the-other-tim-brown commented on code in PR #13602:
URL: https://github.com/apache/hudi/pull/13602#discussion_r2228654902
##########
hudi-common/src/main/java/org/apache/hudi/metadata/HoodieBackedTableMetadata.java:
##########
@@ -496,41 +503,82 @@ private HoodieData<String> repartitioningIfNeeded(
return keys;
}
- private HoodieFileGroupReader<IndexedRecord>
buildFileGroupReader(List<String> sortedKeys,
- FileSlice
fileSlice,
- boolean
isFullKey) {
+ private ClosableIterator<IndexedRecord> readSliceWithFilter(List<String>
sortedKeys, FileSlice fileSlice, boolean isFullKey) throws IOException {
Option<HoodieInstant> latestMetadataInstant =
metadataMetaClient.getActiveTimeline().filterCompletedInstants().lastInstant();
String latestMetadataInstantTime =
latestMetadataInstant.map(HoodieInstant::requestedTime).orElse(SOLO_COMMIT_TIMESTAMP);
- Schema schema =
HoodieAvroUtils.addMetadataFields(HoodieMetadataRecord.getClassSchema());
// Only those log files which have a corresponding completed instant on
the dataset should be read
// This is because the metadata table is updated before the dataset
instants are committed.
Set<String> validInstantTimestamps = getValidInstantTimestamps();
- InstantRange instantRange = InstantRange.builder()
+ Option<InstantRange> instantRange = Option.of(InstantRange.builder()
.rangeType(InstantRange.RangeType.EXACT_MATCH)
- .explicitInstants(validInstantTimestamps).build();
-
+ .explicitInstants(validInstantTimestamps).build());
+
+ // If reuse is enabled and full scan is allowed for the partition, we can
reuse the file readers for base files and the reader context for the log files.
+ Map<StoragePath, HoodieAvroFileReader> baseFileReaders =
Collections.emptyMap();
+
FileGroupRecordBufferInitializer.ReusableFileGroupRecordBufferInitializer<IndexedRecord>
recordBufferInitializer = null;
+ if (reuse && isFullScanAllowedForPartition(fileSlice.getPartitionPath())) {
+ Pair<HoodieAvroFileReader,
FileGroupRecordBufferInitializer.ReusableFileGroupRecordBufferInitializer<IndexedRecord>>
readers =
+ reusableFileReaders.computeIfAbsent(fileSlice.getFileGroupId(), fgId
-> {
+ try {
+ HoodieAvroFileReader baseFileReader = null;
+ if (fileSlice.getBaseFile().isPresent()) {
+ baseFileReader = (HoodieAvroFileReader)
HoodieIOFactory.getIOFactory(storage).getReaderFactory(HoodieRecord.HoodieRecordType.AVRO)
+ .getFileReader(metadataConfig,
fileSlice.getBaseFile().get().getStoragePath(),
metadataMetaClient.getTableConfig().getBaseFileFormat(), Option.empty());
+ }
+ return Pair.of(baseFileReader,
buildReusableRecordBufferInitializer(fileSlice, latestMetadataInstantTime,
instantRange));
+ } catch (IOException ex) {
+ throw new HoodieIOException("Error opening readers for metadata
table partition " + fileSlice.getPartitionPath(), ex);
+ }
+ });
+ if (fileSlice.getBaseFile().isPresent()) {
+ baseFileReaders =
Collections.singletonMap(fileSlice.getBaseFile().get().getStoragePath(),
readers.getLeft());
+ }
+
+ ValidationUtils.checkArgument(isFullKey, "For Metadata Table Reuse, key
filter should be based on full keys");
+ recordBufferInitializer = readers.getRight();
+ }
+
HoodieReaderContext<IndexedRecord> readerContext = new
HoodieAvroReaderContext(
storageConf,
metadataMetaClient.getTableConfig(),
- Option.of(instantRange),
+ instantRange,
Option.of(
isFullKey
? transformKeysToPredicate(sortedKeys)
- : transformKeyPrefixesToPredicate(sortedKeys)));
- return HoodieFileGroupReader.<IndexedRecord>newBuilder()
+ : transformKeyPrefixesToPredicate(sortedKeys)),
+ baseFileReaders);
+
+ HoodieFileGroupReader<IndexedRecord> fileGroupReader =
HoodieFileGroupReader.<IndexedRecord>newBuilder()
.withReaderContext(readerContext)
.withHoodieTableMetaClient(metadataMetaClient)
.withLatestCommitTime(latestMetadataInstantTime)
.withFileSlice(fileSlice)
- .withDataSchema(schema)
- .withRequestedSchema(schema)
+ .withDataSchema(SCHEMA)
+ .withRequestedSchema(SCHEMA)
.withProps(buildFileGroupReaderProperties(metadataConfig))
- .withStart(0)
- .withLength(Long.MAX_VALUE)
- .withShouldUseRecordPosition(false)
+ .withRecordBufferInitializer(recordBufferInitializer)
.build();
+
+ return new
CloseableIteratorWithReuse<>(fileGroupReader.getClosableIterator(), reuse);
Review Comment:
Yes, makes sense not to wrap it when we're not reusing the iterator. It will
just cause confusion.
I want to avoid adding the method directly on FGreader since this is a
special case and I don't want someone to accidentally cause resource leaks in
other parts of the code.
--
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]
