[SPARK-52240][SQL] Fix VectorizedDeltaLengthByteArrayReader.readBinary to handle current row

[wgtmac]

What changes were proposed in this pull request?

Fix VectorizedDeltaLengthByteArrayReader.readBinary to use currentRow to read the length of binary.

Why are the changes needed?

VectorizedDeltaLengthByteArrayReader.readBinary should use currentRow instead of rowId to read from the internal data stream. The variable rowId is the destination of the output column vector not the position to read binary data from input. This causes dirty read and throws ParquetDecodingException.

Does this PR introduce any user-facing change?

No.

How was this patch tested?

Pass all existing test cases.

Was this patch authored or co-authored using generative AI tooling?

No.

[wgtmac]

PTAL @sunchao @parthchandra

[pan3793]

Same fix but was not lucky to get reviewed. #46928

[wgtmac]

@pan3793 Thanks for letting me know! So is it a good time to revive that one (because it has good test)?

[pan3793]

@wgtmac #46928 gets reopened now, if you don't mind, we can wait for responses(maybe one or two days) from the original author of #46928, and see if he still wants to move the PR forward.

[wgtmac]

I'm very happy to wait for it.

cc @dongjoon-hyun

[dongjoon-hyun]

Thank you for making a PR and pinging me, @wgtmac . Do you think we can add a test case for that dirty read, ParquetDecodingException?

[sunchao]

Thanks for the fix @wgtmac !

[wgtmac]

Thanks @dongjoon-hyun and @sunchao! I think #46928 is a good fix with test cases. Does it make sense to merge that? (I haven't been working on Spark for a long time so it may take me more time to setup the test cases)

[dongjoon-hyun]

To all, we cannot merge #46928 because it's opened for branch-3.5. We need to merge yours.

Given that, @wgtmac , please add a commit with #46928 's test case with the authorship of him.

Then, I can merge with the script with both author.

[dongjoon-hyun]

cc @djspiewak from #46928 as the branch-3.5 PR author, too.

[wgtmac]

I'll wait for a day or two to see if the original author will work on the master branch.

[djspiewak]

I'm happy to retarget master or do whatever is deemed best. Let me know your preferences. I'll reply to the test case question on the other PR to avoid confusing the threads.

[parthchandra]

lgtm. Thanks for the fix @wgtmac, @djspiewak

[dongjoon-hyun]

I'm happy to retarget master or do whatever is deemed best. Let me know your preferences. I'll reply to the test case question on the other PR to avoid confusing the threads.

Thank you, @djspiewak . We always prefer the original one (yours in this case).

[dongjoon-hyun]

Please open a new one based on the master branch and ping us, @djspiewak .

[sunchao]

Wow, what a surprise to see you here @djspiewak! We were in the same John Boyland’s type theory class 10+ years ago 😂

[djspiewak]

Small world! :D

[djspiewak]

Cross-posting: I've updated the other PR and retargeted on master