danny0405 commented on code in PR #5825:
URL: https://github.com/apache/hudi/pull/5825#discussion_r896359970
##########
hudi-spark-datasource/hudi-spark/src/main/scala/org/apache/spark/sql/hudi/command/payload/ExpressionPayload.scala:
##########
@@ -318,5 +316,26 @@ object ExpressionPayload {
}
})
}
+
+ private val mergeCache =
CacheBuilder.newBuilder().maximumSize(16).build[TupleSchema, Schema]()
+
Review Comment:
`mergeCache` -> `mergedSchemaCache`
##########
hudi-spark-datasource/hudi-spark/src/main/scala/org/apache/spark/sql/hudi/command/payload/ExpressionPayload.scala:
##########
@@ -318,5 +316,26 @@ object ExpressionPayload {
}
})
}
+
+ private val mergeCache =
CacheBuilder.newBuilder().maximumSize(16).build[TupleSchema, Schema]()
+
+ def getMergeSchema(source: Schema, target: Schema): Schema = {
+
Review Comment:
`getMergeSchema` -> `getMergedSchema`
##########
hudi-spark-datasource/hudi-spark/src/main/scala/org/apache/spark/sql/hudi/command/payload/ExpressionPayload.scala:
##########
@@ -318,5 +316,26 @@ object ExpressionPayload {
}
})
}
+
+ private val mergeCache =
CacheBuilder.newBuilder().maximumSize(16).build[TupleSchema, Schema]()
+
+ def getMergeSchema(source: Schema, target: Schema): Schema = {
+
+ mergeCache.get(TupleSchema(source, target), new Callable[Schema] {
+ override def call(): Schema = {
+ val noMetaRight = HoodieAvroUtils.removeMetadataFields(target)
+ val mergedFields =
+ source.getFields.asScala.map(field =>
Review Comment:
Can we reuse the `mergeSchema` method ?
##########
hudi-spark-datasource/hudi-spark-common/src/main/scala/org/apache/spark/sql/hudi/command/payload/SqlTypedRecord.scala:
##########
@@ -17,27 +17,59 @@
package org.apache.spark.sql.hudi.command.payload
+import com.google.common.cache.CacheBuilder
import org.apache.avro.Schema
import org.apache.avro.generic.IndexedRecord
-import org.apache.hudi.{AvroConversionUtils, SparkAdapterSupport}
+import org.apache.hudi.HoodieSparkUtils.sparkAdapter
+import org.apache.hudi.AvroConversionUtils
+import org.apache.spark.sql.avro.HoodieAvroDeserializer
import org.apache.spark.sql.catalyst.InternalRow
+import
org.apache.spark.sql.hudi.command.payload.SqlTypedRecord.{getAvroDeserializer,
getSqlType}
+import org.apache.spark.sql.types.StructType
+
+import java.util.concurrent.Callable
/**
* A sql typed record which will convert the avro field to sql typed value.
*/
-class SqlTypedRecord(val record: IndexedRecord) extends IndexedRecord with
SparkAdapterSupport {
+class SqlTypedRecord(val record: IndexedRecord) extends IndexedRecord {
- private lazy val sqlType =
AvroConversionUtils.convertAvroSchemaToStructType(getSchema)
- private lazy val avroDeserializer =
sparkAdapter.createAvroDeserializer(record.getSchema, sqlType)
- private lazy val sqlRow =
avroDeserializer.deserialize(record).get.asInstanceOf[InternalRow]
+ private lazy val sqlRow =
getAvroDeserializer(getSchema).deserialize(record).get.asInstanceOf[InternalRow]
override def put(i: Int, v: Any): Unit = {
record.put(i, v)
}
override def get(i: Int): AnyRef = {
- sqlRow.get(i, sqlType(i).dataType)
+ sqlRow.get(i, getSqlType(getSchema)(i).dataType)
}
override def getSchema: Schema = record.getSchema
}
+
+object SqlTypedRecord {
+
+ private val sqlTypeCache =
CacheBuilder.newBuilder().maximumSize(16).build[Schema, StructType]()
Review Comment:
If 16 is a default size, why we still configure it here ?
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
