xushiyan commented on code in PR #9041:
URL: https://github.com/apache/hudi/pull/9041#discussion_r1246655960
##########
hudi-client/hudi-client-common/src/main/java/org/apache/hudi/index/HoodieIndexUtils.java:
##########
@@ -310,6 +312,56 @@ public static <R> HoodieData<HoodieRecord<R>>
mergeForPartitionUpdates(
return Arrays.asList(deleteRecord, getTaggedRecord(merged,
Option.empty())).iterator();
}
});
- return taggedUpdatingRecords.union(newRecords);
+ return taggedUpdatingRecords.union(taggedNewRecords);
+ }
+
+ public static <R> HoodieData<HoodieRecord<R>> tagGlobalLocationBackToRecords(
+ HoodieData<HoodieRecord<R>> incomingRecords,
+ HoodiePairData<String, HoodieRecordGlobalLocation>
keyAndExistingLocations,
+ boolean mayContainDuplicateLookup,
+ boolean shouldUpdatePartitionPath,
+ HoodieWriteConfig config,
+ HoodieTable table) {
+ final HoodieRecordMerger merger = config.getRecordMerger();
+
+ HoodiePairData<String, HoodieRecord<R>> keyAndIncomingRecords =
+ incomingRecords.mapToPair(record -> Pair.of(record.getRecordKey(),
record));
+
+ // Pair of incoming record and the global location if meant for merged
lookup in later stage
+ HoodieData<Pair<HoodieRecord<R>, Option<HoodieRecordGlobalLocation>>>
incomingRecordsAndLocations
+ = keyAndIncomingRecords.leftOuterJoin(keyAndExistingLocations).values()
+ .map(v -> {
+ final HoodieRecord<R> incomingRecord = v.getLeft();
+ Option<HoodieRecordGlobalLocation> currentLocOpt =
Option.ofNullable(v.getRight().orElse(null));
+ if (currentLocOpt.isPresent()) {
+ HoodieRecordGlobalLocation currentLoc = currentLocOpt.get();
+ boolean shouldPerformMergedLookUp = mayContainDuplicateLookup
+ || !Objects.equals(incomingRecord.getPartitionPath(),
currentLoc.getPartitionPath());
+ if (shouldUpdatePartitionPath && shouldPerformMergedLookUp) {
+ return Pair.of(incomingRecord, currentLocOpt);
+ } else {
+ // - When update partition path is set to false,
+ // the incoming record will be tagged to the existing record's
partition regardless of being equal or not.
+ // - When update partition path is set to true,
+ // the incoming record will be tagged to the existing record's
partition
+ // when partition is not updated and the look-up won't have
duplicates (e.g. COW, or using RLI).
+ return Pair.of((HoodieRecord<R>) getTaggedRecord(
Review Comment:
this was having Option.empty() as right of the pair and it won't be
merged-lookup candidates
--
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]
