viirya commented on a change in pull request #29457:
URL: https://github.com/apache/spark/pull/29457#discussion_r474160197
##########
File path:
sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/orc/OrcFiltersBase.scala
##########
@@ -49,19 +51,21 @@ trait OrcFiltersBase {
*/
protected[sql] def getSearchableTypeMap(
schema: StructType,
- caseSensitive: Boolean): Map[String, DataType] = {
+ caseSensitive: Boolean): Map[String, OrcPrimitiveField] = {
import
org.apache.spark.sql.connector.catalog.CatalogV2Implicits.MultipartIdentifierHelper
def getPrimitiveFields(
fields: Seq[StructField],
- parentFieldNames: Seq[String] = Seq.empty): Seq[(String, DataType)] = {
+ parentFieldNames: Seq[String] = Seq.empty): Seq[(String,
OrcPrimitiveField)] = {
fields.flatMap { f =>
f.dataType match {
case st: StructType =>
getPrimitiveFields(st.fields, parentFieldNames :+ f.name)
case BinaryType => None
case _: AtomicType =>
- Some(((parentFieldNames :+ f.name).quoted, f.dataType))
+ val fieldName = (parentFieldNames :+ f.name).quoted
+ val orcField = OrcPrimitiveField(fieldName, f.dataType)
+ Some((fieldName, orcField))
Review comment:
Because we look up attribute name of pushed down predicates in the map
`dataTypeMap(name)`, we need to easily retrieve the file schema's field name.
If we simply use `Map[String, DataType]`, the matched attribute name under
case-insensitive analysis could be possibly in different letter case to the
file schema's field name.
E,g, If ORC file has 'ABC` field, pushed down predicate is "abc > 0",
`dataTypeMap.contain("abc")` is true and we can get `dataTypeMap("abc")` back.
But we need file schema's field name "ABC".
It is similar to Parquet:
https://github.com/apache/spark/blob/d6a68e0b67ff7de58073c176dd097070e88ac831/sql/core/src/main/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetFilters.scala#L56-L76
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
