GH-48234: [C++][Parquet] Fix overly strict check for BIT_PACKED levels byte size

[pitrou]

Rationale for this change

The original change from data_size to data_size - 4 was made by me in #6848 and it was probably based on a misunderstanding, or just blindly copying the similar check for Encoding::RLE (which, unlike Encoding::BIT_PACKED, does have a 4-byte length header).

Are these changes tested?

Yes, by existing CI tests.

Are there any user-facing changes?

No, just a potential bugfix for an unlikely case.

• GitHub Issue: [C++][Parquet] Overly strict check in LevelDecoder::SetData #48234

[mapleFU]

How was this found? Does someone meet this or we have came across this?

[pitrou]

Yes, I forgot to mention: this was found out in this PR: https://github.com/apache/arrow/pull/48205/files#r2553142118

[pitrou]

@github-actions crossbow submit -g cpp

[mapleFU]

I give a quick glance and this is ok to me, but I may need some time to find out why it doesn't trigger any bug previously...

[github-actions]

Revision: 0390720

Submitted crossbow builds: ursacomputing/crossbow @ actions-d4511141a8

Task	Status
example-cpp-minimal-build-static
example-cpp-minimal-build-static-system-dependency
example-cpp-tutorial
test-build-cpp-fuzz
test-conda-cpp
test-conda-cpp-valgrind
test-cuda-cpp-ubuntu-22.04-cuda-11.7.1
test-debian-12-cpp-amd64
test-debian-12-cpp-i386
test-fedora-42-cpp
test-ubuntu-22.04-cpp
test-ubuntu-22.04-cpp-20
test-ubuntu-22.04-cpp-bundled
test-ubuntu-22.04-cpp-emscripten
test-ubuntu-22.04-cpp-no-threading
test-ubuntu-24.04-cpp
test-ubuntu-24.04-cpp-bundled-offline
test-ubuntu-24.04-cpp-gcc-13-bundled
test-ubuntu-24.04-cpp-gcc-14
test-ubuntu-24.04-cpp-minimal-with-formats
test-ubuntu-24.04-cpp-thread-sanitizer

[pitrou]

I give a quick glance and this is ok to me, but I may need some time to find out why it doesn't trigger any bug previously...

Probably because data_size is the entire page size, and it also contains the page values. It's rare for page values to take less than 4 bytes.

[pitrou]

I'll merge now.

[mapleFU]

I forgot to approve yesterday night, I've tried to set the encoder to BIT_PACKED but it failed to encode. Perhaps we don't have testing data for this, and currently writer would merely produce file like this. So the patch LGTM

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit 52d258b.

There were no benchmark performance regressions. 🎉

The full Conbench report has more details. It also includes information about 2 possible false positives for unstable benchmarks that are known to sometimes produce them.