commented on a diff in the pull request:
https://github.com/apache/spark/pull/8026#discussion_r42583944
--- Diff:
sql/core/src/main/scala/org/apache/spark/sql/sources/interfaces.scala ---
@@ -544,11 +544,35 @@ abstract class HadoopFsRelation
private[sql](maybePartitionSpec: Option[Partitio
}
private def discoverPartitions(): PartitionSpec = {
- val typeInference =
sqlContext.conf.partitionColumnTypeInferenceEnabled()
// We use leaf dirs containing data files to discover the schema.
val leafDirs = fileStatusCache.leafDirToChildrenFiles.keys.toSeq
- PartitioningUtils.parsePartitions(leafDirs,
PartitioningUtils.DEFAULT_PARTITION_NAME,
- typeInference)
+ userDefinedPartitionColumns match {
+ case Some(schema) =>
+ val spec = PartitioningUtils.parsePartitions(
+ leafDirs, PartitioningUtils.DEFAULT_PARTITION_NAME, false)
+
+ // Without auto inference, all of value in the `row` should be
null or in StringType,
+ // we need to cast into the data type that user specified.
+ def castPartitionValueWithGivenSchema(row: InternalRow, schema:
StructType)
+ : InternalRow = {
+ InternalRow((0 until row.numFields) map { i =>
+ Cast(Literal.create(row.getString(i), StringType),
schema.fields(i).dataType).eval()
+ }: _*)
+ }
+
+ assert(schema.length == spec.partitionColumns.length &&
+ schema.fieldNames.sameElements(spec.partitionColumns.fieldNames),
+ s"Auto infer partition column is not match with user specified,
" +
--- End diff --
The wording of this error message might be slightly confusing to users
since this branch is explicitly _disabling_ inference. I think that it might be
slightly clearer to say something like "Actual partitioning column names did
not match user-specified partitioning schema; expected ... but got ...", since
as far as I know the inference is really only done for the types of the
columns, not their names.
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
