boneanxs commented on code in PR #6046:
URL: https://github.com/apache/hudi/pull/6046#discussion_r970247306
##########
hudi-client/hudi-spark-client/src/main/java/org/apache/hudi/client/clustering/run/strategy/MultipleSparkJobExecutionStrategy.java:
##########
@@ -273,6 +330,60 @@ private HoodieData<HoodieRecord<T>>
readRecordsForGroupBaseFiles(JavaSparkContex
.map(record -> transform(record, writeConfig)));
}
+ /**
+ * Get dataset of all records for the group. This includes all records from
file slice (Apply updates from log files, if any).
+ */
+ private Dataset<Row> readRecordsForGroupAsRow(JavaSparkContext jsc,
+ HoodieClusteringGroup
clusteringGroup,
+ String instantTime) {
+ List<ClusteringOperation> clusteringOps =
clusteringGroup.getSlices().stream()
+ .map(ClusteringOperation::create).collect(Collectors.toList());
+ boolean hasLogFiles = clusteringOps.stream().anyMatch(op ->
op.getDeltaFilePaths().size() > 0);
+ SQLContext sqlContext = new SQLContext(jsc.sc());
+
+ String[] baseFilePaths = clusteringOps
+ .stream()
+ .map(op -> {
+ ArrayList<String> readPaths = new ArrayList<>();
+ if (op.getBootstrapFilePath() != null) {
+ readPaths.add(op.getBootstrapFilePath());
+ }
+ if (op.getDataFilePath() != null) {
+ readPaths.add(op.getDataFilePath());
+ }
+ return readPaths;
+ })
+ .flatMap(Collection::stream)
+ .filter(path -> !path.isEmpty())
+ .toArray(String[]::new);
+ String[] deltaPaths = clusteringOps
+ .stream()
+ .filter(op -> !op.getDeltaFilePaths().isEmpty())
+ .flatMap(op -> op.getDeltaFilePaths().stream())
+ .toArray(String[]::new);
+
+ Dataset<Row> inputRecords;
+ if (hasLogFiles) {
+ String compactionFractor =
Option.ofNullable(getWriteConfig().getString("compaction.memory.fraction"))
+ .orElse("0.75");
+ String[] paths = CollectionUtils.combine(baseFilePaths, deltaPaths);
+ inputRecords = sqlContext.read()
Review Comment:
Yea, I was also thinking maybe we can change here as well when fixing the
write-path. But I found it quite difficult, if we want to access DefaultSource
here, we may need to move many dependent codes(it might be whole
`hudi-spark-common` package) from package `hudi-spark-common` to
`hudi-spark-client`(like `BaseFileOnlyRelation`, `HoodieFileIndex`, etc), and
I'm not sure whether it works even we move them, it might meet other problems.
--
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]
