jonvex commented on code in PR #12201:
URL: https://github.com/apache/hudi/pull/12201#discussion_r1829582501
##########
hudi-spark-datasource/hudi-spark/src/main/scala/org/apache/spark/sql/hudi/command/payload/ExpressionPayload.scala:
##########
@@ -164,6 +163,34 @@ class ExpressionPayload(@transient record: GenericRecord,
}
}
+ private def doRecordMerge(incomingRecord: GenericRecord,
+ existingRecord: IndexedRecord,
+ schema: Schema,
+ properties: Properties): HOption[IndexedRecord] = {
+ val originalPayload = properties.getProperty(PAYLOAD_ORIGINAL_AVRO_PAYLOAD)
+ if
(originalPayload.equals(classOf[OverwriteWithLatestAvroPayload].getName)) {
+ HOption.of(incomingRecord)
+ } else if
(originalPayload.equals(classOf[DefaultHoodieRecordPayload].getName)) {
+ if (needUpdatingPersistedRecord(existingRecord, incomingRecord,
properties)) {
+ HOption.of(incomingRecord)
+ } else {
+ HOption.of(existingRecord)
+ }
+ } else {
+ val orderingField = ConfigUtils.getOrderingField(properties)
+ if (StringUtils.isNullOrEmpty(orderingField)) {
+ HOption.of(incomingRecord)
+ } else {
+ val consistentLogicalTimestampEnabled =
properties.getProperty(KeyGeneratorOptions.KEYGENERATOR_CONSISTENT_LOGICAL_TIMESTAMP_ENABLED.key,
+
KeyGeneratorOptions.KEYGENERATOR_CONSISTENT_LOGICAL_TIMESTAMP_ENABLED.defaultValue).toBoolean
+ val incomingRecordPayload =
HoodieRecordUtils.loadPayload(originalPayload, incomingRecord,
+ HoodieAvroUtils.getNestedFieldVal(incomingRecord, orderingField,
true, consistentLogicalTimestampEnabled)
+ .asInstanceOf[Comparable[_]]).asInstanceOf[HoodieRecordPayload[_
<: HoodieRecordPayload[_]]]
+ incomingRecordPayload.combineAndGetUpdateValue(existingRecord, schema,
properties)
Review Comment:
The payload takes in the record as a param so I'm not sure we can do any
caching. It is also still functionally correct if we get rid of the special
handling for overwrite and default payloads and load the payload for
everything. This could be a case of over optimization but I would think getting
rid of reflection that happens for every record would be non negligible. So
that's why I did it
--
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]
