HDDS-15348. OmMultipartPartKeyCodec should not use UTF8.decode(..)

[Russole]

What changes were proposed in this pull request?

• Replace StandardCharsets.UTF_8.decode(...) in OmMultipartPartKey with StringCodec for decoding the persisted upload ID bytes.
• The round-trip UTF-8 validation is needed because StringCodecBase.decode(...) catches strict decoding exceptions and falls back to compatibility decoding, which may use replacement characters. Re-encoding the decoded upload ID and comparing it with the original bytes lets us detect non-lossless decoding and reject malformed UTF-8 input.
• Add a unit test to verify malformed UTF-8 upload IDs throw IllegalArgumentException.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-15348

How was this patch tested?

• Added/updated unit test:
  • TestOmMultipartPartKey#testDecodeRejectsMalformedUtf8UploadId
• Ran locally:
  • mvn -pl :ozone-common -Dtest=TestOmMultipartPartKey test
• All CI checks passed
  • https://github.com/Russole/ozone/actions/runs/26320555387

[szetszwo]

@Russole , thanks for working on this! We should throw CodecException. It needs a new variant of StringCodec. Filed HDDS-15355.

BTW, we should also change the other IllegalArgumentException in OmMultipartPartKeyCodec to CodecException.

[szetszwo]

@Russole , thanks for the update. The change looks good.

We should also update toCodecBuffer(..) and toPersistedFormat(..) to use StringCodec.getCodecNoFallback() since String.getBytes(..) could replace the unsupported character silently.

[szetszwo]

+1 the change looks good.

[adoroszlai]

Thanks @Russole for the patch.

[adoroszlai]

@szetszwo Can we use StringCodecBase.decodeNoFallback(ByteBuffer) directly, to avoid buffer copy?

ozone/hadoop-hdds/common/src/main/java/org/apache/hadoop/hdds/utils/db/StringCodecBase.java

Lines 199 to 200 in cfb8ade

	public String fromPersistedFormat(byte[] bytes) throws CodecException {
	return decodeNoFallback(ByteBuffer.wrap(bytes));

ozone/hadoop-hdds/common/src/main/java/org/apache/hadoop/hdds/utils/db/StringCodecBase.java

Lines 115 to 117 in cfb8ade

	String decodeNoFallback(ByteBuffer buffer) throws CodecException {
	try {
	return newDecoder().decode(buffer.asReadOnlyBuffer()).toString();

[szetszwo]

@adoroszlai , this is a good point! However, it will need more changes to make it work properly. I believe the other codecs have similar problems. So, how about we do it separately?