GH-44345: [C++][Parquet] Add Decimal32/64 support to Parquet

[HuaHuaY]

Rationale for this change

As described in #44345, Decimal32/Decimal64 have been implemented but Parquet has poor support. This change allows to write Decimal32/Decimal64 into Parquet file the same way as Decimal128/Decimal256 and to read Decimal32/Decimal64 from an existing Parquet file.

What changes are included in this PR?

1. Support writing Decimal32/Decimal64 as INT32/INT64/BYTE_ARRAY/FIXED_LEN_BYTE_ARRAY into Parquet file.
2. Support reading Parquet column with logical type Decimal. Either reading type from metadata or infering Arrow Decimal type is supported.

Are these changes tested?

Yes.

Are there any user-facing changes?

Yes. A flag named smallest_decimal_enabled_ is added in ArrowReaderProperties. To maintain backward compatibility, only when the flag is true, Arrow will infer Decimal with small precision to Decimal32/Decimal64 instead of Decimal128.

• GitHub Issue: [C++][Parquet] arrow Decimal32/Decimal64 write Parquet and testing #44345

[Copilot]

Pull Request Overview

This PR adds support for reading and writing Arrow Decimal32 and Decimal64 types in Parquet files. The implementation extends the existing Decimal128/256 support to include smaller decimal types, allowing for more efficient storage of decimal values with lower precision.

• Extends Parquet I/O to support Decimal32/64 alongside existing Decimal128/256 types
• Adds reader property for enabling smallest decimal type inference from Parquet
• Consolidates decimal serialization logic to support all decimal types uniformly

Reviewed Changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
cpp/src/parquet/properties.h	Adds smallest_decimal_enabled_ flag to ArrowReaderProperties for backward compatibility
cpp/src/parquet/column_writer.cc	Extends decimal serialization to support Decimal32/64 with unified template logic
cpp/src/parquet/arrow/test_util.h	Refactors decimal test utilities to be generic across all decimal types
cpp/src/parquet/arrow/schema_internal.h	Updates function signatures to accept ArrowReaderProperties parameter
cpp/src/parquet/arrow/schema_internal.cc	Implements smallest decimal type selection logic using new reader property
cpp/src/parquet/arrow/schema.cc	Adds Decimal32/64 cases to schema conversion and metadata restoration
cpp/src/parquet/arrow/reader_internal.cc	Extends decimal reading logic to support all decimal types through generic templates
cpp/src/parquet/arrow/arrow_reader_writer_test.cc	Adds comprehensive test coverage for Decimal32/64 roundtrip scenarios

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[HuaHuaY]

@pitrou @mapleFU Could you spare some time to review this PR?

[wgtmac]

+1

Left some nits. Thanks @HuaHuaY!

[pitrou]

Thanks for doing this @HuaHuaY . Can we also update https://arrow.apache.org/docs/dev/cpp/parquet.html#logical-types ?

[HuaHuaY]

Can we also update https://arrow.apache.org/docs/dev/cpp/parquet.html#logical-types ?

Is this web page generated by docs/source/cpp/parquet.rst? I have pushed a commit to add Decimal32/Decimal64.

[wgtmac]

All CI failures are unrelated for the same reason below:

CMake Error at /opt/conda/envs/arrow/share/cmake-4.1/Modules/FindPackageHandleStandardArgs.cmake:227 (message):
  Could NOT find LLVMAlt (missing: LLVM_PACKAGE_VERSION CLANG_EXECUTABLE
  LLVM_FOUND LLVM_LINK_EXECUTABLE)

I think it is ready to merge. Do you have more comments? @pitrou

[pitrou]

LGTM, but can you please rebase or merge from main to avoid the current CI failures @HuaHuaY ?

[HuaHuaY]

I rebased the branch.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 4 benchmarking runs that have been run so far on merge-commit a444380.

There weren't enough matching historic benchmark results to make a call on whether there were regressions.

The full Conbench report has more details.