Github user budde commented on a diff in the pull request:
https://github.com/apache/spark/pull/16944#discussion_r102850475
--- Diff:
sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveMetastoreCatalog.scala ---
@@ -161,22 +164,45 @@ private[hive] class
HiveMetastoreCatalog(sparkSession: SparkSession) extends Log
bucketSpec,
Some(partitionSchema))
+ val catalogTable = metastoreRelation.catalogTable
val logicalRelation = cached.getOrElse {
val sizeInBytes =
metastoreRelation.stats(sparkSession.sessionState.conf).sizeInBytes.toLong
val fileIndex = {
- val index = new CatalogFileIndex(
- sparkSession, metastoreRelation.catalogTable, sizeInBytes)
+ val index = new CatalogFileIndex(sparkSession, catalogTable,
sizeInBytes)
if (lazyPruningEnabled) {
index
} else {
index.filterPartitions(Nil) // materialize all the
partitions in memory
}
}
val partitionSchemaColumnNames =
partitionSchema.map(_.name.toLowerCase).toSet
- val dataSchema =
- StructType(metastoreSchema
- .filterNot(field =>
partitionSchemaColumnNames.contains(field.name.toLowerCase)))
+ val filteredMetastoreSchema = StructType(metastoreSchema
+ .filterNot(field =>
partitionSchemaColumnNames.contains(field.name.toLowerCase)))
+
+ // Infer a case-sensitive schema when the metastore doesn't
return one, if configured.
+ val inferredSchema = inferSchema(
+ catalogTable,
+ metastoreSchema,
+ options,
+ defaultSource,
+ fileType,
+ fileIndex)
+
+ // If configured, save the inferred case-sensitive schema to the
table properties and
+ // fetch the updated CatalogTable record for use in the
LogicalRelation.
+ val updatedCatalogTable = updateCatalogTable(catalogTable,
inferredSchema)
+
+ val dataSchema = inferenceMode match {
+ case (INFER_AND_SAVE | INFER_ONLY) if
(!catalogTable.schemaPreservesCase) =>
+ inferredSchema.getOrElse {
+ logWarning(s"Unable to infer schema for table
$tableIdentifier from file format " +
+ s"$defaultSource (inference mode: $inferenceMode); using
metastore schema.")
+ filteredMetastoreSchema
+ }
+ case _ =>
+ filteredMetastoreSchema
+ }
--- End diff --
I'll see if I can clean this up some more. It gets a bit messy since we'll
want to replace ```catalogTable``` in the ```LogicalRelation``` returned by
this method with the updated catalog table returned by
```updateTableCatalog()``` if the schema is successfully written to the
metastore. Essentially, we need to return an ```Option[StructType]```
representing if the schema was inferred or not and an
```Option[CatalogTable]``` representing if the catalog table was updated.
---
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]
